Vous êtes sur la page 1sur 13

Electr Eng (2008) 90:229–241

DOI 10.1007/s00202-007-0071-5

ORIGINAL PAPER

Single-phase transformer model including magnetic hysteresis


and eddy currents
A. D. Theocharis · J. Milias-Argitis · Th. Zacharias

Received: 21 December 2006 / Accepted: 21 April 2007 / Published online: 24 May 2007
© Springer-Verlag 2007

Abstract In this paper, a single-phase core-type and In a first group the description of the transformer is based
shell-type transformer model is proposed on the level of on the electromagnetic field analysis in the entire volume of
state equations. The magnetic hysteresis and the eddy cur- the magnetic core using numerical analysis as the finite ele-
rents effects are successfully included based on the analytical ments method [1–4]. These methods are suitable when we
description of the magnetic core topology. Predicted values focus on the design of the transformer but are impractical for
from simulations are in very good agreement with published the calculation of transients and steady state response since
measurements and hence, the developed transformer model they would give time-expensive simulations.
is a powerful tool for transient and steady-state studies. In a second group the transformer modeling is based on the
principle of duality between electric and magnetic circuits
Keywords Eddy currents · Magnetic circuits · and representative references are included in [5–9]. These
Magnetic hysteresis · State equations · Transformers descriptions do not follow mathematical approaches but they
deal with an electrical equivalent circuit, which sometimes
may be very complicated where many ideal transformers are
used.
1 Introduction
The models in a third group are based on the mathematical
description of the magnetic core topology and the windings
Transformer modeling is a very difficult task due the intro-
configuration [10–17]. They are represented by a set of state
duction of the magnetic core characteristics as the saturation
equations in which the coupling between the electric and
effect, the hysteresis and the eddy currents losses as well as
the magnetic circuit takes into account the dimensions and
to the magnetic core topology and the windings configura-
the material characteristics of the magnetic core. The main
tion. Also, in cases where the skin and eddy currents effects
difference between them is the way in which the complete
in the coils and the capacitive effects have to be included the
set of equations is derived.
representation is very complicated. Thus, a variety of meth-
The models in a fourth group are incorporated in the
ods for the transformer description have been developed in
EMTP/ATP-type programs as the BCTRAN Model [18] or
the literature in the past decades, which can be classified into
the STC Model [19]. The frequent problem of these models is
several groups.
the numerical instability that they present and several efforts
have been done in this direction [20,21]. Papers that belong
to the second [6], to the third [22], [23] group or individual
This work was supported by the Research Committee of the
works [24] are introduced in this category. Also, in [25] there
University of Patras under Grant B106. is a successful implementation of a transformer model in the
PSPICE program.
A. D. Theocharis (B) · J. Milias-Argitis · Th. Zacharias In this paper, single-phase core-type and shell-type trans-
Department of Electrical and Computer Engineering,
former models of the third group are presented, which include
University of Patras, Patras, 26500 Rion, Greece
e-mail: theoxar@ece.upatras.gr hysteresis and eddy currents effects of the magnetic core.
URL: http://www.ece.upatras.gr Particularly, the magnetic hysteresis is introduced by two

123
230 Electr Eng (2008) 90:229–241

distinct models (a) using the Jiles-Atherton macroscopic given by


model of hysteresis [26–28] or (b) using the mathemati-
N12
cal model of hysteresis presented by Tellinen. [29]. Also, core-type : re = (Ap.8)
the eddy currents effects, as they are described by Bertotti ke
[30,31], are incorporated by a non-linear resistor in the pri- ⎧ ⎫
mary winding side of the transformer, which is dependent on ⎨3  2 ⎬
∂Φ j
the magnetic core topology of the transformer that is under shell-type : re = N12 / kej (Ap.10)
⎩ ∂Φ1 ⎭
consideration. The proposed models are verified by measure- j=1
ments published elsewhere for the transients and the steady where the terms ke and kej are defined in the Appendix.
state. The equations of the electrical part of the transformers in
Fig. 1 can be written in matrix form as
⎡ ⎤ ⎡ ⎤⎡ i 1 ⎤
v1 r1 + re −re 0
2 Description of the models ⎢ ⎥ ⎢ ⎥
⎣ v1 ⎦ = ⎣ r1 0 0 ⎦⎣ i 1 ⎦
−v2 0 0 r2 i2
2.1 Electrical equations
⎡ ⎤ ⎡ i1 ⎤ ⎡ ⎤
0
L1 0 0
The core-type and shell-type transformers in Fig. 1 are ⎣ ⎦ d ⎢ ⎥ d⎢ ⎥
+ L1 0 0 ⎣ i ⎦ + ⎣ ψ1 ⎦ (1)
examined where v1 and v2 are the voltages at the trans- dt 1 dt
0 0 L2 ψ2
formers terminals, N1 , i 1 , r1 , L 1 and N2 , i 2 , r2 , L 2 are the i2
number of turns, the currents, the resistances and the leak- where ψ1 and ψ2 are the flux linkages.
age inductances of the primary and the secondary windings, The time derivatives of flux linkages are expressed as
respectively. The resistance re , for both of the transformers, non-linear functions of the time derivative of the windings
introduces the eddy currents effects in each case. The deter- currents of the transformers in the form
mination of the resistance re , for both of the transformers,    
is described in detail in the Appendix of this work, in order d ψ1 d i1
= Ld
to simplify the paper’s representation. Particularly, using the dt ψ2 dt i 2
description of the eddy currents effects proposed by Berto- where the matrix Ld is
tti, the resistance re is determined and it is depended on the  
magnetic core dimensions and topology of the transformer L d11 L d12
Ld =
that is under consideration in each case. This resistance re is L d21 L d22
and is depended on the topology of the magnetic core in each
case. The elements L d11 and L d22 represent the self induc-
tances and the elements L d12 and L d21 represent the mutual
inductances. The electrical equations can be represented in
compact form as
DI = L−1 (E − RI) (2)
where D = d/dt is the time derivative operator, E =
  T
[v1 v1 − v2 ]T , I = i 1 i 1 i 2 ,
⎡ ⎤ ⎡ ⎤
r1 + re −re 0 L1 0 0
R = ⎣r1 0 0 ⎦ and L = ⎣L 1 L d11 L d12 ⎦.
0 0 r2 0 L d21 L 2 + L d22

