View
216
Download
0
Embed Size (px)
Citation preview
8/11/2019 False Position.pdf
1/4
Modified JilesAtherton model and parameters identification using falseposition method
M. Hamimid a,b,, M. Feliachi a, S.M. Mimoune b
a IREENA-IUT, CRTT, 37 Boulevard de lUniversite, BP 406, 44602 Saint Nazaire Cedex, Franceb Laboratoire de modelisation des systemes energetiques LMSE, Universite de Biskra, 07000 Biskra, Algeria
a r t i c l e i n f o
Article history:
Received 10 December 2009Accepted 12 January 2010
Keywords:
Hysteresis
Classical and modified JilesAtherton
model
Parameters identification
False position method
a b s t r a c t
In this paper, a modified JilesAtherton model is proposed. This model uses a physical meaning by
introducing the magnetizationMinstead of the irreversible magnetization Mirrin the effective magneticfieldHe-magnetic fieldHrelationship. The false position method is coupled to the iterative algorithm to
identify the JilesAtherton parameters for both classical and modified JilesAtherton model. These
parameters are evaluated by the resolution of three nonlinear equations obtained from three
conditions. The validity of the modified model is done by comparing the obtained hysteresis loops to
the experimental ones.
& 2010 Elsevier B.V. All rights reserved.
Contents
1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1947
2. Modified JilesAtherton hysteresis model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1948
3. Determination of the modified JilesAtherton parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1948
4. Identification technique. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19495. Results and discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1949
6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1950
Work context. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1950
Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1950
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1950
1. Introduction
Different models have been proposed to describe the hyster-
esis phenomenon[13]. Some of them are physical in nature[1,2]
and others ignore the physical behavior of materials [3]. Amongthe physical models, the JilesAtherton is the most and widely
used[1].
To evaluate the effective magnetic field, Jiles and Atherton use
the irreversible magnetization instead of the total magnetization
to derive a simplified formula of the total differential suscept-
ibility [4]. Our purpose in this work is to present a modified
JilesAtherton (MJA) model to take into account the total
magnetization to evaluate the equivalent magnetic field. The
expression of the total differential susceptibility obtained from
this modification is completely different to that obtained by Jiles
and Atherton (JA). The integration of this model in a calculationcode requires the knowledge of the exact parameters. The
parameters of both models JA and MJA require an identification
technique. Several techniques are used to identify the JA model
parameters [59].
In this work, we have used a technique which combines an
iterative algorithm and a false position method FPM for both
models. The parameters that we want to identify in the MJA are
grouped into three nonlinear equations and solved by the FPM.
This technique needs some measurements data such as the
initial, the coercive and the remanence susceptibilities. It requires
also the introduction of the coercivity field intensity, the remanence
magnetization and the coordinates of the saturation point.
ARTICLE IN PRESS
Contents lists available atScienceDirect
journal homepage:www .elsevier.com/locate/physb
Physica B
0921-4526/$- see front matter& 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.physb.2010.01.078
Corresponding author at: IREENA-IUT, CRTT, 37 Boulevard de lUniversite, BP
406, 44602 Saint Nazaire Cedex, France.
E-mail addresses: [email protected] (M. Hamimid), mouloud.
[email protected] (M. Feliachi),[email protected] (S.M. Mimoune).
Physica B 405 (2010) 19471950
http://-/?-http://www.elsevier.com/locate/physbhttp://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.physb.2010.01.078mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.physb.2010.01.078http://www.elsevier.com/locate/physbhttp://-/?-8/11/2019 False Position.pdf
2/4
ARTICLE IN PRESS
Experimental results of the hysteresis loops are conducted to
validate the MJA model with comparison to the JA model.
