Author’s Accepted Manuscript

Hysteresis modelling of GO laminations for

arbitrary in-plane directions taking into account the
dynamics of orthogonal domain walls

A.P.S. Baghel, B. Sai Ram, K. Chwastek, L.

Daniel, S.V. Kulkarni
www.elsevier.com/locate/jmmm

PII: S0304-8853(16)30109-3
DOI: http://dx.doi.org/10.1016/j.jmmm.2016.02.007
Reference: MAGMA61133
To appear in: Journal of Magnetism and Magnetic Materials
Received date: 12 November 2015
Revised date: 31 January 2016
Accepted date: 4 February 2016
Cite this article as: A.P.S. Baghel, B. Sai Ram, K. Chwastek, L. Daniel and S.V.
Kulkarni, Hysteresis modelling of GO laminations for arbitrary in-plane
directions taking into account the dynamics of orthogonal domain walls, Journal
of Magnetism and Magnetic Materials,
http://dx.doi.org/10.1016/j.jmmm.2016.02.007
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Hysteresis Modelling of GO laminations for arbitrary in-plane
directions taking into account the dynamics of orthogonal domain
walls
A. P. S. Baghel1, B. Sai Ram1, K. Chwastek2, L. Daniel3, and S. V. Kulkarni1

1
Department of Electrical Engineering, Indian Institute of Technology Bombay, Mumbai, 400076, India
2
Department of Electrical Engineering Czestochowa University of Technology, Poland
3
Group of Electrical Engineering - Paris (GeePs), CNRS(UMR8507)/CentraleSupelec/UPMC/Univ Paris-Sud,
11 rue Joliot-Curie, 91192 Gif-sur-Yvette ,France

Abstract:
The anisotropy of magnetic properties in grain-oriented steels is related to their microstructure. It

results from the anisotropy of the single crystal properties combined to crystallographic texture. The

magnetization process along arbitrary directions can be explained using phase equilibrium for domain

patterns, which can be described using Neel’s phase theory. According to the theory the fractions of

180o and 90o domain walls depend on the direction of magnetization. This paper presents an approach

to model hysteresis loops of grain-oriented steels along arbitrary in-plane directions. The considered

description is based on a modification of the Jiles-Atherton model. It includes a modified expression

for the anhysteretic magnetization which takes into account contributions of two types of domain

walls. The computed hysteresis curves for different directions are in good agreement with

experimental results.

Keywords— Hysteresis loops, grain-oriented laminations, anisotropy, domain walls, Jiles-Atherton model.

1.1. INTRODUCTION

The magnetic properties of grain-oriented steels (hysteresis loops, losses, permeability, etc) exhibit

substantial anisotropy due to the strong anisotropy of Iron-Silicon crystals combined to a very

pronounced crystallographic texture [1, 2]. From a practical point of view, the development of

physics-based descriptions of magnetization processes, taking into account anisotropy effects, is

crucial for proper material characterization. Grain-oriented (GO) Iron-Silicon steels are widely used
1
in electromagnetic devices such as power transformers and large rotating machines [3-6]. Standard

characterization techniques are most often applied along the two principal directions, namely the

rolling direction (RD) and the transverse one (TD) [7-9]. However, in many applications (e.g. core

joints and corners in transformers) the magnetization process happens along directions different from

the aforementioned two ones [10,11]. It is thus important to develop appropriate tools for material

behavior description in these conditions.

Among numerous hysteresis models three of them have gained much attention of scientists and

engineers [12]. These are Preisach-Mayergoyz, Stoner-Wohlfarth and Jiles-Atherton models. The

Preisach model is based on the concept that hysteresis loop can be described from the superposition

of contributions from elementary entities called hysterons [13, 14]. This description has been studied

in detail since its inception in 1935 and substantially improved by I. D. Mayergoyz [14]. A thorough

discussion of individual modifications of the Preisach model is beyond the scope of the present paper

and the readers are referred to the monograph [14]. For the subject of the present paper, it suffices to

focus on the vector extension of the Preisach model [15]. It considers individual distributions of

hysterons for different magnetization directions. Hence, the vector model is constructed as a

superposition of several scalar Preisach models. The idea has been verified and further extended for

instance to consider possible couplings between distributions of individual axes [16, 17], to extend

the problem to 3D models [18], to deal with efficient numerical implementation [19,20], or to

describe the phenomena of accommodation and aftereffect [21].

