IEEE Transactions on Power Delivery, Vol. 8, No.

4, October 1993 1811

DYNAMIC MODEL OF A THREE-PHASE POWER TRANSFORMER

D. Dolinar, Member, IEEE J. Pihler, Student Member, IEEE B. GrEar, Associate Member, IEEE
Faculty of Technical Sciences
University of Maribor
Maribor, Slovenia

Abstract - An adequate mathematical model of a is performed. The approach mentioned above enables
three-phase power transformer is one of the important the two authors to evaluate the losses in the steel core.
elements in the programs for the computer analysis of The author [3] efficiently calculates the inrush current
power system transients. Featured in this paper is the by the quasi-stationary approach in the harmonic do-
simulation model of a three-phase, three-limb core-type main. Three-phase two- and three-winding transformers
power transformer. Non-linear effects of saturation, hys- are composed of single-phase ones in different connec-
teresis and eddy currents are considered. Two ways of tions, and the impedance matrix representation is used
creating major and minor hysteresis loops are presented. for transformers [4]. The saturation effect is considered
The transformer model, described by a system of time by the piecewise linear magnetizing characteristics.
dependent differential equations, is solved by an efficient The works of [5,6,7]discuss in detail the effects of mag-
numerical algorithm. The behaviour of the transformer netic nonlinearities, but the transformer model is not
model during switching-in and fault transients, as well brought to its final form and the results of simulation
as other types of transients, has been tested. The com- are not given. In reference [8], a clearly presented model
puted transient waveforms are compared with the mea- of a three-phase, three-limb power transformer enables
sured ones and there exists very close agreement between the computation of the inrush currents, but the non-
them. linearities of the steel core are not taken into account.
The three-phase model which also takes into account the
INTRODUCTION mutual coupling between the phases and the steel core
is presented in reference [9], but the emphasis is put
The use of digital computers for the analysis of tran- on the iterative solution of the nonlinear part and on
sients in electric power systems and for the design of the incorporation of the transformer model into the net-
transformers protection systems requires, among other work simulation program. The modelling of nonlinear-
things, an adequate dynamic model of the three-phase ities presented in recent works [10,11] and in reference
transformers. Despite their relatively simple design the [12] is rather complex and can probably not be used in
accurate and realistic representation of nonlinear effects universal electromagnetic transient programs.
of the laminated steel core represents a challenge to The model of a single-phase steel-core transformer in
numerous authors. Several of them discuss the mod- the form of algebraic and differential equations is per-
elling of single-phase transformers considering the non- haps best presented by Ewart [13], and as the author
linear effects caused by the steel core. The approach claims, may be a solid basis for the derivation of a three-
used in papers [1,2] is based on equivalent electric cir- phase model which is not presented in the paper. Hys-
cuits. Nonlinearities of the steel core are considered in teresis and saturation are taken into account as the com-
the form of additional non-linear elements in the mag- position of dead band elements and limiters.
netizing branch; they are more numerous and are even The simplified power transformer model with the
frequency dependent [2]. In reference [l], a comparison piecewise linear magnetization curve is presented in pub-
of the results obtained by the harmonic analysis based lication [14]. The model is relatively simple and enables
on the so-called frequency and time-domain approach a quality analysis of steady and transient states. A ma-
jor deficiency of this model results from the considerable
93 WM 048-9 PWRD A paper recommended and approved discrepancy between direct as well as harmonic compo-
by the IEEE Transformers Committee of the IEEE Power nents of the calculated and the measured currents. The
Engineering Society for presentation at the IEEE/PES
1993 Winter Meeting, Columbus, OH, January 31 - main reason for this is the insufficient accuracy in con-
February 5, 1993. Manuscript submitted August 28, sidering the nonlinear effect of the steel core.
1992; made available for printing November 10, 1992. The model of a three-phase, three-limb multi-winding
core-type transformer described in this paper was de-
veloped as a test tool to be used by the designer of the
digital differential protection with the ambition to incor-
porate it into the program for the analysis of power sys-
tem transients. As regards the testing of the transformer