2.2 Magnetic equations and determination of the matrix Ld

2.2.1 The matrix Ld of the core-type transformer

Fig. 1 a A core-type transformer. b A shell-type transformer. Ac , lc , Φ The corresponding magnetic circuit of the magnetic
are the cross-section, the mean length paths of the magnetic lines and core in this case is shown in Fig. 2 where F1 , F2 are the
the magnetic flux, respectively magnetomotive forces of the primary and the secondary

123
Electr Eng (2008) 90:229–241 231

where the term µd is defined as


db
µd = (10)
dh
and is called incremental permeability.
Substituting the term dµ from (9) in (7), the differential
of the magnetic potential df is
 
Fig. 2 The electrical equivalent of the magnetic circuit of the core-type lc dΦ Φµ Φµ
transformer df = − 2 d dh + 2 dh (11)
Ac µ µ h µ h

The differential of the magnetic flux Φ can be determined as


windings respectively, Rm the magnetic resistance and f the
follows
magnetic potential. The time derivatives of the flux linkages
ψ1 and ψ2 are expressed via the number of turns of the wind- Φ = b Ac (12)
ings N1 and N2 and the core flux Φ and can be written as
dΦ = Ac db = Ac µd dh (13)
   
d ψ1 N1 dΦ Thus, by substituting in (11) the terms of the magnetic flux
= (3)
dt ψ2 −N2 dt from (12) and using Eqs. (8), (10) and (13), the derivative of
the core flux with respect to the magnetic potential results
The total magnetomotive force F is equal to the difference    
of the magnetomotive forces due to the primary and the sec- dΦ d f = µd Ac lc = G md (14)
ondary windings currents
where G md is called incremental magnetic conductance.
F= N 1 i 1 − N2 i 2 (4) From the magnetic circuit in Fig. 2 the partial derivative
of the magnetic potential with respect to the total magneto-
The core flux Φ is, in general, a non-linear function of the motive force results as
total magnetomotive force and can be written as 
∂f ∂F = 1 (15)
dΦ ∂Φ ∂ f dF
= (5) and from Eq. (4) the time derivative of the total magnetomo-
dt ∂ f ∂ F dt
tive force is equal to
The differential of the magnetic potential can be determined
dF di  di 2
from the equation = N1 1 − N2 (16)
dt dt dt
f = Φ Rm = Φ (lc /µAc ) (6) Substituting (14), (15) and (16) into (5), the time deriva-
tive of the core flux is written as
as
dΦ di  di 2
lc µdΦ − Φdµ = N1 G md 1 − N2 G md (17)
df = (7) dt dt dt
Ac µ2
Substituting Eq. (17) into Eq. (3), the matrix Ld of the core-
where µ = b(h)/ h is the magnetic permeability of the core- type transformer results as
material, b the magnetic flux density and h the magnetic field  2 
N1 G md −N1 N2 G md
strength of the core. Ld =
The derivative of the permeability µ with respect to the −N1 N2 G md N22 G md
magnetic filed strength h inside the core can be determined Finally, the Eqs. (2) and (17) are the complete set of state
from equations for the core-type transformer model.