The Stoner-Wohlfarth description is derived from micromagnetics and naturally introduces the

anisotropy of individual grain as a key parameter [22]. It is inherently vectorial in nature and it has

been extensively used by scientists working on thin films. In practice it is very often combined with

the Preisach model [15, 23]. Other approaches - so called multiscale approaches -have also been

proposed to deduce the macroscopic behavior of magnetic materials from energetic considerations at

the magnetic domain scale [24-26].

2
The third popular hysteresis model is the Jiles-Atherton approach [27]. It introduces a limited number

of parameters to define the anhysteretic magnetization and the hysteresis contribution. It is given in

the form of a set of algebraic equations and an ordinary differential equation, which can be

implemented in numerical codes. It provides a macroscopic material response to a given loading. The

original model has been further extended for instance to consider vectorial loadings [28-30], or to

introduce advanced description of the anhysteretic curve [31, 32]. A recent modification in the JA

model is proposed for describing hysteresis loops of grain-oriented steels along the two principal

directions [33]. In previous works the usefulness of the model was assessed for the characterization of

hysteresis loops, including minor loops, for two principal directions either in quasi-static or dynamic

conditions [34, 35].

The magnetization process can be described in terms of complex variations of the domain pattern

with directions of the applied field [3-4]. A theoretical study of the angular dependence of magnetic

properties in GO materials is reported in [4] using the domain phase theory. An experimental study,

reported in [3], supports the theory with observation of different proportions of 180 o and 90o domain

walls (DWs) in different directions. Magnetic properties along RD can be modelled using the

contribution of 180o DWs only, whereas, in other directions, these can be explained in terms of 180o

and 90o DWs [4]. Anisotropy in the magnetic behavior of GO materials is evident in Fig. 1 showing

measured quasi-static hysteresis loops (details of measurement procedure is given in Section-IV) at

angles 0o, 45o and 90o from the rolling direction.

Hysteresis loops along different directions can be described from the complicated interplay between

two magnetization mechanisms viz., the irreversible motion of domain walls and the irreversible

rotation of domain magnetization vectors [36]. Lower initial values of permeability and higher

coercivity values along TD can be attributed to an initial domain configuration mainly oriented

towards RD so that the magnetization process involves a significant amount of magnetization rotation

[7, 37]. When a field is applied along TD, the magnetization process is completed first by nucleation

3
and growth of domains perpendicular to RD, and then rotation to align them with TD [7]. Therefore, a

hysteresis model must have the ability to consider these mechanisms (wall translation and

magnetization rotation) and to relate the internal parameters to corresponding microstructure. Such a

model will allow considering the case of rotating flux in electromagnetic applications. When

integrated into system design CAD tools, it will allow an accurate computation of fields in

electromagnetic devices.

Fig. 1 Measured quasi-static hysteresis loops for a GO material at different angles (0O, 45O, and 90O)

This paper presents the modeling of hysteresis loops in arbitrary in-plane directions for a sheet of

grain-oriented steel on the account of contributions of 180o and 90o DWs motions and the

magnetization rotation in the magnetization process. The model adapts the assumptions from the

Neel’s phase theory [38]. It computes the anhysteretic magnetization using a relationship that

includes energy contributions from both types of walls using two orthogonal components of the

applied field along RD and TD. Hysteretic effects are included in the model using the pinning

concept of the JA approach. An additional parameter, which represents the volume fractions of the

two types of DWs in the material, is used in the model. The proposed model is validated using

measured hysteresis loops along different directions. Computed and experimental results are in close

agreement.