0885-8977/93/$03.000 1993 IEEE

1812

differential protection, the main objective was a precise

presentation of inrush currents and currents at various
loading levels, from the no-load to various short circuits.
The need for a reliable harmonic reproduction of cur-
rents conditioned the consideration of nonlinear effects
of saturation and hysteresis. Our intention to include the
transformer model into the simulation programs required
a verified numerical approach and a relatively simple en-
try of a not too large number of data.
The improved version of the power transformer model
Fig. 1. Three-phase transformer scheme
presented here differs from the original one [14] in the
formulation of the nodal equations for fluxes. The link-
age fluxes are the sums of leakage and mutual fluxes. The R p 1 , 2 , 3 , R s 1 , 2 , 3 are ohmic resistances of primary ( p ) and
leakage fluxes inside the transformer tank are neglected secondary (s) windings of individual phases and R e 1 , 2 , 3
and only those in the air are considered. Due to the pre- are ohmic resistance of short-circuited windings to rep-
vious assumption, the linkage inductance becomes con- resent the eddy-current effects in the core. N p 1 , 2 , 3 ,
stant. The leakage and mutual fluxes are treated sep- N s 1 , 2 , 3 , N e 1 , 2 , 3 are the number of turns of the primary,
arately and all nonlinear effects due to steel core are secondary and short-circuited windings. @ 1 , 2 , 3 are the
considered in the determination of mutual fluxes. The mutual fluxes, and @ U p 1 , 2 , 3 , @ O s 1 , 2 , 3 , @ u e 1 , 2 , 3 the leak-
procedure according to which the magnetization curve age fluxes of the primary, secondary and of the short-
and the hysteresis loop is calculated by polynomial ap- circuited windings. The short circuited windings are
proximation from the data provided by the steel core located on each of the three limbs, however, they are
manufacturer is presented in this contribution. The non- omitted in Fig. 1. In order to simplify the considera-
linear, non-symmetrical, history dependent hysteresis- tion of magnetic nonlinearities, the total linking fluxes
saturation relationship between the net magneto-motive in equations (1) are expressed by the sums of mutual and
forces (MMFs) and the fluxes is taken into account. The leakage fluxes.
use of the described model on its own, or incorporated The magnetic conditions of the transformer in Fig. 1
into a network simulation program that treats voltages are described by two equations that express the relation-
and currents as instantaneous quantities with arbitrary ship between the net MMFs and the mutual fluxes
waveforms, facilitates studies of inrush currents, ferrores-
onance, harmonic generation, chopping, rectifier loads i p l Npi + i s 1 Ns 1 S i e 1 N e -i p 2 N p 2 -is 2 N s 2 - i e 2 N e
and others. =@1Rml-@~Rm2
The modelling procedure described below is not en- i pz Npz + i s 2 Ns2 +i e 2N e - i p 3 N p 3 - i s3 N s 3 - i e 3 N e (2)
tirely new, but it is a good compromise between the ac- =@2Rm2-@3Rm3
curacy and the excessive complexity found in some excel-
lent previously quoted works. The significant contribu- Rm1,2,3 in (2) represent the reluctances of individual
tion of our paper is a simple and efficient incorporation magnetic circuit branches and they are generally vari-
of nonlinear effects of the steel core into the three-phase able. The following node equation is valid for mutual
transformer model. fluxes
a 1 @2 @3 = 0 + + (3)
MATHEMATICAL MODEL OF A The condition stated in equation (3) simplifies the trans-
THREE-PHASE TRANSFORMER former model considerably, it does, however, restrict its
use to cases in which both the primary and the secondary
The mathematical model of the transformer is de- neutrals of the wye-wye connected windings are either
veloped for a three-limb core-type transformer whose grounded or not grounded. Equations (2) can be writ-
primary and secondary windings are star connected. ten in matrix form, and the flux @ 3 can be expressed by
A transformer model of this kind represents the basic the linear combination of @ 1 and @ 2
mathematical model of a three-phase transformer from
which a model with differently connected primary and
secondary windings and a different steel core geometries
are developed. Fig. 1 presents the general scheme of a
= [ R~~
-Rm2
+R ~ I[: ~ ] (4)
where
three-phase, three-limb core-type transformer. The mag-
netic circuit branches are according to Fig. 1 featured by
the mean flux paths of different lengths 1 1 , 1 2 , 13. Voltage
equations of the transformer in Fig. 1 are given by ~,
The mutual fluxes are obtained from system (4)
 1813