2. Modified JilesAtherton hysteresis model
In the JilesAtherton model, the magnetization M is decom-
posed into reversible component Mrev, representing the transla-
tion and the reversible rotation of the walls within ferromagneticmaterials, and the irreversible componentMirrwhich corresponds
to domain wall displacement against the pinning effect, M MrevMirr. In their model, Jiles and Atherton suppose that the
magnetization M is equivalent to the irreversible magnetizationMirrin the relation which binds the magnetic effective fieldHeand
the magnetic field H:
He HaMirr 1
In this relation, they suppose that the reversible magnetization is
negligible compared to the irreversible magnetization Mirr and
they obtained the differential irreversible susceptibility in the
simplified form:
dMirr
dH
ManMirr
kdaManMirr 2
And finally they determined the total differential susceptibilitydM=dH in the usually form:
dM
dH
1cManMirr
kdaManMirrc
dManHe
dH 3
The hypothesis of neglecting Mrev in the relation (1) has not a
physical meaning and conduct to imprecisely result in general
case. Our purpose is to use the physical relation relating the
magnetic effective field He and the magnetic field H:
He HaM 4
By using Eq. (4) the differential irreversible susceptibility is
obtained as
dMirrdH
ManMirr 1acdMandH
kda1cManMirr
5
We can see clearly that ifc=0 in Eq. (5) this relation becomes as
the same form as Eq. (2) because the parameter cis related to the
reversible magnetization as
Mrev cManMirr 6
After several manipulations, given in Appendix A, we arrived
at the following relation of magnetic total differential suscept-
ibility:
dM
dH
1cManMirr kcddMandHe
kda1cManMirrkcaddMan
dHe
7
As we can see, this relation (7) of MJA model is completely
different from the relation (3) of JA model.
3. Determination of the modified JilesAtherton parameters
The total differential susceptibility can be written by using
1cManMirr ManMas
w dM
dH
ManM kcddMandHe
kdaManMkcaddMan
dHe
8
or
w Z
kdaZ 9
with the following complementary relationships:
Z ManM kcddMandHe
10a
Man Ms coth He
a
a
He
10b
dMandHe
Ms
a 1coth2
Hea
a
He
2 ! 10c
d 1;
dH
dtZ0
1;dH
dto0
8>>>: 10d
The parameters of the MJA model have different expressions
compared to those of JA model and are determined from
experimental data using the differential susceptibilitywinat initialpoint (H=0,M=0), the differential susceptibility wrat remanence
point (H=0, M=Mr), the differential susceptibility wc at thecoercive point (H=Hc, M=0) and the differential susceptibility
wmat the loop tip (H=Hm, M=Mm).From these measured magnetic properties, the parameters
governing the hysteresis equations are obtained from the new
expression of the total differential susceptibility (9).
The initial susceptibility is calculated at the initial point (H=0
andM=0) of the initial magnetization curve with d=+1:
win dM
dH
H 0; M 0
ZinkaZin
11
The parameter c is associated to the reversible magnetization, at
the beginning of the initial magnetization curve and is deter-mined using the initial susceptibility
With
dMandHe
H 0;M 0
Ms3a
In this conditionZin kcMs=3a, whena and a are initially known,the parameter c is determined by using the Eq. (11) and is
expressed as
c 3awin
1awinMs12
At the coercive point (H=Hc, M=0) with d= +1, the differential
susceptibilitywcis given by
wc dM
dH
H Hc;M 0
ZckaZc
13
with
Zc ManHc ckdManHc
dHe
The relationship between the parameters k and wc is obtainedusing Eq. (13), whena,aandcare previously determined, Eq. (13)can be written as follows:
fck Zc1awckwc 0 14
This is the first nonlinear equation, the term Zcis function of the
parametersk, cand a .
