Advances in Engineering Mechanics and Materials

Magnetic Hysteresis under an Applied

Continuous External Magnetic Field
Hamza HOUASSINE1
1

Djelloul MOUSSAOUI2

Samir MOULAHOUM1

Electrical and Automatic Research Laboratory (L.R.E.A) University of Medea Algeria

2

Laboratory Of Electromagnetic Systems, Emp, Bordj El Bahri, Algiers, Algeria

Abstract: Electromagnetic systems such as AC machines have to support supply voltages containing a DC component
which induces both an increase of the total magnetic losses and the premature saturation of the magnetic core. In the
present paper, we present an approach for predicting the hysteresis loop of a magnetic material such as non oriented
FeSi 3% which is subjected to a DC bias. The measurements are carried out with a bench test built around an Epstein
frame. The material is excited with a damped sinusoidal flux density superimposed to a known continuous field. We
obtain superimposed asymmetrical hysteresis loops. The cycles are modeled via the Preisach Model (PM) [1], which
provides both a mathematical model for the B(H) curve and an analytical approach which identifies and predicts the
parameter behavior needed by the PM.
Key words: Hysteresis, Preisach Model, continuous field, magnetic material.

1. Introduction
The magnetic properties of soft material such as
Fesi 3% are very sensitive to both applied
mechanical and thermal stresses as well as the
external electromagnetic field. Magnetic materials
such as Fesi 3% are commonly used in the
manufacturing of electromagnetic devices which are
supplied by a sinusoidal electrical network. The flux
density is symmetrical, a very important property for
the good functioning of this type of device.
Many devices operating in the low- and mediumfrequency range, ranging from rotating machinery
(dc machines, permanent magnet motor) to magnetic
cores in electronic drives, are employed under
special supply conditions, which may include dc
bias and generally non-sinusoidal induction with
local minima. The pre- diction of the core losses
under such conditions, which are far removed from
the standard conditions required for conventional
magnetic testing of materials [1], may present
special difficulties, inherent to the general features
of hysteresis (e.g. Non-local memory effects) and
their evolution with the magnetizing frequency [2].
This work attempts to investigate the function b(h)
take account the continue component of the
excitation. The pm model is chosen because, on one
hand, it merely accounts for all the parameters
involved in process of magnetization of magnetic
materials such as saturation, minor cycles, etc, and
on the other hand, the simplicity of its
implementation.
In order to validate the simulations, we

ISBN: 978-1-61804-241-5

carried out tests on two types of sheets: Fesi 3%

with oriented-grains 0.5mm thickness on the
Epstein bench.
2. Preisach Model
The classical Preisach model together with
quite a few generalizations has been efficiently
applied to describe hysteresis phenomena under
various conditions [1,2]. It consists in a set of
hysteresis operators (ideal switches with
rectangular characteristics) and features some very
favourable properties: fast and memory sparing
numerical implementation based on the everett
integrals, well de
ned and reliable experimental
identication procedure based on measurement of
rst order reversal curves. It can adequately
describe the static magnetic behavior of various
ferromagnetic materials.
Combined with eddy
current computation, the dynamic operation of
homogeneous materials or thick parts having
randomly oriented, tiny domains can be simulated
with acceptable accuracy [3]. This method fails
however to encompass the effects of domain wall
displacement [5], important in the case of electrical
steel sheets and it is also costlyin terms of
computing time and memory requirement.
Preisach represented the magnetic state of
magnetic material, at any time, has two possible
states of magnetization (m=1 and m= -1), defined
by a rectangular elementary cycle on the inputoutput diagram. The latter is characterized by the
inversion fields of and (with ) for which

221

Advances in Engineering Mechanics and Materials

Several analytical expressions can be used. One

of these approximations is the Lorentz function
given by [ 1 ], [ 2 ]:

there is an irreversible transition from the high state

(m=1) to the low state m= -1 and vice versa; that is,
and correspond to up and down switching values
of the input, respectively. The calculation of total
magnetization requires knowledge of the statistical
distribution of the elementary cycles. This function
is called Preisach function [1], [2][4]. Assuming the
input and output variables as function of time, the
Preisach function of magnetization resulting from
the application of an h(t) field is given by:

( , ) =

With
K: constant of standardization adjusted to have
M(Hs(t)) = Ms ,
Hc: the coercitive field.
For a better approximation of the experimental
loop, the Lorentzian function is modified by
adding parameters and takes the form:

With: (the value is (+1) if h= and (-1) if h=)

(,) : Density functionalso referred as Preisach
measure-. It depends on the nature of the material
[1].
,[H(t)] : Operator associated to the magnetic
entities referred to as elementary Preisach hysteron
operator.