Coefficients a l l , a 1 2 , a 2 1 and a22 are the elements of MAGNETIZATION CURVE AND

the inverse reluctance matrix. The mutual fluxes from HYSTERESIS LOOP
expression (6) are given by
The properties of reluctances R m 1 , 2 , 3 in equation (4)
I[: = [ bii biz b i 3 b i 4 b i 5 b i 6 6 1 7 bis b i g
b 2 1 b 2 2 b 2 3 b 2 4 b 2 5 b z 6 b 2 7 b 2 8 b29 ] .- (7)
have not been discussed yet. The nonlinearity of the
magnetization curve and the hysteresis loop appears as
variable reluctances, therefore reluctances R m 1 , 2 , 3 are
where the vector of the current is defined as
nonlinear dependent on MMFs. They must be calculated
- . . .
-iT - [ % P I i, p 2 , 2 p 3 , z s 1 ~is21 i s 3 , i e l l i e 2 , i e 3 ] (8) in each integration step.
Additional difficulties in the calculation of reluctances
The leakage fluxes are also expressed by currents and of individual magnetic circuits branches are due to the
leakage inductances
fact that the flux of one phase is no longer solely a func-
Qupi = (Lopi/Npi)ipi tion of the net MMF of an individual phase, but of the
Qgsi = (Losi/Nsi)isi i = I, 2 , 3 (9) net MMFs of all three phases. Consequently, the reluc-
Qoei = (Luei/Nei)iei tances of all three branches are calculated separately.
The leakage fluxes are inserted into the basic voltage 72
equations (1). The system of equations obtained in this HIA"l t 1
way is expressed by the time derivatives of primary, sec-
ondary and eddy currents and in general they can be
written in the matrix form
U = c-
-
where

-
U =
"egatlve
region
*aturatlon~

segment f-
v
egotive saturation
point

Fig. 2 . Typical major hysteresis loop characteristic

C= tI
Determinine the magnetization curve and the maior
hysteresis loop by polynomial approximation

Although conventionally the magnetization curve and

the hysteresis loop are plotted as B versus H the H-B
Coefficients cij are not constant because they include representation in Fig. 2 is more useful for further calcula-
the combination of reluctances of individual magnetic tion of magnetic reluctances. The points, beyond which
circuit branches and of a number of primary, secondary the characteristics in Fig. 2 becomes single-valued are
and short-circuited turns. u P 1 , 2 , 3 are the instantaneous defined as the positive and negative saturation points.
values of primary voltages, and the secondary voltages The magnetization curve and the major hysteresis loop
~ , 1 , 2 , 3are expressed as ohmic voltage drops i s i R b i (i = in the H - B plane are divided into five segments. The
1 , 2 , 3 ) . L a i ( i = 1 , 2 , 3 ) are the inductances of the load. first three are approximated by polynomials of the 13th
The time derivatives of currents are obtained from equa- order. The functional and derivative values of their com-
tion (10) by mon positive and negative saturation points are equal so
i
=c - 1 g (12) their fitting is not problematical. The fourth and the
The calculation of matrix C-' is possible only if the fifth segment in the positive and negative saturation re-
values of leakages L o p 1 , 2 , 3 , L o s 1 , 2 , 3 and L o e 1 , 2 , 3 do not gion are represented by parabolas. The functions ex-
equal zero, otherwise the third, sixth and ninth row of pressing the segments must be defined so that d H / d B is
matrix C are a linear combination of the others, and the at least continuous in the entire definition region of the
determinant of matrix C is equal to zero. The vector H-B plane.
4 is composed of time derivatives of primary, secondary Let us first see how the polynomial approximation for
and eddy currents. Equation (12) represents the final the first three segments was carried out. We assume that
power transformer model and can be solved by standard the magnetization curve and the major hysteresis loop
numerical integration procedures. are given point by point in tabular form on the basis of
1814