M. Hamimid et al. / Physica B 405 (2010) 194719501948
8/11/2019 False Position.pdf
3/4
ARTICLE IN PRESS
At remanence point (H=0, M=Mr), the differential suscept-
ibilitywris given with d= 1:
wr dM
dH
H 0;M Mr
Zr
kdaZr15
with
Zr ManMrMr ckddManMr
dHe The relationship between the parameter a and the susceptibilityat the remanence point wris given by
fra Zr1awr kwr 0 16
frrepresents the second nonlinear equation and is extracted from
Eq. (15), where the term Zr is also function ofk ,a , cand a.Finally, the use of the coordinates of loop tip (H=Hm,M=Mm),
withd= +1, the differential susceptibilitywmcan be written as
wm dM
dH
H Hm ;M Mm
ZmkaZm
17
with
Zm ManHm; MmMm ckdManHm; Mm
dHe
The parameterais determined using the differential susceptibilitywm at the loop tip; on the other hand, parameters c, a and k areknown and the parametera is obtained by solving:
fma Zm1awmkwm 0 18
The last nonlinear equation is given by Eq. (18) where the termZmis function ofk ,c,a and a .
The parameter Ms is the magnetization value at strong
saturation. It is probably the easiest parameter to obtain; this is
often known for a particular material; in the most cases it is given
by the constructor of the metallic sheet.
The parameters c, k, a and a are nonlinearly coupled. Todetermine them, we propose to solve successively the Eqs. (12),
(14), (16), and (18) using an iterative algorithm coupled with theFPM.
4. Identification technique
This technique is based on the coupling of an iterative
algorithm with the FPM to identify the parameters of both MJA
and JA models. The expressions giving the JA model parameters
are the same as in Ref.[4]. In the MJA the parametercis obtained
by using Eq. (12) with a given initial values of the parameters aanda, and we introduce all of them into Eq. (14) which is resolved
by FPM to find a first estimation ofk.
The parametersa and a are calculated from Eqs. (16) and (18)using the same FPM. The current values ofa and a are used to
determine the new value of the parameter c. The calculationprocess is repeated until the satisfaction of the criterion
jcnewcold=cnewjre, wheree is a given small number.The FPM starts with two pointsaandb such thatc(a) andc(b)
are of opposite sign, implying that the function c has at least one
zero in the interval [a, b] [10]. The method has to produce a
decreasing sequence of intervals [ak,bk], which all contain a zero
ofc . In step k, the numberbk is calculated by
bk ak akbk
cakcbkcak 19
Ifc(ak) andc(bk) have the same sign, thenak+1=bkand bk+1=bk,
elsebk+1=bkandak+1=ak. This process is repeated until c(bk)E0.
For example we calculate the parameter a with FPM using
Eq. (19): we have from Eq. (16), fra Zr1awr kwr and we
seek a such as f(a)=0; let start with two points a1=a anda2=bsuch asf(a1) andf(a2) are of opposite sign. In step k, the numberbk, where bk is the solution in step k, is calculated by
1. bk akakbk=fakfbkfak,
2. iff(ak) f(bk)40 thenak+1=bkand bk+1=bk; else bk+1=bk andak+1=ak,
3. if |f(bk)|rethen end; else go to 1 (eis a given small number).
5. Results and discussion
To identify MJA and JA parameters, the algorithm proposed
needs some measured parameters that are presented in Table 1.
These ones are extracted from the experimental loop. The
identified parameters are presented inTable 2.
The JA model and the MJA model hysteresis loops, using the
identified parameters, respectively, ofTable 1are compared to the
experimental ones.Figs. 1 and 2show the hysteresis loops of both
models. A good agreement between measured and simulated
hysteresis loops is obtained for both MJA and JA models; this
Table 1
Parameters extracted from experiment.
Parameters Measured
win 184.12wm 0.0443wr 1.9725 10
3
wc 9.974 105
Hm (A/m) 1.039 103
Hc(A/m) 69.37
Mm (A/m) 1.134 106
Mr(A/m) 8.905 105
Table 2
Identified JA model parameters.