µ=b h (8)
2.2.2 Determination of the matrix Ld of the shell-type
as transformer

dµ µd − µ Figure 3 shows the equivalent magnetic circuit of the shell-


= (9)
dh h type transformer. The time derivatives of the flux linkages ψ1

123
232 Electr Eng (2008) 90:229–241

The Kirchhoff’s law for the magnetic potentials for the


left loop, which includes parts 1 and 2 of the core, declares
that

f1 + f2 = F

The partial derivative with respect to the total magnetomotive


F, for this relation results as

∂ f1 ∂ f2
+ =1 (25)
Fig. 3 The electrical equivalent of the magnetic circuit of the shell- ∂F ∂F
type transformer
On the same way, from the right loop of the magnetic circuit,
which includes parts 1 and 3, reveals that
and ψ2 are expressed via the number of turns of the windings
N1 and N2 and the core flux Φ1 and are written as f1 + f3 = F
    ∂ f1 ∂ f3
d ψ1 N1 dΦ 1 + =1 (26)
= (18) ∂F ∂F
dt ψ2 −N2 dt
Equations (24), (25) and (26) constitute a system with
The time derivative for the core flux in the part j, ( j = unknown quantities the three partial derivatives of the mag-
1, 2, 3), of the core can be expressed as ∂f
netic potentials ∂ F1 , ∂∂ fF2 and ∂∂ fF3 . In compact matrix form, the
dΦ j ∂Φ j ∂ f j dF above system is represented as
= (19)
dt ∂ f j ∂ F dt
f = A−1
1 A2 (27)
where the total magnetomotive force F is given by (4) and
the magnetic potential f j across the jth part of the magnetic where  is the partial derivative operator  = ∂/∂ F and the
circuit is expressed in terms of the magnetic resistance Rm j matrices
⎡ f, A1 and A⎤2 are f = [ f 1 f 2 f 3 ]T , A1 =
and the core flux Φ j as −G md1 G md2 G md3
⎣ 1 1 0 ⎦ and A2 = [0 1 1]T . From (27),
f j = Rm j Φ j (20) 1 0 1
one can write
The Kirchhoff’s law for the core fluxes at the node of the
∂ f1 G md2 + G md3
magnetic circuit declares that = (28)
∂F G md1 + G md2 + G md3
− Φ1 + Φ2 + Φ3 = 0 (21) ∂ f2 G md1
= (29)
∂F G md1 + G md2 + G md3
The partial derivative with respect to the total magnetomotive
force of this equation results as ∂ f3 G md1
= (30)
∂F G md1 + G md2 + G md3
∂Φ1 ∂ f 1 ∂Φ2 ∂ f 2 ∂Φ3 ∂ f 3
− + + =0 (22)
∂ f1 ∂ F ∂ f2 ∂ F ∂ f3 ∂ F Substituting Eqs. (23), (28), (29), (30), (16) into (19), the
time derivatives of the core fluxes, in a compact matrix form,
Introducing the incremental conductance for each one of the are
three parts of the core, in the same manner that the incremen-
tal conductance has been defined in the case of the core type D = DI (31)
transformer
∂Φ j Acj where the matrices are  = [Φ1 Φ2 Φ3 ]T ,
= G md j = µd j (23) ⎡ ⎤
∂fj lcj N1 G md1 (G md2 +G md3 ) N2 G md1 (G md2 +G md3 )
0 G md1 +G md2 +G md3 − G md1 +G md2 +G md3
⎢ ⎥
Eq. (22) is written as ⎢ ⎥
=⎢
⎢0
N1 G md1 G md2
G md1 +G md2 +G md3 − N2 G md1 G md2
G md1 +G md2 +G md3


∂ f1 ∂ f2 ∂ f3 ⎣ ⎦
− G md1 + G md2 + G md3 =0 (24) 0 N1 G md1 G md3 N2 G md1 G md3
− G md1
∂F ∂F ∂F G md1 +G md2 +G md3 +G md2 +G md3

123
Electr Eng (2008) 90:229–241 233

By substitution of the first equation of the system (31), 2.3.1 Incorporation of the J-A model of hysteresis
which corresponds to the time derivative of the flux Φ1 , into
Eq. (18) the matrix Ld of the shell-type transformer results Using the well-known expression b = µ0 (h + M) where the
as magnetization M is the sum of two components, the revers-
⎡ 2 ⎤
N1 G md1 (G md2 +G md3 ) N1 N2 G md1 (G md2 +G md3 ) ible magnetization Mrev and the irreversible magnetization

⎢ G md1 +G md2 +G md3 G md1 +G md2 +G md3 ⎥ Mirr , the term µd can be written as
Ld = ⎣ ⎦
− N1 GN2 G md1 (G md2 +G md3 ) N 2 md1 (G md2
2G +G md3 )  
md1 +G md2 +G md3 G md1 +G md2 +G md3 dM rev dM irr
µd = µo 1 + + (35)
dh dh
Therefore, Eqs. (2) and (31) constitute the complete set of
state equations for the shell-type transformer model. where µo is the permeability of free space. According to J-A
At this point, model
 it shouldbe noted that, in view of (Ap.10),
the terms ∂Φ2 ∂Φ1 , ∂Φ3 ∂Φ1 can be determined as follows dM irr Man − Mirr
= (36)
dh kδ − a 1 (M an − Mirr )
∂Φ2 ∂Φ2 ∂ f 2 ∂ F ∂ f 1  
= (32) dMrev dMan dMirr
∂Φ1 ∂ f 2 ∂ F ∂ f 1 ∂Φ1 =c − (37)
dh dh dh
Using Eqs. (23) for j = 1 and j = 2 as well as Eqs. (28) and and the term Man reflects the anhysteretic magnetization. In
(29), then (32) becomes terms of h it is represented by the Langevin equation as
   
∂Φ2 G md2 h + a1 M a2
= (33) Man (h) = M s coth − (38)
∂Φ1 G md2 + G md3 a2 h + a1 M

In an analogous manner, one can writte and the term dMan /dh is derived by differentiation of the
above equation. The parameters a1 , a2 , c and k are constants
∂Φ3 G md3 for the material being used, Ms is the saturation magnetiza-
= (34) tion and the directional parameter δ is dependent on the sign
∂Φ1 G md2 + G md3
of dh/dt.
The vector of the state variables for the core-type trans-
The matrix Ld , in both cases, is dependent on the incre-  
T
mental permeability µd (through G md ) representing the core former model is i 1 i 1 i 2 Mrev Mirr h . Equations (2)
material characteristic. When the core material is described are the state equations of the electrical part of the transformer
by its magnetization curve h = g(b), the incremental perme- while, the state equations of the magnetic core are
ability is caclulated in each time step as µd = (dh/db)−1 .
dM rev dM rev dh
When magnetic hysteresis is to be taken into account, the = (39)
expression for µd depends on hysteresis model. dt dh dt
dM irr dM irr dh
= (40)
dt dh dt
2.3 Incorporation of the magnetic hysteresis
The time derivative of the magnetic field strength dh/dt is
There are several descriptions for the magnetic hysteresis evaluated from
which, in general, can be classified into three main cate- dh 1 dΦ
gories. In the first, the hysteresis loop is described through = µ−1
d (41)
dt Ac dt
the physical mechanisms in microscopic scale. In the second
category, experimental data are fitted by mathematical func- In the case of the shell-type transformer, the vector of the

tions to obtain the major hysteresis loop and minor loops. state variables is [i 1 i 1 i 2 Mrev1 Mrev2 Mrev3 Mirr1
In the third category [32], the magnetic hysteresis is ana- Mirr2 Mirr3 h 1 h 2 h 3 ]T . The incremental permeabilities
lyzed on a macroscopic level and equations are revealed for are determined from Eq. (35) for each one of the three parts
the magnetization M. For the incorporation of the hysteresis of the magnetic core of the transformer. Equations (2) are
loop in this work two different methods are used. At first, the state equations of the electrical part of the transformer
the Jiles-Atherton (J-A) model is selected among the mod- while, the state equations of the three parts of the magnetic
els of the last category. Next, the model that is proposed by core have the same form as Eqs. (39) and (40). Also, the time
Tellinen (TLN) is selected among the models of the second derivatives of the magnetic field strengths in the three parts
category. are of the same form as Eq. (41).