the data supplied by the producer of the steel core, and we are presenting just the fitting of the first and fourth
that all states may lie only inside the major hysteresis segments. Let us assume that the first segment of the
loop or on it. The first segment of function g(B) are to magnetization curve is approximated by a polynomial of
be given in tabular form: 13th order, and the fourth segment by the equation of
the parabola running through points TI and T2 (Fig. 3).
TI is the last point of the first segment and also the first
point of the fourth segment. Tz is the selected point from
the fourth segment of the magnetization curve. In this
common point the polynomial and the parabola must
The approximation polynomial is to be have the same functional and derivative values. The
n
parabola that fulfils the requirements can be obtained
f ( B )= ciBi from the system of equations
i=O
+
H(T1) = L(B(T1)- B(To)y H(T0)
where ci are the coefficients of the approximation poly- +
H(Tz) = k(B(T2)- B(To))' H(To) (18)
nomial sought. The following system of equations is set %(TI) = WB(T1) - B(T0))
up for the determination of coefficients ci
k , B(T0) and H ( T 0 ) are the parameters obtained as the
solution of the system of equations (18). The equation
of the sought parabola is

H - H ( T 0 ) = L(B - B(To))2 (19)

All five functions that approximate the individual seg-

ment from Fig. 2 are connected by the procedure de-
1 B1 B; ... scribed above so that the derivation d H / d B is smooth
B= . . .. ..
throughout the entire definition range of H-B plane.
1 Bn BZ . . . B,"
The coefficients ci are derived from equation (14) on the Creation of the magnetization curve and the hysteresis
basis of error defined as ~ O O Dbv
B-Sdine curve

The second possible way of determining the magneti-

zation characteristic and the hysteresis loop is more gen-
which must be minimal in the quadratic sense. We get eral than the first one. It enables a simple creation of the
the well-known expression minor hysteresis loop. Let us assume that the magneti-
zation characteristic, the major hysteresis loop, and the
c = [BTB]-lBTg(B)
-
curve of hysteresis loss depending on flux density for the
The calculated coefficients are inserted into polynomial nominal frequency are given point by point in tabular
(13). Thus the approximation value H = f ( B ) is ob- form. The magnetization curve and the hysteresis loop
tained for function H = g(B). are, according to Fig. 2, divided into five segments. The
fourth and the fifth are approximated by two parabolas,
similarly as in the previous case, whereas the magneti-
zation curve, the major hysteresis loop and the curve of
hysteresis loss are approximated by nonperiodic B-spline
curves of the fourth order [15]. These are curves of class
C2, which means that they are two times continuously
derivable. From here on +B, and -B, will denote the
value of the flux density in the last reversal point. In
accordance with Fig. 2, +B,,, and -B,,, denote both
points of saturation. In the case when by reversal of B
the value +B,,, and -B,,, in the positive or nega-
Fig. 3. Fitting procedure of two polynomials tive direction is not achieved, similarly as in Fig. 9 at
switching-in, a minor hysteresis loop is reached. Minor
The remaining segments from Fig. 2 are represented hysteresis loops lie inside the major loop and are similar
by polynomials of the 2nd order. The same technique is to it. Their area represents the hysteresis loss per cycle.
used for the fitting of the fourth and fifth segments, so During the variation of B from -BR to BR the energy,
 1815

Of course, the minor hysteresis loop inside -B, and

Fig. 6. Hysteresis loss per half-cycle within the values

-B, and Bm defined by major hysteresis loop

Bm has a smaller area W,(Bm). It is defined according

to the loss from the diagram in Fig. 5. The difference
between the surfaces

Fig. 4. Hysteresis loop and energy relation per half-cycle

defines the factor
equal to the shaded area in Fig. 4 defined by equation AWw(Bm)
(20), is dissipated per half-cycle. AH(B,) =
2Bm

W, = lBR LIm
B,
HdB - HdB = for which the major hysteresis loop should be reduced
(Fig. 7).
The hysteresis power loss per cubic meter at a constant
frequency f is defined by
t HIA/ml m
PFe = 2Wvf (21)
The producer of the steel core gives it in the form of

P, [w/m31
t I
Fig. 7. Graphical interpretation of the major hysteresis
loop reduction

B,[T] The points belonging to the minor hysteresis loop (Fig.