Identified parameters JA model MJA model
Ms(A/m) 1.18 106 1.18 106
a (A/m) 46.858 46.9605
k(A/m) 81.10 79.001
a 1.4843 104 1.507 104
c 0.0219 0.0214
-1500 -1000 -500 0 500 1000 1500
-1.5
-1
-0.5
0
0.5
1
1.5
magnetic field H (A/m)
magneticfluxdensityB
(T)
measured
JA
MJA
Fig. 1. Measured and simulated magnetic flux density.
M. Hamimid et al. / Physica B 405 (2010) 19471950 1949
8/11/2019 False Position.pdf
4/4
ARTICLE IN PRESS
shows the effectiveness of the technique used to identify modelparameters on the one hand, on the other hand we can see that
the MJA model hysteresis loop fits better the experimental one.
The parameters obtained by this technique have a physical
interpretation, such as the parameter k, in the case of soft
magnetic materials, it is like similar to the coercive field.
6. Conclusion
The employment of the exact expression of the effective
magnetic field relation by using the total magnetization instead of
the irreversible one improves the hysteresis evaluation of the
JilesAtherton model. A technique based on iterative algorithm
coupled with the false position method is introduced to identify
the parameters of the MJA and the JA models to derive thehysteresis loops. These parameters are compared to the experi-
mental data. The MJA model hysteresis loop fits better the
experimental loop compared to the JA model hysteresis loop.
This model can be improved by taking into account the
dynamic behavior of soft and hard magnetic materials.
Work context
The present work is carried out within an AlgerianFrench
cooperation PHI-Tassili program.
Appendix A
The total magnetization given by JilesAtherton is
M MrevMirr A:1
By using Eq. (6) the total magnetization becomes
M 1cMirr cMan A:2
The total differential susceptibility is given by
dM
dH 1c
dMirrdH
cdMandH
A:3
We can also write this susceptibility as the following form:
dM
dH 1c
dMirrdHe
cdMandHe
dHedH
A:4
wheredMan=dHe is given by Eq. (10c), andHe is given by Eq. (4)
dMirrdHe
ManMirr
kd A:5
The expression (A.5) is given in Ref. [4], by insert Eq. (A.5) in
Eq. (A.4) we obtained
dM
dH 1c
ManMirrdHe
cdMandHe
dHedH
A:6
By replacing 1cManMirr ManM and dHe=dH by
1adM=dHthe expression (A.6) becomes
dM
dH
Zkd
1a
dM
dH
A:7
Z is given by Eq. (10a).Finally the total differential susceptibility is obtained as given
by Eq. (7).
References
[1] D.C. Jiles, D.L. Atherton, J. Appl. Phys. 55 (1984) 2115.[2] E. Dlala, J. Saitz, A. Arkkio, IEEE Trans. Magn. 42 (8) (2006) 1963.[3] M.L. Hodgdon, IEEE Trans. Magn. 24 (6) (1988) 3120.[4] D.C. Jiles, J.B. Thoelke, M.K. Devine, IEEE Trans. Magn. 28 (1992) 27.[5] J. Izydorczyk, J. Magn. Magn. Mater. 302 (2006) 517.[6] K. Chwastek, J. Szczyglowski, J. Magn. Magn. Mater. 314 (2007) 47.[7] E.D.M. Hernandez, C.S. Muranaka, J.R. Cardoso, Physica B 275 (2000) 212.[8] P. Andrei, L. Oniciuc, A. Stancu, L. Stoleriu, J. Magn. Magn. Mater. 316 (2007)
330.[9] A. Salvini, F. Riganti Fulginei, IEEE Trans. Magn. 38 (2002) 873.
[10] A. Quarteroni, A. Sacco, F. Saleri, Me thode Numeriques, Algorithmes, Analyseet Applications, Springer, 2007.
0 200 400 600 800 1000
0
0.2
0.4
0.6
0.8
1
1.2
1.4
magnetic field H (A/m)
magneticflu
xdensityB(T)
measured
JA
MJA
Fig. 2. Measured and simulated magnetic flux density (zoom up).
M. Hamimid et al. / Physica B 405 (2010) 194719501950