International Journal of Thermal Sciences 60 (2012) 195e204

Contents lists available at SciVerse ScienceDirect

International Journal of Thermal Sciences

journal homepage: www.elsevier.com/locate/ijts

Coupled electro-magneto-thermal model for induction heating process

of a moving billet
Kee-Hyeon Cho*
Energy & Resources Research Department, Research Institute of Industrial Science & Technology (RIST), San 32, Hyoja-dong, Namgu, Pohang, Kyungbuk 790-330, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: The purpose of this study is to present a coupled electro-magnetic and thermal model for numerical
Received 11 November 2011 analysis of an induction heating system including the workpieces moving relative to the inductors. In this
Received in revised form paper, a finite element method-based numerical analysis of a low-frequency (60 Hz) induction heating
4 March 2012
system for the one-dimensional solution of a stationary circular billet and the two-dimensional solution
Accepted 2 May 2012
considering the dynamic effect of circular billets moving along the skid rails with constant speed are
Available online 8 June 2012
presented and compared against each other. The non-linearities of both the electro-magnetic and
thermal material properties are also taken into account in the model. The computational results have
Keywords:
Induction heating
been compared with experimental data. As a result, it is suggested that the presented numerical model
Electro-magneto-thermal may be a very cost-effective tool in predicting the temperature of a workpiece in a variable flux field
Eddy currents where the interested workpieces undergo an arbitrary change in the electro-magnetic fields. It is possible
FEM (finite element method) to obtain some preliminary results more accurate than those calculated from previous works using
Numerical simulation a stationary model on electro-magnetic field and temperature distribution of workpieces by applying the
Modeling presented numerical model.
Ó 2012 Elsevier Masson SAS. All rights reserved.

1. Introduction Maxwell equations for electro-magnetic phenomena. Numerical

techniques in electroheat can be traced way back to the early 1960s
Induction heating is a process that uses an AC electro-magnetic with the FDM (Finite Difference Method) to determine the
field to induce heat in a workpiece and has been used to heat temperature in ferromagnetic cylinder [3]. Excellent reviews of
electrical workpieces since the early 1930s. Industrial applications numerical methods in induction heating and melting applications
of the technology include heat treating and metal melting, single can be found by Lavers [4]. Recently, the Finite Element Method
crystal growing, semiconductor wafer processing, high speed (FEM) as presented in this paper has been used extensively in many
sealing and packaging, and curing of organic coatings [1]. problems associated with electroheat applications such as induc-
If an AC field is placed in close vicinity to a conductor, eddy tion heating. The numerical calculation of induction heating
currents are induced in the workpiece. The induced eddy currents processes has become increasingly common [5].
release energy in the form of heat, which is then distributed Many researches have been done concerning the theoretical
throughout the workpiece. The hysteresis loss also forms an addi- model and the numerical simulation of induction heating devices.
tional source for heating when dealing with a magnetic problem. Enokizono and Tanabe [6] studied a numerical analysis of the
This kind of heating is in addition to those caused by eddy currents. magnetic field on high-frequency induction heating. The depen-
It originates from magnetic domain friction in ferromagnetic dence of various magnetic properties on temperature was included.
materials [2]. The required characteristics were obtained experimentally. Sade-
The induction heating is a complex process involving both ghipour et al. [7] investigated the induced power density profile
electro-magnetic and thermal phenomena. Electrical energy and transient temperature distribution, using finite element anal-
research and development deals with the use of electrical, ysis, for a flanged tube of AISI 1040 steel. Ernst et al. [8] presented
magnetic and electro-magnetic phenomena to produce, transfer a new and original electromagnetic-thermal coupling modeling
and transform this type of energy. This process is governed by the tool based on the Matlab two-dimensional finite elements solver
for the optimization of a single SiC crystal inductive elaboration
* Tel.: þ82 54 279 5267; fax: þ82 54 279 6309. process. Arita et al. [9] analyzed the results of finite element anal-
E-mail address: khcho@postech.ac.kr. ysis of high-frequency induction heating problems considering

1290-0729/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ijthermalsci.2012.05.003
196 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204