4
1.2. DOMAINS AND MAGNETIZATION PROCESSES

As a first approximation GO Fe-Si materials can be considered as single crystals because the axes of

individual crystallites are aligned with the global cubic axes [39]. In single crystals, prior knowledge

of the space orientation of easy magnetization axes can be used to define domain structures/patterns

[3-4, 40]. The domain structure of a material sample can be described in terms of the coexistence of

phases provided that the external field is large enough to overcome pinning fields. Typical hysteresis

loops of GO materials can be reproduced using various forms of this phase equilibrium [4].

Magnetocrystalline anisotropy plays a major role in the phase equilibrium for these materials [40].

The anisotropy energy for a cube with a positive anisotropic constant (K1), produces six equivalent

energy minima, corresponding to the magnetization vector m pointing along the positive or negative

direction of the three orthogonal easy axes. Thus, there are six magnetic phases and a given domain

structure represents a particular mixture of these phases [40]. Domains in demagnetized GO

laminations are mostly oriented along RD and can be described with two phases ([001] and [ 001 ]).

The domains change in a complex manner when the field is applied along directions other than RD.

The domain pattern of a sheet along its plane and cross section in a demagnetized state is shown in

Fig. 2, with magnetization ms directed along [001] and [ 001 ] (i.e. RD). Hence, in the demagnetized

state, the sample has only two phases with magnetization aligned with RD (positive and negative).

When an applied field is applied along RD, the domains with favourable orientation will expand at

the expense of other domains. In the saturation state, the sample can be represented by a single

domain structure. Therefore the RD loops of GO laminations can be defined in terms of these two

phase evolutions ([001] and [ 001 ]). On the other hand, when a magnetic field is applied along TD,

the domains will change in a more complex fashion. As the applied field increases, the basic 180o

domains will get transformed through 90o domain wall processes. The domains in the sheet cross

section with the magnetization directed along [100] and [ 010 ] are shown in Fig. 2. This domain

structure reduces the magnetization component normal to the sheet surface to zero [4, 7]. Hence, the

5
sample has a mode (known as mode-I) of four phases in this procedure which terminates at maximum

magnetization according to Kaya’s rule [4, 40]. If the applied field further increases, the two phases

coexist in equilibrium and the magnetization further increases by domain magnetization rotation.

Therefore, in the magnetization process along TD, the sample will have a mixture of both 180 o and

90o DWs and it exhibits a hysteresis loop with a complex shape. The magnetization process and the

corresponding domain structures have consequences at the macroscopic level in terms of magnetic

induction, coercivity, permeability and losses. The angular dependence of these variables with the

texture of materials is discussed in [41]. The value of maximum magnetization (ms) depends on the

domain structure (i.e., the balance between different phases) of the material sample [4]. The value of

the maximum (technical saturation) magnetization in different directions for these materials can be

calculated according to Kaya’s rule [4, 40]. This "maximum" magnetization is also referred to as

―apparent saturation magnetization‖ and its directional dependence can be explained by

demagnetising effects due to the complex domain structure. It is important here to note that the

―final‖ or ―true‖ saturation magnetization is the same for any direction and it is markedly higher than

the ―technical‖ or ―apparent‖ saturation magnetization‖ [40]. An attempt to explain the anisotropy in

the coercive field and losses using subdivision of hysteresis loss into low induction and high

induction loss components is made in [42].