(4)

1.1. Geometrical Interpretation of the Model

Experimental measurements show that the two

parameters a and b depend on the the nature of the
material, i.e. remanent induction, coercitive field
and permeability of the material. On the other
hand, the area of B(H) increases with the
frequency [3]. An adjustment of the parameters a
and b is necessary to have a correct modelling of
the B(H) behaviour. The parameters a and b are

There is a known one-to-one correspondence

between the operator ,[H(t)] and the points (,)
located in the half plan . Geometrically, S can
be subdivided into two parts, which are separated by
the border l(t), which is itself time dependent. The
surface S+(t) represents all the entities whose state of
magnetization is (+1), while S-(t) represents those
with state of magnetization (-1). Model (1) can then
be written in the following form:

( , )dd

S+

( , )dd

defined as follow: a*+ and b 1 , Hs

Hc

1.4. Mathematical Modified Formulation

With the PM model and the modified
Lorentz function, we obtain the following
expression.

(2)

1.2. The Distribution Function

The complete determination of the Preisach
model requires the knowledge of the density
function (,) , which is the basis for the calculation
of the total magnetization. For this purpose and at a
given time, intuitively two approaches contrast.
The first method relies on extensive
experimental hysteretic loops while the second
method consists of approximating real loops by
means of some analytical function. In our study, we
consider the second approach.

(5)
Mi-1 stands for the previous magnetization
moment. This formulation makes the calculation
easier. Figure 1 shows the influence of the
continuous component of the magnetic field on the
hysteresis loop of the ferromagnetic material.

I.3. Analytical Approach

ISBN: 978-1-61804-241-5

1 + H + 0.5

c

(3)

(1)

M (t ) =

K
2

1 + H 0.5

c

222

Advances in Engineering Mechanics and Materials

1.5

1
0.8
0.6

0.4
0.2

B[T]

B[T]

0.5

0
-0.2

-0.4
-0.6
-0.8

-0.5

-1
-300

-1
-100

-50

-100

100
H[A/m]

200

300

400

500

Fig.3. Hysteresis loop under DC bias combined whit AC

magnetization field excitation

200

150

100

50
H[A/m ]

-200

The figure 4 shows the anysteretic curve

Fig.1: Hysteresis loop under the continuous component

of magnetic field

obtained by varying the amplitude of the

dc bias magnetization.

This figure shows that the hysteresis loop is

asymmetrical. Hence, the magnetic properties of the
material are modified under the solicitation field
with constant component.

1.5

2. Simulation of the Hysteresis Loops under

Solicitation with Constant Component
B[T]

0.5

The solicitation considered is a combination

of dc bias and ac magnetization with linearly
diminishing amplitude. In order to modelize the
hysteresis, we subdivide the considered signal
into other periodical signal, and we use the
Preisach Model for the simulation.
The combined signal from dc bias and ac
magnetization as well as the linearly
diminishing amplitude and the hysteresis loops
obtained using PM are shown in Fig.2 and Fig.3
respectively.

-0.5

-1

-1.5
-500

-400

-300

-200

-100

0
H[A/m]

100

200

300

400

Fig.4. the Anysteretic curve obtained by varying the amplitude

of the dc bias magnetization

3. Experimental Bench

500

The Figure 5 illustrates the experimental

measurement setup composed from:
An Epstein frame standardized
A power amplifier KEPCO
A numerical generator programmable
WAVETEK 39.
A digital scope (MXOX2000) for the
visualisation and for data acquisition.
A PC for data acquisition and processing.
A current sensor.