Nomenclature t time, s
T temperature, K
A magnetic vector potential, Wb m1
A outer area of steel workpiece, m2 Greek symbols
B magnetic flux density, T Dt time step size, s
Cp specific heat, J kg1 K1 3 emissivity
E electric field intensity, V m1 mm relative magnetic permeability
f frequency, Hz r density of workpiece, kg m3
g gravitational acceleration, m s2 sc electrical conductivity, U1 m1
H magnetic field intensity, A m1 s StefaneBoltzmann constant, W m2 K4
Js excitation current density,
pffiffiffiffiffiffiffi A m2 u angular frequency, rad s1
j complex number ð 1Þ
k thermal conductivity, W m1 K1 Subscripts
L length of workpiece, mm N ambient
n normal direction eddy eddy current
q000
eddy
heat power by eddy currents in the workpiece, W m3 i inner
qs energy losses by radiation, W o outer
r, q, z cylindrical coordinates, m, rad, m s surface

temperature dependence of material characteristics using three- the No. 3 coil and the holding chamber, as illustrated in Fig. 1. It will
dimensional finite element analysis in order to correctly express be assumed that the pure heating parts (the No. 1 through the No. 3
the induction heating coil’s shape. Cajner et al. [10] developed coil and the gap parts including the air region between the pre and
a simulation program for the axially symmetric workpieces and post coil), except for the holding chamber, have an axial symmetry.
compared with the measured values of surface hardness and For the simplicity, we describe a model only consisting of work-
hardening depths. Kawaguchi et al. [11] presented results of finite piece, coil, and air in the computational domain, as illustrated in
element analysis of induction heating problems considering Fig. 2.
temperature dependence of material characteristics. Recently, Jiang We consider that the properties of the materials used in this
et al. [12] proposed an optimal heating/cooling strategy to achieve calculation, i.e. the relative magnetic permeability, the resistivity,
uniform temperature distribution along the radius of a cylinder. the thermal conductivity, the density, and the specific heat may be
Tibouche et al. [13] developed a software tool on the base of the dependent on temperature.
finite difference method couples with a semi-analytic one. Kranjc The model of this analysis utilizes an indirect method in which
et al. [14] investigated the characteristics of induction heating both the electro-magnetic and thermal problems are solved sepa-
process both numerically and experimentally. More recently, Bar- rately. Because of the different time constants of the electro-
glik [15] investigated three-dimensional analysis of electro- magnetic and thermal problems, the eddy current problems are
magnetic and temperature fields in transverse flux induction solved as a time-harmonic and the thermal problems are solved as
heater for thin strips and compared the result with measured data. a non-stationary [17].
However, most of theoretical model or numerical simulation
that has been done so far is for the cases applied in the stationary 2.1. Electro-magnetic field
high-frequency inductor systems. Furthermore, all these studies
can not be expected to provide a good solution in a region close to The electro-magnetic phenomena under consideration are
the inductors because induction heating has its inherent drawback governed by Maxwell equations. Therefore, the electro-magnetic
of non-uniform heating due to the so-called skin effect, end effect field is calculated using the Maxwell equations to predict the
and electro-magnetic transverse edge effect. The moving conductor generated heat power in the workpiece by using an induction
or coil effects, however, are of importance when one of them is heating. The Maxwell equations can be simplified using a quasi-
moving relative to the other and goes through the end of the steady state approximation as follows [18]:
inductors. This is because the magnetic field gradients become very
high in that region and finally result in great inaccuracy in pre- VH ¼ J (1)
dicting the temperature distribution of workpieces.
Therefore, in this paper, a method using a commercial package, V  E ¼ juB (2)
ANSYS [16], is given for analyzing the one-dimensional and two-
dimensional problems of an induction heating system. The V$B ¼ 0 (3)
primary goal of the present work is to present a coupled and more
accurate mathematical model taking the dynamic effect of moving where H is the magnetic field strength, J is the current density, B is
workpieces into account for an induction furnace system, thereby the magnetic field intensity, E is the electric field intensity, u is the
obtaining results that are more accurate than those calculated from angular frequency of the current (u ¼ 2pf, f ¼ 60 Hz in the present
previous works on electro-magnetic field and temperature distri- study), and j is the complex number defined by j2 ¼ 1. Hysteresis
bution of workpieces. and anisotropy are neglected.
Introducing a magnetic vector potential A as
2. Mathematical model
B ¼ VA (4)
We solve eddy current and thermal problems in order to it is assumed that the current in the inductors is time-harmonic and
simulate an induction heating system, which is used to produce heat is generated in the workpiece by eddy currents. From the
seamless tubes and usually consists of four parts: the No. 1 through above equations, we derive the governing equation as follows:
 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204 197

Fig. 1. Three dimensional model of an induction heating furnace system with a holding chamber.