6
 Fig. 2 Domain pattern of a single crystal (110) [001] when field is applied along (a) RD (b) TD [7].

1.3. HYSTERESIS MODEL

For an ideal Goss-textured material, the (110) plane is parallel to the surface of the sheet and the

[001] direction lies along RD [43]. The magnetic field in GO laminations is generally analyzed in two

dimensions because of their small thickness. A modified JA hysteresis model to describe hysteresis

loops along RD and TD of GO steels has been proposed in [33]. The model uses a single lamination

model as shown in Fig. 3. Two magnetic moments m1 and m2 in the plane of the lamination, need to

rotate from 0o to 180o due to symmetry in the anisotropy energy [44]. The energy equation, in

presence of anisotropy, is [44]:

E  0ms  (H   M )  Ean . (1)

According to Ref. [45], the anisotropy energy expression can be written as

 sin4   cos2   sin4 

Ean =K0  K1  cos2   sin2     K2 (2)
 4  4

where φ is the angle between the magnetization direction and RD. The anisotropy constants K 0 and

K2 can be neglected without a significant loss of accuracy [45].

7
 Fig. 3 A single sheet model and field components

The model, reported in [33], includes the magnetocrystalline anisotropic energy on the account of

modified anhysteretic magnetization. Since the two basic types of domain walls are responsible for

the magnetization process in these materials, the model can accurately predict hysteresis loops along

the two principal directions. In other directions, magnetization process can be described by

introducing the proportions of these two basic DWs [3].

For an applied field at an arbitrary angle ϕ from RD, the magnetic field can be decomposed into two

orthogonal components as:

H  H cos   and H  H sin   (3)

The first term of the energy equation (1) can be rewritten as the sum of two energy terms

corresponding to the two orthogonal field components as:

Es  [(0 ms  (H   M ))  (0 ms  (H   M )) (4)

The equation is consistent with the basic model given in [33] because when the field is along RD,

only the first term will contribute to the total energy and when the field is along TD only the second

term will be non-zero. Now, the total anisotropy energy can also be rewritten as the sum of two

energy terms corresponding to these two field components. The anisotropic energies of the two

magnetic moments m1 and m2 can be expressed as [33]:

8
   sin4 11   2 sin4 12 
Ean 1  Kan rx  cos2 11  sin2 11    1  rx   cos 12  sin 12  
2
(5)
  4   4 

  sin4 21   2 sin4 22 

Ean  2  Kan rx  cos2 21  sin2 21    1  rx   cos 22  sin 22  
2
(6)
  4   4 

where 11  1  ; 12  2  ; 21  1  and 22  2  , whereas θ is the angle between the

applied field and the magnetic moments. ϕ 1 and ϕ 2 are 00 and 900, respectively. rx is the proportion

parameter which is related to the volume fractions of the two domain walls.

When a field is applied along RD, the parameter is 1 and when the applied field is along TD, it is

zero. For the intermediate directions it will take different values between 0 and 1. Now, the total

energy equation becomes:

E  Es  Ean 1  Ean  2 (7)

The anhysteretic magnetization can be expressed as a function of the magnetization direction [44]:


 exp  E / k T  sin cos d

B

M an  ms 0

. (8)
 exp  E / k T  sin d
0
B

Equation (8) is consistent with the Langevin function in the case of zero average anisotropy energy

(K1 = 0), as per the modification reported in [46]. This equation can be integrated numerically in

order to compute the anhysteretic magnetization. The hysteresis curve may be modelled similarly like

in the JA approach. An appropriate offset to the anhysteretic magnetization may be introduced in

order to take into account the irreversible magnetization processes [27]. The relationship for the

differential susceptibility in the JA model is given as,

dM  M an ( H )  Mirr (H )  dM an dM 
 c  
dH  (k  0 )    M an (H )  Mirr (H )   dH dH 
(9)

where M is the total magnetization, H is the applied magnetic field, Man is the anhysteretic

magnetization, Mirr is the irreversible magnetization, μ0 is the magnetic permeability of vacuum, and

δ is the directional parameter taking the values: +1 for dH/dt >0 and −1 for dH/dt < 0.