8) are expressed by
Fig. 5. Hysteresis power loss per cubic meter

a graph in Fig. 5. The minor hysteresis loop is defined

on the basis of the instantaneous value of B, so that its where HMaxHyst(B) is the equation for the major hys-
area corresponds to the loss in the diagram in Fig. 5. Let teresis loop. By the appropriate selection of value AB
us assume that the minor hysteresis loop is defined by within the values -B, and B, we can get the adequate
the value B, and that it lies inside the major hysteresis number of points on the minor hysteresis loop which are
loop which is defined by the values of B,,, . One half of then approximated by B-spline curve of the fourth order.
the area of the major hysteresis loop between the reversal In this way the equation of the minor hysteresis loop de-
points is, in accordance with Fig. 6, defined by
fined by the instantaneous value of flux density B, can

WuMaxHyst(Bm) = /-B,
Bm
H dB (22)
be determinated.
1816

Fig. 8. Determination of the minor hysteresis loop

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

Calculation of reluctances ml
The procedures described above enables the calcula- Fig. 9. Hysteresis loops a t switching-in (phase 1)
tion of the analytical expression for the magnetization
curve and the major and minor hysteresis loops in two RESULTS
ways. The values of differential reluctance R m d of in-
dividual magnetic branches i can be calculated in each The described transformer model was verified by com-
integration step by paring the calculated currents with the currents mea-
sured on a 100 kVA transformer a t different operating
conditions. Precise data of the transformer, including
the magnetization curve and hysteresis loop data, were
where li and Si are the mean values of lengths and the provided by the manufacturer. With the exception of the
cross-section area of the individual magnetic branches. magnetization curve and the hysteresis loop, the data are
The relationship between the net MMFs and the fluxes presented in the Appendix.
in equations (2) is defined by the total value of reluc- The comparison of calculated and measured steady
tances R m i ( i = 1 , 2 , 3 ) . In order to attain an accurate states currents of the primary and secondary side of the
description of nonlinear magnetic conditions in an oper- nominally loaded transformer is not given in the paper.
ating point, these equations (2) must be complemented However , the difference between them does not exceed
by the values @ i ( O ) , as shown in Fig. 3, and calculated the limit of 2 percent. The time response curves of cal-
from the relevant values Bi(0). The modified system culated and measured inrush currents are given in Fig.
(2) in the operating point TI is linear again and can be 10 and 11. Fig. 12 and 13 show the time response of
calculated by harmonics for calculated and measured current in phase
1. The voltage in phase 1 was zero at the moment
i p 1 Np 1 +is 1 Ns 1 +ie 1 Ne -ipzNpz -is 2 Ns 2 +i e Ne
2 of switching-in and the transformer was demagnetized.
+@l(O)Rmdl-@Z (O)RmdZ=@lRmdl-@ZRmdZ The comparison of responses of calculated and measured
(27) inrush currents shows a very good agreement as regards
$2 Npz +is z Ns z +iez Ne -ip3Np3 -is 3 Ns 3 +ie3 Ne
+@Z(O)RmdZ-@3(0)Rmd3=@ZRmd2-~3Rmd3 the amplitude and harmonics. The comparison of har-
monics in Fig. 12 and 13 shows that the calculated am-
The above description discusses the symmetric position plitude values and the time constant of decreasing dc
of the hysteresis loop to the origin; the symmetry, how- component differ from the measured ones by less than
ever, is not necessary, as it is illustrated by the next 4 percent. The comparison of calculated and measured
example. Let us assume the switching-in of the three- values in the remaining two phases gave a similar result,
phase transformer in the case when residual fluxes in but due to the limited extent of the paper, the responses
all three phases are zero and the first phase voltage at are not presented.
zero is going positive, and symbolically observe the con- The harmonic analysis of measured and calculated
ditions in the first phase (Fig. 9). When the positive currents was carried out by means of the nonrecursive fil-
reversal point 1 is reached by the magnetization curve, ter obtained by the least square method. Consequently,
then the flux density does not decrease along the same the calculated responses of instantaneous harmonic com-
curve but along the major hysteresis loop. The hystere- ponents are delayed. The measured steady state no-load
sis loop follows the polynomial of 13th order through the current in phase 1 is presented in Fig. 14, and its har-
characteristic points B , and -H, until the negative re- monics components, in Fig. 15. The calculated steady
versal point 2 is reached, when again the magnetic field state no-load current is presented in Fig. 16 and its har-
intensity H increases along the minor hysteresis loop, monic components in Fig. 17. Also in this case the cal-
but only up to point 3. The cyclic variation of H is re- culated values differ from the measured ones by less than
peated, until the dc component of the flux disappears, 5 percent. Approximation by B-spline curves was used
thus enabling the transition to the symmetrical hystere- to represent the magnetization curve and the hysteresis
sis loop between points 5 and 6. loops in the above calculations.
 1817