  2.2. Temperature field

1
jusc A þ V  V  A ¼ Js (5)
mm
The mathematical model that represents the thermal
phenomena is based on a diffusion equation as follows:
where sc is the electrical conductivity of the material, mm is the
relative magnetic permeability of ferromagnetic materials    
depending on the temperature T, and Js is the excitation current vT 1v vT v vT
rCp ¼ rk þ k þ q000 (10)
density loaded in the inductors [19]. vt r vr vr vz vz eddy

It is assumed that the electro-magnetic field is axi-symmetric so

where T is temperature, t is time, r is mass density, Cp is specific
that the magnetic vector potential A and the current density Js,
heat, and k is thermal conductivity.
which is constant and independent of time, have only one angular
For one-dimensional analysis, the heat loss due to the surface
component.
radiation effects of workpiece is considered; however, it is not
As stated above, it was modeled using an axi-symmetric
considered for two-dimensional analysis. As described in Fig. 2, the
approximation and the geometry depicting this approximation is
thermal boundary conditions for a one-dimensional outer surface
shown in Fig. 2. As depicted in Fig. 2, the boundary conditions
are as follows:
associated with the governing Eq. (5) are as follows:
 
Flux parallel condition : A ¼ 0 (6) one  dimensional outer surface : qs ¼ A3 s T 4  TN
4
;
at r ¼ ro ð11Þ
vA
Flux normal condition : ¼ 0 (7)
vn
where n is the unit normal to the boundary. On the boundaries of vT
one  dimensional inner surface : ¼ 0; at r ¼ ri
the computational domain which are parallel to the symmetry axis, vn
flux-parallel conditions are imposed (A ¼ 0 at r ¼ 0) because the (12)
direction of induced magnetic field is parallel to the longitudinal
direction of the workpiece. It is also assumed that the magnetic vT
field external to the coils is zero, so the nodes on the outer edge are two  dimensional outer surface : ¼ 0; at r ¼ ro
vn
coupled (A ¼ constant) such that the proper flux-parallel boundary (13)
condition is imposed. Unlike this, on the other boundaries which
are normal to the symmetry axis, flux normal boundary conditions
vT
which are the natural boundary conditions are imposed. two  dimensional inner surface : ¼ 0; at r ¼ ri
Solving the above equation, we can get the generated volu-
vn
metric heat power (W/m3) by the eddy currents which are induced (14)
by a time-harmonic electro-magnetic field as follows:
where qs (W) is the radiated power from a body of area A (m2) at
1 temperature T (K), 3 is the emissivity (0.7 in the present study), s is
q000
eddy ¼ sc u2 ðAr þ jAi ÞðAr  jAi Þ (8) the StefaneBoltzmann constant (W/m2/K4), TN is the ambient
2
temperature (K), ri is the inner radius of an interested workpiece,
1   and ro is the outer radius of an interested workpiece. It is
q000
eddy ¼ sc u2 A2r þ A2i (9) noted that the convection boundary condition is neglected since
2
its order is 3% compared to the radiation effect when there is
where r denotes the real part and i denotes the imaginary part in a large difference in temperature between the workpiece and the
the complex conjugate. ambient [20].
Fig. 2. 2-D axi-symmetric approximation, boundary conditions and FE mesh plot for an induction heating furnace: (a) 1-D axi-symmetric electro-magnetic and thermal domain; (b)
2-D axi-symmetric electro-magnetic and thermal domain; (c) mesh plot of electro-magnetic domain (2-D).
 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204 199