9
The extended model contains seven parameters, including an additionally introduced parameter (rx).

Their values have to be determined using an estimation procedure, for example with a direct iterative

method [47], a heuristic [48] or a hybrid technique [49]. It is important to remark that the inverse

form of the model may be written analytically. This form may be easily implemented into finite

element analysis tools. Moreover it is useful in order to compare the modeling results with

measurements carried out under standard conditions, as described in the IEC 60404 Standard (Epstein

frame and Single Sheet Tester) in which time dependence of induction B is controlled. The inverse

from can be obtained from relationship (9) as [43]:

  c dM an   dMirr  
     1  c   
dM   0 dHe   dBe  

dB   dM an  
. (10)
dM an  
1   1    c     1   1  c  
  dHe   0 dHe  

1.4. MODELING OF HYSTERESIS LOOPS IN ARBITRARY DIRECTIONS

Measurements are carried out using a standard single sheet tester (SST) (Model: MGP 200 D).

Material samples are cut at angles of 0o, 10o, 20o, 30o,.., 90o with respect to RD of a highly grain-

oriented (HGO, grade:27-MOH) material. The thickness of the samples is 0.27 mm, the length is 200

mm, and the width is 29.5 mm. The measurement frequency is set to 1 Hz. It is assumed that the

effects of eddy currents on the hysteresis loops can be neglected at this frequency [50].

First, the extended model is applied to quasi-static hysteresis loops along the principal directions. The

model parameters for the measured RD and TD curves are obtained using the hybrid technique

reported in [49]. The optimization procedure is repeated 15 times in order to ensure repetitiveness of

the results. The best-fit values for model parameters are given in Table I.

TABLE I
OPTIMIZED PARAMETERS FOR RD AND TD
Angle ms (A/m) a(A/m) k(A/m) Α C K1(J/m3) rx

θ =0o 1.425×106 30 20 5.1×10-5 0.33 1.4×102 1

θ =90o 1.11×106 33 29 5.1×10-5 0.33 1.7×102 0

10
For the rolling direction loops, only 180o DWs (two phases) will be present in the sample. For the

transverse direction the domains are arranged in a rather complex manner, as 90o domain walls are

also present. The measured and computed RD and TD curves are shown in Fig. 4. Maximum values

of mean squared error do not exceed 6.7 % (the highest error is obtained at the knee region on the

transverse direction loop).

Fig. 4 Measured and computed hysteresis loops (along RD and TD)

Next the model is applied to hysteresis loops along other directions. As discussed in the previous

section, a sequence of different phases appears in the magnetization process, with a mixture of both

180o and 90o DWs. When the field is applied at any angle with respect to RD, the domain patterns (in

the presence of 90o DWs) lead to a sequence of different modes [4]. Some of the model parameters

are chosen to vary with the direction of the applied field. Since the maximum value of the

magnetization due to domain wall motion varies with direction according to Kaya’s rule [4], the

model parameter representing saturation magnetization, ms, depends on the direction of the applied

field. The model parameters a (form factor) and k (pinning factor) depend on the domain structure of

the materials [51]. Therefore, these parameters are also kept free to vary with respect to the direction

of the applied field. The model parameter K1 is used to adjust the insufficient description of the

material anisotropy and particularly of the role of the initial domain configuration in the

magnetization process. Hence it is made to depend on the magnetization angle. The proportion
11
parameter (rx) will also vary to take into account different proportions of the two types of DWs in

different directions. The two remaining parameters of the model, α (mean field parameter) and c

(domain wall bowing parameter), can be assumed to be independent of the direction of the applied

field and remain the same for different magnetic phases [52]. Hence, five model parameters (ms, a, k,

K1, and rx) have to be estimated, whereas α and c are kept fixed. The optimized parameters for

hysteresis loops at different angles are given in table II. The magnetization process is basically due to

movements of tilted 180o DWs for a magnetic field orientation up to θ ≈ 30o as it is confirmed in [1]

using the normalized hysteresis loss. At higher θ values, a transition of 90o DWs occurs and a sharp

increase in the hysteresis loss due to irreversible rotation of domain magnetization can be observed.

It is also confirmed by sharp changes in the values of the parameters (K1 and rx) at θ ≈ 30o in the

proposed model.