7
-2nd Harmonic
8

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
f[4 r1.1

Fig. 10. Calculated inrush currents Fig. 12. Harmonic components of the calculated inrush
currents (phase 1)

- 1st Harmonic
-2nd Harmonic

I , I
0 0.02 0.04 0.08 0.08 0.1 0.12 0.14 0.16 0.10
r1.l
Fig. 11. Measured inrush currents
Fig. 13. Harmonic components of the measured inrush
currents (phase 1)
CONCLUSION

The presented model of the three-phase power trans-

primarily by a comparison of the inrush currents. The
former is developed from the system of voltage equations
that are coupled in dependence of the transformer con- calculation of the inrush currents represents perhaps the
nection. The star- connection of a transformer is dis- severest test for the model because in this case the effects
cussed in the paper, whereas other models of standard of saturation and hysteresis are most explicit. In order
three-phase two- and three-winding transformers are in- to make the comparison possible, a demagnetization of
cluded in the software package described in [14]. The the transformer was performed before the switching-in.
model considers nonlinearities due to saturation] hys- Measured and calculated steady states at nominal load
teresis and eddy currents. Special attention is paid to the and no-load agree practically perfectly. Also, the re-
modelling of the magnetization curve and the hysteresis sponses of magnetizing inrush currents agree very well.
loop. The nonlinearity of the B = f ( H ) characteristic is The fact that the harmonic content of inrush currents
considered in the basic transformer model by taking into of the transformer matches with that of the model en-
account the variable values of reluctances. Presented are ables us to use it efficiently in the design of the digital
two ways of the mathematical description of the mag- transformer protection. However, the model is gener-
netizing characteristic] as well as the major and minor ally applicable. A detailed description of the model and
hysteresis loops. The two ways differ in the complexities. a large number of applications is available in reference
They give identical results in the calculation of inrush ~41.
and no-load currents. In case of substantial voltage vari-
ations] which are not analyzed in the paper, the second APPENDIX
method yields better results. The model of the trans-
former was verified by comparing the calculated currents Transformer data: Y-Y Connection
with the currents measured on a 100 kVA transformer. U,, = 15000 V Ne1,2,3 = 1 turn
The verification presented in the paper was carried out B = 1.69 T Rp1,2,3 = 25.9 R
1818

Fig. 14. Measured steady state no-load current Fig. 16. Calculated steady state no-load current
(phase 1) (phase 1)

Fig. 15. Harmonic components of the measured steady Fig. 17. Harmonic components of the calculated steady
state no-load current (phase 1) state no-load current (phase 1)