123
234 Electr Eng (2008) 90:229–241

2.3.2 Incorporation of the TLN model of hysteresis


200 161.7 A
161.7 A
The ascending branch of the major hysteresis loop is denoted 150

Inrush Current (A)


as basc (h) and the descending branch of the major hysteresis
loop is denoted as bdesc (h). It is well known that due to the 100 65 A
symmetry about the origin of the axes 60 A 38 A
25 A
50 39.6 A
31.3 A
basc (h) = −bdesc (−h) (42)
0
The incremental permeability µd through which the hystere-
sis loop is introduced in the previously defined transformers -50
0.00 0.01 0.02 0.03 0.04 0.05 0.06
models, can be evaluated as
time (s)
 
bdesc (h) − b dbasc (h)
µd = µ0 + − µ0 dh > 0 Fig. 4 Inrush current waveform. The peak values from [33] (bold) and
bdesc (h) − basc (h) dh from the simulation (normal)
(43)

and
  and using the proposed transformer model, all the measured
b − basc (h) dbdesc (h) waveforms have been confirmed. A representative result is
µd = µ0 + − µ0 dh < 0
bdesc (h) − basc (h) dh shown in Fig. 4, where the four first cycles of the simulated
(44) inrush current and the peak measured values are shown. By
comparison of the peak measured and simulated values as
Using this model of hysteresis, the state variables of the well as by comparison with the published measured wave-
 T
core-type transformer are i 1 i 1 i 2 Φ h . Equations (2) are form in [33] (L air = 0.85643 mH) the agreement is very
the state equations of the electrical part of the transformer good. The slight differences at the peak values and the dis-
while, Eq. (17) is the equation for the magnetic core and similarity in the regions of the minimum values are due to
the term dh/dt is evaluated from (41). The state variables of the fact that only the magnetization curve was available.
  T
A further agreement in this region would be accomplished if
the shell-type transformer are i 1 i 1 i 2 Φ1 Φ2 Φ3 h 1 h 2 h 3 .
the hysteresis loop were used.
Equations (2) are the state equations of the electrical part
of the transformer while, Eqs. (31) are the state equations
for the magnetic core and the terms dh j /dt are evaluated
from (41). 3.2 Second case: TLN model of hysteresis

In this case, hysteresis loop and eddy currents are taken into
account. Data for the core material are taken from [34]. In
3 Results [34] the core material power losses Pc as well as the hyster-
esis loop power losses Ph per unit mass are reported. They
Simulations have been conducted in order to verify the pro- correspond to a sinusoidal excitation with 50 Hz frequency at
posed transformers models. Three cases are investigated. maximum flux densities 1.7 and 1.5 T. Moreover, the major
In the first case, the core material is represented by its hysteresis loop in [34] is given as a graph on the b − h plane.
magnetization curve. Consequently, the ascending branch of the hysteresis loop of
In the second case, the core material is represented by its the core material is approximated by
hysteresis loop using the TLN model.
In the third case, the core material is represented by its
hysteresis loop using the J-A model. 5
q=1 w1q (h − H c )q
b(h) = 4 (45)
q=0 w2q (h − H c )q
3.1 First case: magnetization curve model

In this case, an unloaded, single-phase, shell-type transformer where Hc is the coercive force. The constants w1q and w2q
is considered. Eddy currents are not taken into account. The for h < Hc and h > Hc are given in Table 1. The descend-
set of state equations are (2) and (31). In [33] a number ing branch is determined from Eq. (42). The corresponding
of measured waveforms for the inrush current are reported curve is shown in Fig. 5. The hysteresis loop is incorporated

123
Electr Eng (2008) 90:229–241 235

Table 1 Constants w1q and w2q Table 3 Magnetic cores dimensions


q h > Hc h < Hc Length (m) Cross-section (10−4 m2 )

w1q w2q w1q w2q Core-type


1.4 63.36
0 – 627.98 – 330.28
Shell-type
1 159.4 −271.3 217.78 48.18
Part 1 0.35 63.36
2 −93.68 −37.6 105.84 −50.12
Part 2 1.05 31.68
3 5.01 2.6 3.13 −1.32
Part 3 1.05 31.68
4 –0.14 −0.0842 0.33 0.191
5 −1.13 × 10−5 – 7.98 × 10−6 –

Table 4 Comparison of the core power losses for Bmax = 1.7 T


2.0
Magnetic Flux Published in [34] Predicted
Density (T) 1.5
Core-type Shell-type
1.0
Ph (w/kg) 0.6 0.615 0.613
0.5 Pe (w/kg) 0.9 0.927 0.926
Pc (w/kg) 1.5 1.542 1.539
-300 -200 -100 100 200 300
-0.5 Magnetic Field
Table 5 Comparison of the core power losses for Bmax = 1.5 T
-1.0 Strength (A/m)
Published in [34] Predicted
-1.5
Core-type Shell-type
-2.0
Ph (w/kg) 0.37 0.388 0.386
Fig. 5 Hysteresis loop. Data points from [34] and the fitted curve Pe (w/kg) 0.65 0.673 0.671
Pc (w/kg) 1.02 1.061 1.057
Table 2 Windings data
Primary Secondary
component of power losses due to eddy currents is a signifi-
Resistance () 64.2 0.0457 cant percentage of the total power core losses.
Leakage Inductance (10−3 H) 243.6 0.1732
Number of turns 4165 111 3.2.2 Eddy currents effect on the primary winding current
waveform