TABLE II
OPTIMIZED MODEL PARAMETERS AT DIFFERENT ANGLES
Angle ms (A/m) a(A/m) k(A/m) α C K1(J/m3) rx

θ =10o 1.30×106 32 25 5.1×10-5 0.33 1.5×102 0.92

θ =20o 1.20×106 42 35 5.1×10-5 0.33 1.7×102 0.85

θ =30o 1.02×106 43 38 5.1×10-5 0.33 1.4×103 0.48

θ =40o 1.01×106 54 47 5.1×10-5 0.33 1.8×103 0.46

θ =50o 9.5×105 56 49 5.1×10-5 0.33 1.9×103 0.45

θ =60o 9.9×105 57 50 5.1×10-5 0.33 2.2×103 0.43

θ =70o 9.92×105 58 53 5.1×10-5 0.33 1.93×103 0.42

θ =80o 1.05×106 60 56 5.1×10-5 0.33 1.91×103 0.40

The measured and computed curves at different intermediate angles are shown in Figs. 5-6 for angles

(10o, 20o, 30o…..90o) from the rolling direction.

12
 Fig. 5 Measured (dashed line) and computed (solid line) hysteresis loops at different angles (10o, 20o, 30o, and

40o)

Fig. 6 Measured (dashed line) and computed (solid line) hysteresis loops at different angles (50o, 60o, 70o, and 80o)

Computed and experimental curves are in reasonably close agreement as evident from the figures.

The highest mean squared error value is not more than 8.2% for loops in arbitrary directions (the

highest error is for θ = 80o). The angular variations of the two model parameters k and a, which

strongly depend on the domain structure of the material, are shown in Fig. 7.

13
 Fig. 7 Angular variations of the model parameters (k and a)

The trends obtained for these parameters can also be compared with Orientation Distribution

Functions (ODF)-based description recently examined in [53]. This is particularly so for the k

parameter, as it is interpreted as the product of the pinning site density and their average energy [27].

Assuming as a first approximation, that this parameter is approximately equal to coercivity, which

controls power losses, it follows that the angular dependence of parameter (k) should be somewhat

similar to that described by the Lankford parameter in [53]. The trend is indeed followed quite neatly

up to 50o-60o. However there are some discrepancies above 60o. The discrepancy between the values

obtained from the estimation procedure and evaluated from the ODF theory did not exceed 15% up to

60o. This may indicate that for higher angles the roles of 180o and 90o DWs are to some extent

interchanged which could indicate another dominant dissipation mechanism, different from the

domain wall motion.

1.5. Conclusions

Grain-oriented laminations show direction-dependent magnetic properties due to their strong

crystallographic structure. The macroscopic properties can be described in terms of 180o and 90o

domain wall movements and associated phase equilibrium. Bases on the Jiles-Atherton model, this

paper presents an approach to model hysteresis loops along arbitrary directions using the domain

phase theory to describe average magnetic domain structures. The modeling along arbitrary directions

14
is achieved by considering separately the contributions of 180o and 90o domain walls. A new

proportion parameter (rx) is defined in the model to take into account different volume fractions of the

two basic types of DWs. The model has the ability to approximate hysteresis loops for any

magnetization angle from RD to TD in GO laminations. The measured and computed curves at

different angles are in close agreement. The model can be useful for the analysis of rotational

hysteresis losses in many applications where the flux is not aligned with the principal directions, for

example, at T-joints of power transformers and teeth regions of large rotating machines.

The proposed approach could be further extended to consider the effect of mechanical stress. It could

shed some light on the physics of coupled magneto-mechanical effects after slight modifications, and

notably it may lead to an appropriate extension in the definition of the effective field. In the case of

magneto-mechanical loadings, the two types of domain walls are active whichever is the

magnetization direction as pointed out in [54-55]. A more complex description of the domain

microstructure and its evolution is then required as shown in [56]. The development of a coupled

anisotropic extension of the JA model is a potential direction for further study in this research area.