400
300

H[A/m]

200
100
0
-100
-200
-300

0.02

0.04

0.06
t[s]

0.08

0.1

0.12

Fig.2. DC bias combined whit AC magnetization

field excitation

ISBN: 978-1-61804-241-5

500

223

Advances in Engineering Mechanics and Materials

B[T]

0.5

0.05

0.04
0.03
0.02

B[T]

0.01

-0.5

0
-0.01
-0.02

Zoom

-0.03
-0.04

-1

Fig.5. Experimental bench

-100

-50

-3

50
H[A/m]

-2

-1

0
1
H[A/m]

100

150

200

Fig.6.Exprimental and modeled loops under sinusoidal field

superposed to a known DC voltage component

4. Experimental validation

1.4

The Epstein test method is used to perform

measurements for an FeSi 3% sheet. The primary
winding is supplied with a sinusoidal voltage
superimposed to a known DC voltage component.
The amplitude of the maximum alternating voltage
is decreasing linearly (Figure 4). We keep constant
the parameter b in the Lorentz distribution function
while the parameter a is adjusted to draw hysteresis
loops as is described in Fig.6. We obtain a linear
evolution for the parameter a as a function of the
magnetisation and the loops are converging to the
magnetic state determined by the dc field, as shown
in Figure 7.
The same process is reproduced by
considering an exponential decrease of the
maximum amplitude of the alternating voltage.
Figures 8 and 9 show the comparison between
predicted and experimental hysteresis loops and the
dependence of parameter a wiht maximal induction
Bmax .

ISBN: 978-1-61804-241-5

-0.05
-4

1.2
1

B[T]

0.8
0.6
0.4
0.2
0
0.25

0.3

0.35

0.4

0.45

0.5

0.55

Fig.7. Dependence of a parameter of the Lorentz

function with the magnetic induction

224

0.6

Advances in Engineering Mechanics and Materials

1.2

Conclusion

Measurements have been carried out in

0.5mm thick non-oriented FeSi 3% laminations.
The material is excited with a deadened
sinusoidal field superimposed on a continuous
component. The obtained asymmetric loops are
modeled through the proposed approach based
on the Preisach Model (PM) combined with a
mathematical model which predicts the
parameters that the PM needs. The experiment
validation results are in agreement with several
experiments. As a perspective for this work, the
study of the modeling of the B(H) curve through
the new approach could be extended to predict
magnetic losses.

0.8
0.6
0.4

0.07

Zoom

0.06

0.2

0.05

0.04

0.03

0.02

-0.2
-0.4
-200

200

10

12

14

16

20

1000

800

600

400
H[A/m]

18

Fig.8. Experimental and modeled loops an exponential

decrease of the maximum amplitude of the alternating voltage

References
[1] G. Bertotti Dynamic Generalization of the Scalar
Preisach Model of Hysteresis IEEE Trans.
Magn.,
vol 28, pp. 2599-2601, September 1992
[2] D. Moussaoui, A. Bendjerad, M.Oussalah and H.
Houassine A Neural Network approach for
Determination of Preisach Model parameters Under a
Sinusoidal Induction at Various Frequencies, Physica
B, 372, pp. 106-110, Oct. 2006.
[3] E. Barbisio, O. Bottausciob, M. Chiampi, F.
Fiorillo, C. Ragusa Prediction of magnetic power
losses in soft laminations under DC-biased supply,
Journal of Magnetism and Magnetic Materials 290291
(2005) 14761479.
[4] J.Fzi, Experimental verification of a dynamic
hysteresis model, Physica B: Condensed Matter,
Volume 343, Issues 14, 1 January 2004, pp. 8084
[5] O. Bottauscio, G. Pellegrino, P. Guglielmi, M.
Chiampi, and A. Vagati Rotor Loss Estimation in
Permanent Magnet Machines With Concentrated
Windings IEEE Transactions On Magnetics, Vol. 41,
pp. 3913-3915 N. 10, Oct. 2005.

0.8
0.7
0.6
0.5
B(T)

B[T]

0.08

0.4
0.3
0.2
0.1
0
0.1

0.15

0.2

0.25

0.3
a

0.35

0.4

0.45

0.5

Fig.9. Dependence of a parameter of the Lorentz function

with the magnetic induction

By repeating the process for different DC

voltage values corresponding to the different DC
magnetization, the table 1 illustrates the comparison
between the experimental and predicted values of
the DC magnetization state. We notice that there is
no significant difference between the two.
DC component of the
Experimental
magnetization
0.2587

DC component of the
predicted magnetization
0.22

0.3235

0.35

0.4038

0.45

0.5265

0.55

0.6612

0.7

Table 1 : Comparison between DC magnetization values

(experimental and predicted)
ISBN: 978-1-61804-241-5

225