2.3. Calculation method

In this study, ANSYS [16], a finite element code with Galerkin’s

formulation, was used to predict the power densities and temper-
ature distribution of workpieces. “Program chosen solver” was
chosen as an equation solver and “Newmark algorithm” with 2nd
order transient integration parameter (0.5) was selected as a time
integration scheme for thermal analysis. In the harmonic solution,
“full option” was used as the solution method and “harmonic” was
chosen as the solution type.
The coupled electro-magneto-thermal analysis, as shown in
Fig. 3, starts from the solution of a time-harmonic equation and the
power densities within the workpiece are computed. Then, they are
used together with the initial nodal temperature as the input for
the transient thermal analysis. When carrying out a global finite
element simulation by moving the coil outlines and mesh in the
global mesh, there is a possibility that the mesh undergoes large
distortions leading to inaccurate results and numerical instability.
The strategies we have adopted in order to minimize such
a problem are as follows: (1) The area where the coil will move
through is initially defined and meshed separately. (2) The electro-
magnetic properties of this area are moved back and forth from air
properties to inductor material properties. After updating the value
of electro-magnetic properties, the elements which are resident on
moving areas including the workpiece and air zone are newly
generated, allowing movements by steps with lengths determined
Fig. 3. Flow chart of the numerical analysis of the coupled electro-magneto-thermal
by the time step size entering the coils. Then, the constraint
problem.
equations are coupled with the interface nodes and the elements
that reside on the interface line where the moving domain and the
fixed domain, as illustrated in Fig. 2, coincide geometrically. With
Table 1
this modeling technique and the sequential solution methods, the Physical dimensions.
coils, air and workpieces can be remodeled to get a better repre-
sentation of the real current density distribution. After this is done, Item Parameter Units

a new magnetic field calculation is performed with new elements. Workpiece material Mild steel (0.23% carbon)
Workpiece dimension Billet (47.5(ri)  148(ro)  148 (L)) mm
This iteration is continued until the heating cycle ends.
Workpiece shape Hollow cylinder
In general, the electro-magnetic problem in a typical induction Coil number of turns 93/each stage
heating system is solved as a time-harmonic and is stationary in Coil material Copper
time because the electro-magnetic time constant, based on the Size (Equivalent) 27  1475/1 stage mm
frequency of the electric power supply, is typically of the order of Current density 8.073  106 A/m2
Frequency 60 Hz
10 m s. That is to say, due to the different time scale of the two Total heating time 720 s
phenomena, it is assumed that the solution of the electro-magnetic Initial temperature 25 
C
problem is valid on a time interval during which the physical Computational domain (3 stages) 1480 (W)  4830 (H) mm
properties of the workpieces don’t change too much due to the
increase in temperature resulting from the Joule effect. We will
then use the result to compute the source term to be plugged into
the heat equation [21]. Keeping this problem in mind and consid- magnetic fields. Other simplifying assumptions used are discussed
ering the stability of problems, the appropriate time step Dt is 0.05 s in the following sections.
(lower than 2.1 s obtained from stability condition) at the beginning
of the heating and 2 s since the Curie temperature is exceeded.
Therefore, we can neglect the electro-magnetic transient and safely 3. Results and discussion
assume that, within that time step, the electro-magnetic process is
in a steady state. The iteration numbers needed were 7 for one step 3.1. One-dimensional calculation
of the electro-magnetic field problem and 13 for one step of the
heat transfer problem. The first example is a heating of 1-D having a surface radiation
The mesh independence tests are carried out in order to ensure loss at the outer surface of the workpiece during the residence on
mesh independent results. In general, the computation of eddy the space between the No. 1 coil and the No. 2 coil or between the
currents does not cause major difficulties, provided a sufficiently No. 2 coil and the No. 3 coil, as illustrated in Fig. 1. Computational
fine mesh is used in the skin depth region of the conductors. The domain for 1-D analysis includes workpiece, coil, and air region like
finer mesh was obtained by re-meshing the coarser mesh. The as 2-D domain shown in Fig. 2. Only boundary conditions are
independence of the solution with respect to the mesh size was updated as time goes by.
checked by examining the maximum surface temperature for each Numerical simulations are carried out for physical dimensions,
test. The total number of quadrilateral-shaped elements in the as shown in Table 1, and the electro-magneto-thermal properties
electro-magnetic analysis was 8047, the number of nodes was which are dependent on temperature [22], as illustrated in Fig. 4.
13,758 and the elements number for the corresponding thermal Table 2 also depicts the chemical composition of the workpiece
model was 2002, considering the skin depth and the gradient of used in this calculation. As stated above, the boundary conditions
200 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204

Fig. 4. Thermal and electro-magnetic properties with temperature: (a) resistivity; (b) relative permeability; (c) volumetric specific heat; (d) thermal conductivity.