ACKNOWLEDGEMENTS

The authors would like to thank Crompton Greaves Ltd., Mumbai, India, for providing experimental

facilities. The authors would also be grateful to Dr. Satish Shekhawat, a PhD graduate from

MEMS Dept., IIT Bombay, for his valuable inputs in experimentations.

REFERENCES

[1] L. R Dupre, F. Fiorillo, J. Melkebeek, A. M. Rietto, and C. Appino, J. Magn. Magn. Mater. 215-

216, 112 (2000)

[2] V. Paltanea and G. Paltanea, Mater. Sci. Foru. 670, 66 (2011).

[3] S. Shin, R. Schaefer and B. C. DeCooman, J. Appl. Phys.109, 07A307-1 (2011).

[4] F. Fiorillo, L. R. Dupre, C. Appino, and A. M. Rietto, IEEE Trans. Magn. 38, 1467 (2002)

15
[5] H. Pfützner, IEEE Trans. Magn. 30, 2802 (1994).

[6] A. Basak and P. Caryotis, IEEE Trans. Magn. 20, 1560 (1984).

[7] O. Hubert, and L. Daniel, J. Magn. Magn. Mater 320, 1412 (2008).

[8] F. Fiorillo, Measurement and characterization of magnetic materials, first ed., Elsevier-Academic

Press, Amsterdam, 2004.

[9] R. Szewczyk, Application of Jiles-Atherton model for modelling magnetization characteristics of

textured electrical steel magnetized in easy or hard axis, in: R. Szewczyk et al. (Eds.), Progress in

Automation, Robotics and Measuring Techniques, book series Advances in Intelligent Systems and

Computing, vol. 350, Springer-Verlag, Berlin-Heidelberg, 2015, pp. 293-302.

[10] H. Yoon, and C. S. Koh, COMPEL 33, 14 (2014).

[11] A. Moses, J. Mater. Eng. and Perform. 1, 235 (1992).

[12] E. Cardelli, Advances in hysteresis modeling, chapter, in: K. H. J. Buschow (Ed.), Handbook of

Magnetic Materials, vol. 24, Science Direct Publication, 2015, pp. 323-409.

[13] F. Preisach, Z. Phys. 94, 277 (1935).

[14] I. D. Mayergoyz, Mathematical models of hysteresis and their applications, first ed., Elsevier-

Academic Press, Amsterdam, 2003.

[15] I. D. Mayergoyz, J. Appl. Phys. 63, 2995 (1988).

[16] A. A. Adly, and S. K. Abd-El-Hafiz, IEEE Trans. Magn. 42, 1518 (2006).

[17] A. A. Adly, and S. K. Abd-El-Hafiz, Physica B 407, 1350 (2012).

[18] I. D. Mayergoyz, and G. Friedman, IEEE Trans. Magn. 24, 2928 (1988) .

[19] G. Friedman, and D. Ahya, IEEE Trans. Magn. 30, 4386 (1994).

[20] E. Dlala, A. Belahcen, K. A. Fonteyn, and M. Belkasim, IEEE Trans. Magn. 46, 918 (2010).

[21] E. Della Torre, IEEE Press, New York 1999.

[22] E. C. Stoner, and E. P. Wohlfarth, Philos. Trans. R. Soc. London A240, 599 (1948).

[23] G. Kahler, E. Della Torre, and E. Cardelli, IEEE Trans. Magn. 46, 21 (2010).

16
[24] W.D. Armstrong, J. Appl. Phys. 81, 232 (1997).

[25] H. Hauser, J. Appl. Phys. 96 (5), 2753 (2004).

[26] L. Daniel, M. Rekik, O. Hubert, Archive of Applied Mechanics, 84 (9), 1307 (2014).

[27] D. C. Jiles, and D. L. Atherton, J. Magn. Magn. Mater. 61, 48 (1986).

[28] M. Jagieła, Arch. Electr. Eng. 56, 57 (2007).

[29] A. J. Bergqvist, IEEE Trans. Magn., vol. 32, no. 5, pp. 4213, (1996).