The eddy currents effect on the primary winding current


using the TLN model of hysteresis. The term (σ Gdτ Ho )1/2 , waveform of the core-type tranformer is examined. The sec-
which appears in the eddy currents resistance, is calculated ondary winding is considered to be open. Figure 6 shows the
from the analysis of the measurements as it is shown in the nonlinear characteristic of the eddy currents resistance re on
Appendix. The set of state equations are (2), (17) and (41) the i − v plane. Figure 7 shows the primary winding current
for the core-type and (2), (31) and (41) for the shell-type. waveforms when the eddy currents are neglected as well as
Data for the two types of transformers are shown in Tables 2 when the eddy currents are taken into account (maximum
and 3. The amplitude of the source voltage is adjusted for flux density 1.5 T). As it is shown from the results, the eddy
maximum flux density in each case. currents produce significant changes on the primary winding
current waveform.
3.2.1 Core losses
3.3 Third case: J-A model of hysteresis
Loaded, core-type and shell-type transformers are simulated
in order to predict the magnetic core power losses in each In this case, simulation results are presented for a loaded,
case. The load at the secondary side is an R L series branch shell-type transformer. The load at the secondary side is an
with values 2.7  and 5 mH. Simulation results and mea- R L series branch with values 1.9  and 8.064 mH in parallel
sured values are shown in Tables 4 and 5 where a very with a capacitance 0.5121 mF for the compensation of the
good agreement is achieved. It is worthwhile to note that the reactive power to a power factor value 0.9. The amplitude of

123
236 Electr Eng (2008) 90:229–241

200 (a)
15

Voltage (kV)
Primary
150 Secondary
10
100
5

Current (A)
50

-10 -8 -6 -4 -2 2 4 6 8 10 0
-5 Current (mA)
-50
-10
-100
-15
0.00 0.05 0.10 0.15 0.20 0.25 0.30

Fig. 6 The nonlinear characteristic of the eddy currents resistance re time (s)
in the case of the core-type transformer for maximum flux densities 1.5 200
T (dashed line) and 1.7 T (solid line) (b) R-L branch
150 Capacitor

100
20
50

Current (A)
15
0
10
Primary Current (mA)

-50

5 -100

0 -150

-200
-5
0.00 0.05 0.10 0.15 0.20 0.25 0.30

-10 time (s)


2.0
-15 (c) Part 1
1.5
Magnetic Flux Density (T)

-20
1.92 1.93 1.94 1.95 1.96 1.97 1.98 1.99 1.0

time (s) 0.5

Fig. 7 The primary winding current waveform of the unloaded core- 0.0
type transformer when the eddy currents are neglected (solid line) or
-0.5
included (dotted line)
-1.0

-1.5
the source voltage is 12.25 kV. The parameters of the core
-2.0
material are those in [28] (Fig. 1). The J-A model of hys- -10000 -5000 0 5000 10000
teresis is used, while the eddy currents are neglected. More- Magnetic Field Strength (A/m)
over, for the magnetic core two cases are considered. First, a
symmetrical magnetic core and next an asymmetrical one. Fig. 8 Transient response for the shell-type transformer with a sym-
metrical magnetic core. a Primary and secondary windings currents.
The set of state equations are (2), (39), (40) and (41).
b Load currents. c The b(h) trajectory of the part 1 of the magnetic
core
3.3.1 Shell-type transformer with symmetrical
magnetic core
3.3.2 Shell-type transformer with asymmetrical
The parameters of the transformer are given in Tables 2 and 3. magnetic core
Figures 8 and 9 show transient and steady state responses
of the transformer respectively. Due to the symmetry of the In order to present simulation results of an asymmetrical
magnetic core, the b(h) trajectories of the parts 2 and 3 are shell-type transformer, the length as well as the cross-
identical to that of part 1. The secondary winding current section of the part 3 of the magnetic core in the previous
shows a slight difference from the sinusoidal shape due to case have been changed. The new values are lc3 = 1.5lc2
the presence of the capacitor. and Ac3 = 0.5Ac2 where lc2 and Ac2 are those in Table 3.

123
Electr Eng (2008) 90:229–241 237

250 (a)
Primary (N1/N2) Primary
Secondary 200 Secondary
(a) R-L branch
150
120 Capacitor

Current (A)
100
80
50
40
Current (A)

0
0
-50
-40
-100
-80
-150
0.00 0.05 0.10 0.15 0.20 0.25
-120
time (s)
3.00 3.01 3.02 3.03 3.04 3.05 200 (b) R-L branch
time (s) 150 Capacitor
100
1.5
(b)
Part 1 50

Current (A)
Magnetic Flux Density (T)

1.0 0

-50
0.5
-100
0.0
-150
-0.5 -200

-250
-1.0
0.00 0.05 0.10 0.15 0.20

-1.5 time (s)

-6000 -4000 -2000 0 2000 4000 6000 2.0 (c)


Magnetic Field Strength (A/m) 1.5
Magnetic Flux Density (T)

1.0 Part 1
Fig. 9 Steady state response for the shell-type transformer with a sym-
metrical magnetic core. a Current waveforms. b The b(h) trajectory in Part 2
0.5
the part 1 of the magnetic core Part 3
0.0