U,, = 400 V Rs1,2,3 = 0.016 R in the Harmonic Domain”, IEEE Transactions on

Ipn = 3.85 A Re1,2,3 = 1.11 R Power Delivery, vol. 4, No.4, pp. 2114-2120, Octo-
Isn - 144 A Lup1,2,3 = 0.14 H ber 1989.
Sn = 100 kVA Lus1,2,3 = 0.098 mH [4] H. W. Dommel, “Extension of the Basic Solution
Np1,2,3 = 2025 turns Lue1,2,3 = 0.003 mH Methods”, IEEE Tutorial Course Digital Simula-
Ns1,2,3 = 54 turns tion of Electrical Transient Phenomena, 81. EHO-
173-5-PWR 82/03281, pp. 20-29, 1982.
[5] G. W. Swift, “Power Transformer Core Behavior
Under Transient Conditions”, IEEE Power Engi-
neering Society, 71 TP 88-PWR, pp. 2206-2210,
REFERENCES February 1971.
J . D. Greene, C. A. Gross, “Nonlinear Modeling of [6] E. P. Dick, W. Watson, “Transformer Models for
Transformers”, IEEE Transactions on Industry Ap- Transient Studies Based on Field Measurements”,
plications, vol. 24, No. 3, pp. 434-438, May 1988. IEEE Transactions on Power Apparatus and Sys-
tems, vol. PAS-100, No. 1, pp. 409-419, January
J . Avila-Rosales, F.L. Alvarado, “Nonlinear Fre-
quency Dependant Transformer Model for Elec- 1981.
tromagnetic Transient Studies in Power Systems” , [7] N. Janssens, “Static Models of Magnetic Hystere-
IEEE Transactions on Power Systems, vol. PAS- sis”, IEEE Transactions on Magnetics, vol. MAG-
101, No. 11, pp. 4281-4288, November 1982. 13, No. 5, pp. 1379-1381, September 1977.
N. RajakoviE, A. Semlyen, “Investigation of the [8] M. A. Rahman, “Digital Simulation of Magnetiz-
Inrush Phenomenon a Quasi-stationary Approach ing Inrush Currents in Three-phase Transformer”,
 1819

IEEE Transactions on Power Delivery, No 4., pp. Since 1978 he has been working at the Faculty of Tech-
79-85, October 1986. nical Sciences in Maribor, Department of Electrical En-
[9] B. Stein, “A Transformer model for Electromag- gineering, Institute for Power Systems as a researcher
netic Transients Programs”, Proceedings of the 7th and a Associate Professor. His special field of interest
Power Systems Computation Conference, Lousane, is modelling of electrical machines and control of power
pp. 900-904, July 1981. systems.
Dr. Dolinar is a member of IEEE.
[lo] I. J . Binard, I. J. Maun, “Hysteresis Model for
Power Transformer Transient Simulation Program”,
IMACS-TC1’90 Nancy, pp. 539-544, September
1990. Joie Pihler was born in Ptuj, Slovenia on November
[ll] I. J . Binard, I. J . Maun, “Power Transformer 26, 1955. He received his B.S. and M.S. degrees from
Simulation Including Inrush Currents and Inter- the Faculty of Technical Sciences, Maribor, in 1978 and
nal Faults”, IMACS-TC1’90 Nancy, pp. 57-62, 1991, respectively, all in electrical engineering.
September 1990. In 1978 he was employed by the TSN (production of
switch gears) in Maribor. He was engaged in research
[12] H. L. Nakra, T. H. Barton, “Three Phase Trans-
and development activities related to switch gears. Since
former Transients”, IEEE Power Engineering Soci-
1988 he has been working at the Faculty of Technical Sci-
ety, T74 243-2, pp. 1810-1819, February 1974.
ences in Maribor, Department of Electrical Engineering,
[13] D. N. Ewart, “Digital Computer Simulation Model Institute for Power Systems as a researcher.
of a Steel-core Transformer”, IEEE Transactions on His special field of interest is modelling of electrical
Power Delivery, vol. PWRD-1, No. 3, pp. 174-183, machines and power systems protection.
July 1986. He is a Student member of IEEE.
[14] J. Pihler, Design of a Digital Digerential Protection
f o r Power Transformer, M. Sc. thesis in Slovene,
Faculty of Technical Sciences Maribor, February
Bojan GrEar (AM’83) was born in Maribor, Slovenia,
1991.
on December 7,1952. He received his B.S., M.S., and
[15] C. de Boor, A Practical Guide to Splines, Springer- Ph.D. degrees from the Faculty of Technical Sciences,
Verlag, New York - Heidelberg - Berlin, 1978. Maribor, in 1978, 1983 and 1988, respectively, all in elec-
trical engineering.
Drago Dolinar (M’81) was born in Celje, Slovenia on Since 1983 he has been working at the Faculty of Tech-
November 4, 1953. He received his B.S., M.S., and Ph.D. nical Sciences in Maribor, Department of Electrical En-
degrees from the Faculty of Technical Sciences, Maribor, gineering, Institute for Power Systems in the field of au-
in 1978, 1980, and 1985, respectively, all in electrical tomation, control and power systems protection.
engineering. Dr. Grtar is associate member of IEEE.