[30] J. V. Leite, N. Sadowski, P. Kuo-Peng, N. J. Batistela, J. P. A. Bastos, and A. A. de Espíndola,

IEEE Trans. Magn. , vol. 40, no. 4, pp. 1769–1775, Jul. 2004

[31] P. Dimitropoulos, G. I. Stamoulis, and E. Hristoforou, IEEE Sens. Journ. 6, 721 (2006).

[32] L. Bernard, L. Daniel, IEEE Trans. Magn., 51(9), 7002513 (2015).

[33] A. P. S. Baghel and S. V. Kulkarni, J. Appl. Phys, 113, 043908-1 (2013).

[34] A. P. S. Baghel, K. Chwastek, and S. V. Kulkarni, IET-Electr. Power Appl, 9, 344 (2015).

[35] A. P. S. Baghel, A. Gupta, K. Chwastek, and S. V. Kulkarni, Physica-B, 462, 86, (2015).

[36] C. Appino, M. Valsania, and V. Basso, Physica-B, 275, 103, (2000).

[37] M. F. de Campos, M. A. Campos, F. J. G. Landgraf, L. R. Padovese, J. Phys. 303, 012020

(2011).

[38] L. Néel , J. Phys. Radiat., 5, 241 (1944).

[39] M. Celasko and P. Mazzetti, IEEE Trans. Magn. 5, 372, (1969).

[40] G. Bertotti, Hysteresis in magnetism, Academic Press, San Diego, 1998.

[41] M. F. de Campos, Proc. XVIII IMEKO World Congress, Rio de Janeiro, Brazil, (2006).

[42] M. Emura, M. F. de Campos, F. L .G. Landgraf, and J. C. Teixeira, J. Magn. Magn. Mater. 226,

1524 (2001).

[43] A. P. S. Baghel, S. K. Shekhawat, S. V. Kulkarni, and I. Samajdar, Physica B 448, 349, (2014).

[44] A. Ramesh, D. C. Jiles, and J. M. Roderick, IEEE Trans. Magn. 32, 4234 (1996).

[45] G. H. Shirkoohi and M. A. M. Aricat, IEEE Trans. Magn. 30, 928 (1994).

17
[46] R. Szewczyk, J. Materials, 7, 5109 (2014).

[47] D. C. Jiles, J. B. Thoelke, and M. K. Devine, IEEE Trans. Magn. 28, 27 (1992).

[48] R.-A. Naghizadeh, B. Vahidi, S. H. Hosseinian, COMPEL 31, 1293 (2012).

[49] A. P. S. Baghel and S. V. Kulkarni, IET-Electr. Power Appl. 6, 689 (2012).

[50] T. R. Haller and J. J. Kramer, J. Appl. Phys. 41, 1034 (1970).

[51] K. Chwastek, J. Szczygłowski, and W. Wilczyński, COMPEL 28, 603 (2009).

[52] A. Raghunatan, Y. Melikhov, J. E. Snyder, and D. C. Jiles, J. Magn. Magn. Mater. 324, 20

(2012).

[53] K. Chwastek, A. P. S. Baghel, M. F. de Campos, S. V. Kulkarni, and J. Szczygłowski, IEEE

Trans. Magn. 51, 6000905 (2015).

[54] O. Perevertov and R. Schäfer, J. Appl. Phys-D:App. Phys. 45, 135001-1 (2012).

[55] O. Perevertov and R. Schäfer, J. Appl. Phys-D:App. Phys. 47, 185001-1 (2014).

[56] K. J. Rizzo, O. Hubert, and L. Daniel, IEEE Trans. Magn. 46, 270 (2010).

Highlights
- An extended Jiles-Atherton description is used to model hysteresis loops in GO steels

- The model stresses the role of material anisotropy and different contributions of the two types of
domain walls
- Hysteresis loops can be modelled along arbitrary in-plane directions

- Modeling results are in good agreement with experiments

18