-0.5

Transient and steady state responses are shown in Figs. 10, -1.0
11, respectively. Due to the asymmetry of the magnetic core, -1.5
the operating points on the b − h plane of the parts 2 and
-2.0
3 have been inserted in the saturation region. The windings -5000 0 5000 10000
currents have been further distorted in comparison with the Magnetic Field Strength (A/m)
symmetrical case hence, J-A model has been easily incorpo-
rated into the transformers models of this work. Fig. 10 Transient response for the shell-type transformer with an
asymmetrical magnetic core. a Primary and secondary windings cur-
rents. b Load currents. c The b(h) trajectories of the three parts of the
magnetic core

4 Conclusions

In this paper, detailed models of single-phase, core-type For the validation of the model, simulation results are com-
and shell-type transformers for a state space formulation are pared with published measuremets. The developed model
presented. The proposed models incorporate the topology, permits the various parts of the core having different dimen-
material characteristics and size of the core. The magnetic sions thereby leading to an assymetrical core.
hysteresis of the core material is taken into account using Although the paper is written for single phase transform-
either the model of Jiles-Atherton or the mathematical for- ers the developed theory is suitable for the simulation of any
mulation of Tellinen. The eddy currents effects are modeled three-phase transformer with uniform magnetic flux density
by a nonlinear resistor based on the work of Bertotti. in the various parts of the core.

123
238 Electr Eng (2008) 90:229–241

Primary (N1/N2)
Secondary
160 (a) Load R-L
Capacitor
120

80

40
Current (A)

-40

-80

-120

-160
3.00 3.01 3.02 3.03 3.04 3.05
time (s)
Fig. 12 A magnetic lamination with length l p , thickness d, cross-sec-
2.0 (b) tion A p , width τ and electrical conductivity σ in which the magnetic
1.5 flux density b p is considered uniform
Part 1
Magnetic Flux Density (T)

1.0 Part 2
Part 3
0.5
where h e.c represents the classical eddy currents and given
0.0 by
-0.5
σ d 2 db p
-1.0 h e.c (t) = (Ap.2)
12 dt
-1.5

-2.0 and h e.e represents the excess eddy currents and given by
-10000 -5000 0 5000 10000
Magnetic Field Strength (A)  1/2
db p
Fig. 11 Steady state response for the shell-type transformer with an h e.e (t) = (Gdτ H o σ )1/2 (Ap.3)
asymmetrical magnetic core. a Current waveforms. b The b(h) trajec-
dt
tories in the three parts of the magnetic core
The term G = 0.1356 is a dimensionless constant and Ho is
a parameter with dimensions [A/m] representing the internal
potential experienced by domain walls in the magnetic lam-
The proposed model could also be very useful for the
ination and can be a function of the maximum flux density
simulation of power converters including transformers and
Bmax .
interconnecting a linear or nonlinear source with the power
In the case of the core type transformer the eddy currents
system [35].
magnetic field strength in the magnetic core is

 1/2
σ d 2 db 1/2 db
h e (t) = + (Gdτ H o σ ) (Ap.4)
Appendix 12 dt dt

Determination of the eddy currents resistance where b is the magnetic flux density of the core. The instan-
taneous power core losses are estimated by
In the classical model of the eddy currents effects, the mag-
netic domains of the material are neglected. This is too gross db
pe = h e Ac lc (Ap.5)
a simplification and the consequences of the domain walls dt
are the excess eddy currents effects, which have been inves-
tigated in detail by Bertotti [30,31]. According to Bertotti, in and substituting Eq. (Ap.4) and the magnetic flux density as
a part of length l p of the magnetic lamination in Fig. 12, the b = Φ/Ac the power core losses are written as
total magnetic field due to eddy currents effects is
 
ke dΦ 2
pe = N 1 (Ap.6)
h e (t) = h e.c (t) + h e.e (t) (Ap.1) N12 dt

123
Electr Eng (2008) 90:229–241 239

where
  1/2  −1/2 
σ d 2 lc Gdτ H o σ l 2c dΦ
ke = + (Ap.7)
12 Ac Ac dt
Fig. 13 Equivalent magnetic branch of the part with length l p of the
lamination shown in Fig. 12
The coefficient of the parenthesis of Eq. (Ap.6) has dimen-
sions −1 and represents the eddy currents conductance
which is referred to the primary winding. Hence, the eddy
currents resistance in the case of the core-type transformer is

N12
re = (Ap.8)
ke
In the same manner, in the case of the shell-type trans-
former the eddy currents instantaneous power core losses
are the sum of the losses in each part of the magnetic core
and are written as
⎡ ⎤
3    
kej ∂Φ j 2 dΦ 1 2
pe = ⎣ ⎦ N1 (Ap.9)
N 2 ∂Φ1
j=1 1
dt

