IEEE TRANSACTIONS ON MAGNETICS, VOL. 42, NO. 10, OCTOBER 2006 3159
Using Experimental FORC Distribution as Input
for a Preisach-Type Model
Laurentiu Stoleriu and Alexandru Stancu
Department of Solid State and Theoretical Physics, Alexandru Ioan Cuza University, 700506 Iasi, Romania
In this paper, we present a method of using the experimental first-order reversal curves (FORC) distribution as input for Preisach-type
models. In order to be able to calculate in the model the integral over the FORC distribution we designed an interpolation algorithm
of the three-dimensional (3-D) distribution. The algorithm is validated for a simple case—the output of a classical Preisach model—and
then is used for realistic, asymmetrical FORC distributions. Several Preisach-type models are tested for the same FORC distribution
and it is shown that the PM2 model gives the best results in simulating magnetization curves starting from the FORC diagram.
Index Terms—Magnetic hysteresis, magnetic materials, magnetization processes, modeling.
I. INTRODUCTION
P
HENOMENOLOGICAL approach to magnetic modeling
is able to give good results in a much shorter time and de-
manding lesser computing performance than the physical (mi-
cromagnetic) approach.
The first-order reversal curves (FORC) method was intro-
duced by Mayergoyz [1] as a method to identify the Preisach
distribution [2] of a system which can be fully described by the
classical Preisach model (CPM). In CPM, a particulate magnetic
system is characterized by two independent statistical distribu-
tion of coercive and interaction fields. Mayergoyz proved that,
in order to be correctly described by a CPM, the system should
obey to the wiping out and congruency properties [1]. While
most of the real systems meet the terms of the wiping-out prop-
erty, the congruency property was hardly ever verified on real
systems.
The lack of experimental systems which can be correctly de-
scribed by CPM determined the evolution of many more re-
alistic Preisach-type models—one of the most widely used is
the generalized moving Preisach model which adds to the CPM
hypotheses the reversible component of the magnetization pro-
cesses and a mean field interaction term. In the same time, the
interest in the FORC method diminished, especially because of
the numerical errors introduced by the second-order derivative
of the experimental data requested by this method of identifica-
tion.
Recently, Pike et al. [3] introduced a versatile numerical
method of evaluation of the FORC diagram starting from
measured data. They also have separated the method from
its origin—the CPM—and suggested to use it only as an ex-
perimental tool which can give information about magnetic
systems. Since then, a considerable work has been undertaken
in the field of the experimental FORC diagram interpretation.
The aim of parameter identification in Preisach modeling is
to find algebraic functions which, when used as input for the
model, fit the experimental observations and are able to predict
Digital Object Identifier 10.1109/TMAG.2006.880112
sample’s magnetic behavior. The main point in measuring
FORCs and in the calculation from these experimental data
FORC diagrams is especially related to the evaluation of in-
teraction and coercive fields of the magnetic entities contained
in a sample and to the evaluation of the ratio between the re-
versible and the irreversible magnetization processes. However,
it was recognized by all those who are currently using this
experimental method that the FORC diagram, even it is closely
related to the well known Preisach distribution [2] (of coercive
and interaction fields as well as the reversible distribution) is
not identical to this distribution. It has been shown [4] that for
real systems the interaction field distribution is changing as a
function of the magnetic state so the FORC diagram should be
seen like an averaged photography of a moving distribution.
It is important to emphasize that ideally, if an experimental
FORC is used as the Preisach distribution in a CPM, the set of
first-order reversal curves should be properly simulated. How-
ever we have shown in [5] that if we measure a set of second-
order reversal curves we obtain a different diagram while the
CPM will give a diagram identical with the FORC diagram. As
in many cases we have observed that the differences are not
too important, this gave us the idea to implement a Preisach-
type model which uses the experimental FORC distribution as
Preisach distribution and to analyze the quality of the predic-
tions made by this model.
II. THE MODEL
To obtain a point on the FORC starting on the descending
branch of the major hysteresis loop (MHL), the following field
sequence is applied to the sample: , where
is a field sufficient to saturate the sample, , named reversal
field, is a field in the domain and is the ac-
tual field at which the sample magnetic moment—noted with
—is measured; the sign minus is indicating that
the FORC is measured on the descending branch of MHL. The
FORC distribution is given by the second-order mixed deriva-
tive
(1)
0018-9464/$20.00 © 2006 IEEE