where the subscript j declares the number of the part of the


magnetic core. Thus, the eddy currents resistance which is
referred to the primary winding of the shell-type transformer
is defined as Fig. 14 The electrical equivalent of the magnetic circuit of the core-
⎧ ⎫ type transformer with “n” laminations including the eddy currents
⎨ 3  2 ⎬ according to the representation in the Fig. 13
∂Φ j
re = N12 / kej (Ap.10)
⎩ ∂Φ1 ⎭
j=1
Supposing that the magnetic core of the core-type trans-
The resistances in Eqs. (Ap.8) and (Ap.10) are two suit- former consists of “n” laminations of length lc , the magnetic
able predetermined resistances to fit the magnetic core power circuit, which includes the eddy currents effects according to
losses in each case. Following a detailed representation of the above representation, results as that shown in Fig. 14. In
the magnetic circuit of the magnetic core for each one of the the same manner, the magnetic circuit of the sell-type trans-
transformers, the relations (Ap.8) and (Ap.10) can be found. former supposing that each one of the three parts consist of
The part of length l p in Fig. 12 is represented by an equiva- “n” laminations of length lcj is as Fig. 15 shows.
lent magnetic branch in order to prove this interpretation. The In the case of the core-type transformer, the magnetic
magnetomotive force Fep due to eddy currents in the length resistance Rmi is given by
l p of the magnetic lamination in Fig. 12 can be defined as lc
Rmi = (Ap.13)
µi Aci
Fep = h e l p (Ap.11)
where Aci the cross-section, µi the magnetic permeability
and by substitution of the Eqs. (Ap.1), (Ap.2) and (Ap.3) it and Φi the magnetic flux in the ith lamination. The magne-
is written as tomotive force Fei is given by Eq. (Ap.12) where the term
  Φ p is replaced by Φi and the kep is replaced by kei which
dΦ p
Fep = kep (Ap.12) is referred to the quantities of the ith lamination. Since the i
dt laminations of the core are identical one can write
where Φ p is the magnetic flux of the part under consideration 
Φi = Φi+1 = Φ n (Ap.14)
and the term kep is in the same form as ke in Eq. (Ap.7) with 
the difference that the quantities are referred to the part under Aci = Aci+1 = Ac n (Ap.15)
consideration of the lamination. Using Eqs. (Ap.14) and (Ap.15), for the kei factors and
Thus, the part of length l p of the magnetic lamination in the Fei magnetomotive forces it results
Fig. 12 can be represented as a magnetic branch with mag-
netic flux Φ p and a magnetic resistance Rmp in series with kei = kei+1 = ke (Ap.16)
the magnetomotive force Fep as Fig. 13 shows. Fei = Fei+1 = Fe (Ap.17)

123
240 Electr Eng (2008) 90:229–241

Fig. 15 The electrical equivalent of the magnetic circuit of the shell-


type transformer with “n” laminations including the eddy currents
according to the representation in the Fig. 13

where ke is given from (Ap.7) and Fe from (Ap.12) by replac-


ing kep with ke and Φ p with Φ. Applying the Ampere’s law Fig. 16 a A first equivalent of the magnetic circuit in Fig. 14 due to
in the loop that contains the branches i and i + 1 of the mag- Eq. (Ap.20). b The equivalent magnetic circuit for the core-type trans-
former including eddy currents
netic circuit in Fig. 14, it arises that the magnetic resistances
of the loops are equal
Rmi = Rmi+1 (Ap.18) the total magnetomotive force is written as

and the magnetic permeabilities are equal too F = N 1 i 1 − N2 i 2 (Ap.24)
µi = µi+1 (Ap.19)
and the eddy currents resistance represented by Eq. (Ap.8)
Hence, the magnetic potentials f i and f i+1 across the mag- revealed.
netic resistances are equal In the case where only the classical eddy currents effects
are taken into account the eddy currents resistance presented
f i = f i+1 (Ap.20)
in Eq. (Ap.8) is identical to that in [36].
and the magnetic resistances can be considered to be in par-
allel. Thus, the magnetic circuit is simplified in that shown Determination of the term (σ Gdτ H o)1/2
in Fig. 16a. The parallel combination of the magnetic resis-
tances further simplifies the magnetic circuit in that shown in The total power losses per unit mass Pc in a magnetic lami-
Fig. 16b. The total magnetomotive force F from the primary nation are
and the secondary windings currents as well as from the eddy
currents is given by Pc = Ph + Pe (Ap.25)

F = N 1 i 1 − N2 i 2 − ke (Ap.21) where Ph are the dc hysteresis loop power losses and Pe
dt
are the eddy currents power losses. The eddy currents power
Rearranging the terms in the above equation results in losses are the sum of the classical, Pcl and the excess, Pexc
  
ke dΦ eddy currents losses
F = N 1 i 1 − 2 N1 − N2 i 2 (Ap.22)
N1 dt
Pe = Pcl + Pexc (Ap.26)

Defining the current i 1 as
  In the case of a sinusoidal voltage supply, which means that
 ke dΦ the magnetic flux density b p is also sinusoidal, the classical
i 1 = i 1 − 2 N1 (Ap.23)
N1 dt losses per unit mass are given by

123
Electr Eng (2008) 90:229–241 241

 2
π dB pmax fr σ 12. Hatziantoniu C, Galanos GD, Milias-Argitis J (1988) An incre-
Pcl = (Ap.27) mental transformer model for the study of harmonic overvoltages
6dens in weak ac/dc systems. IEEE T Power Deliv 3:1111–1121
while the excess losses are given by, [31] 13. Rajakovic N, Semlyen A (1989) Investigation of the inrush phe-
nomenon a quasi-stantionary approach in the harmonic domain.
8(σ Gdτ H o)1/2  3/2 IEEE T Power Deliv 4:2114–2120
Pexc = Bpmax fr (Ap.28)
dens 14. Chen XS, Neudorfer P (1992) Digital model for transient studies
of a three-phase five legged transformer. IEE Proc-C 139:351–358
where Bpmax is the maximum flux density in the lamination, 15. Dolinar D, Pihler J, Grcar B (1993) Dynamic model of a three-
fr the excitation frequency and dens the density of the mate- phase power transformer. IEEE T Power Deliv 8:1811–1819
rial of the lamination. 16. Yacamini R, Bronzeado H (1994) Transformer inrush calculations
The manufacturer in [34] gives the total losses Pc and the using a coupled electromagnetic model. IEE Proc-A 141:429–498
17. Hatziargyriou ND, Prousalidis JM, Papadias BC (1993) General-
dc hysteresis losses Ph at a specific frequency and maximum ized transformer model based on the analysis of its magnetic core
flux density. Hence, the eddy currents losses, the classical circuit. IEE Proc-C 140:269–278
losses and the excess losses can be calculated, accordingly, 18. EMTP-ATP rule book, Chapter XIX-C
from Eqs. (Ap.25), (Ap.27) and (Ap.26). Consequently, in 19. EMTP-ATP rule book, Chapter IV-E
20. Henriksen T (2002) How to avoid unstable time domain responses
Eq. (Ap.28) the term (σ Gdτ H o)1/2 is calculated as caused by transformer models. IEEE T Power Deliv 17:516–522
an unknown quantity. The derived values are 72.36 × 10−2 21. Chen X (2000) Negative inductance and numerical instability of
[A/(Volt)1/2 ] for Bmax = 1.7 T and 60.44 × 10−2 [A/(Volt)1/2 ] the saturable transformer component in EMTP. IEEE T Power
for Bmax = 1.5 T. The different values show that the internal Deliv 15:1199–1204
22. Chen X (1996) A three phase multi legged transformer model in
potential Ho is dependent on the maximum flux density for ATP using the directly formed inverse inductance matrix. IEEE T
this material. Power Deliv 11:1554–1562
23. Chadrasena W, McLaren PG, Annakkage UD, Jayasinghe RP
(2004) An improved low-frequency transformer model for use in
GIC studies. IEEE T Power Deliv 19:643–651
References 24. Thomas DWP, Paul J, Ozgonenel O, Christopoulos C (2006) Time-
domain simulation of non-linear transformers displaying hystere-
1. Silvester P, Chari MVK (1970) Finite element solution of saturable sis. IEEE T Magn 42:1820–1827
magnetic field problems. IEEE T Power Ap Syst PAS- 89:1642– 25. Pedra J, Sainz L, Corcoles F, Lopez R, Salichs M (2004) PSPICE
1651 computer model of a nonlinear three phase three legged trans-
2. Chari MVK, D’angelo J, Palmo MA, Sharma DK (1986) Appli- former. IEEE T Power Deliv 19:200–207
cation of three-dimensional electromagnetic analysis methods to 26. Jiles DC, Atherton DL (1984) Theory of ferromagnetic hysteresis.
elecrtical machinery and devices. IEEE T Energy Conver EC- J Appl Phys 55:2115–2120
1:145–157 27. Jiles DC, Thoekle JB (1989) Theory of ferromagnetic hysteresis:
3. Demerdash NA, Nehl TW, Fuand FA, Mohammed OA (1981) determination of model parameters from experimental hysteresis
Three dimensional finite element vector potential formulation of loops. IEEE T Magn 25:3928–3930
magnetic fields in electrical apparatus. IEEE T Power Ap Syst PAS- 28. Jiles DC, Thoekle JB, Devine MK (1992) Numerical determina-
100:4104–4111 tion of hysteresis parameters for the modeling of magnetic prop-
4. Demerdash NA, Nehl TW, Fuand FA, Mohammed OA (1981) erties using the theory of ferromagnetic hysteresis. IEEE T Magn
Experimental verification and application of the three dimensional 28:27–35
finite element magnetic vector potential method in electrical appa- 29. Tellinen J (1998) A simple scalar model for magnetic hysteresis.
ratus. IEEE T Power Ap Syst PAS- 100:4112–4122 IEEE T Magn 34:2200–2206
5. Arturi CM (1991) Transient simulation of a three phase five limb 30. Bertotti G (1988) General properties of power losses in soft ferro-
step-up transformer following an out-of-phase synchronization. magnetic materials. IEEE T Magn 24:621–630
IEEE T Power Deliv 6:196–207 31. Bertotti G (1998) Hysteresis in magnetism, 1st edn. Academic,
6. De Leon F, Semlyen A (1994) Complete transformer model for San Diego
electromagnetic transients. IEEE T Power Deliv 9:231–239 32. Liorzou F, Phelps B, Atherton DL (2000) Macroscopic models of
7. Narag A, Brierley RH (1994) Topology based magnetic model for magnetization. IEEE T Magn 36:418–428
steady state and transient studies for three-phase core type trans- 33. Chen SD, Lin RL, Cheng CK (2005) Magnetizing inrush model of
formers. IEEE T Power Syst 9:1337–1349 transformers based on structure parameters. IEEE T Power Deliv
8. Chen X, Venkata SS (1997) A three-phase winding core type trans- 20:1947–1954
former model for low frequency transient studies. IEEE T Power 34. Akcay H, Gokhan Ece D (2003) Modeling of hysteresis and power
Deliv 12:775–782 losses in transformer laminations. IEEE T Power Deliv 18:487–492
9. Mork BA (1999) Five legged wound core transformer model: deri- 35. Theocharis AD, Menti A, Milias-Argitis J, Zacharias Th (2005)
vation, parameters, implementation and evaluation. IEEE T Power Modeling and simulation of a single-phase residential photovoltaic
Deliv 14:1519–1525 system. In: Proceedings of the IEEE power tech 2005 conference,
10. Nakra HL, Barton TH (1973) Three phase transformer transients. St. Petesburg, paper nr. 206
IEEE T Power App Syst 93:1810–1818 36. Avila-Rosales J, Alvarado F (1982) Nonlinear frequency depen-
11. Ewart DN (1986) Digital computer simulation model of a steel core dent transformer model for electromagnetic transient studies in
transformer. IEEE T Power Deliv PWRD- 1:174–183 power systems. IEEE T Ap Syst PAS- 101:4281–4288

123

Vous aimerez peut-être aussi