and mesh plot used for the numerical analysis are represented in losing its magnetism locally, i.e. it passes from a ferromagnetic state
Fig. 2. to a paramagnetic state.
A circular billet with the size listed in Table 1 was heated inside The temperature drop is about 34  C due to radiation heat loss at
an inductor of 93 turns (length 1475 mm, pitch 102 mm) and was the outer surface during the open zones of the furnace. Also, as
modeled as an equivalent rectangular shape with 27 by 1475 mm. shown in Fig. 5, the soaking time is about 70 s and the temperature
The speed of the billet movement is 6.72 mm/s inside the induction
heating devices. It is also assumed that the ambient temperature
around the furnace TN is 50  C (323 K) and the process is carried out
over a total residence time of 720 s (12 min); the end effects are
1400
negligible.
Fig. 5 shows the profiles of unsteady temperature distribution in
the workpiece along the radial direction at three different posi- 1200
tions: inner, center and outer surface of the workpiece. As can be
seen in Fig. 5, relatively large differences in temperature variation 1000
occurred; specifically, it was heated unevenly along the surface
Temperature ( C)

during the residence time at the No. 1 coil. The peak profile 800
o

occurring at the outer surface is caused by the intensive magnetic

fields which are concentrated on the outer surface of the work- Inner surface
600 Center
piece. Also, the temperature at the outer surface rises very quickly
Outer surface
to around the Curie temperature which passes from a ferromag-
netic state to a paramagnetic state, about 761  C for the mild steel. 400
There seems to be a sudden cooling because the workpiece starts
200

Table 2 0
0 100 200 300 400 500 600 700
Chemical composition of workpiece material used in the calculation (unit: wt%).
Time (sec)
Type C Si Mn S P
Mild steel 0.23 0.3 0.6 0.05 0.05 Fig. 5. Predicted temperature history of each position with residence time for 1-D case
(stationary).
 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204 201

Fig. 6. Magnetic potential contours with residence time: (a) at t ¼ 1 s; (b) at t ¼ 4 min; (c) at t ¼ 8 min.
202 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204

distribution in the workpiece along the radial direction of it of field in the air gap which is empty between the workpiece and
becomes nearly uniform. the inductor is unaffected by the workpiece but eddy currents in
the workpiece reduce to the total flux by the eddy current theory.
3.2. Two-dimensional calculation As a result, the bulging profiles in the magnetic flux occur because
the temperature has exceeded the Curie temperature and they
The second example is a heating of 2-D having adiabatic result in a loss of magnetism and increasing the skin depth (about
conditions at the inner surface of the workpiece like the 1-D case. 1.8 mm in this calculation) with a similar pattern as time goes by.
And the other conditions are the same as the first example (one- Fig. 7 shows the isothermal lines within the workpiece in which
dimensional analysis). Unlike the one-dimensional analysis, the we are interested with the time variation after charging it into the
particular treatment of the interested workpiece in the computa- induction furnace. In Fig. 7a, the temperature distributions were
tional domain, as shown in Fig. 2, is given: the dummy workpieces nearly parallel along the length direction of the workpiece just after
are used in order to escape, more or less, an inaccurate magnetic charging the workpiece (at t ¼ 30 s). However, the isothermal lines
field generated near the ends of the interested workpiece. The lie across the workpiece as the workpieces are transferred to the
purpose is to obtain as accurate a temperature distribution as furnace with a continuous speed and the temperature differences
possible for only a target workpiece inside the induction heating become lower within the interior region of workpiece, as shown in
devices due to the end effect, as illustrated in Fig. 2a. Fig. 7b, c, and d.
The typical contours of the magnetic vector potential repre- Fig. 8 shows the temperature history within the workpiece with
senting the dynamic effect of the moving workpiece with the time the positions (1)e(9) of the workpiece, as shown in the schematic
variation are shown in Fig. 6. As depicted in Fig. 6, the magnetic diagram. It is obvious that the heating pattern is similar to the result
vector potential in the far field region where it is far from the outer of one-dimensional calculation (see Fig. 5) but the profile of the
surface of the workpiece is diminished regardless of time variation temperature history at position (6), the peak profile, is steeper than
after charging the workpieces. On the other hand, the contour lines that of 1-D case. We think that this result can be caused by the
are concentrated along the outer surface of workpiece. The closed reason that the fluxes formed inside the moving workpiece are
loop patterns of magnetic flux density are changed extremely as more heavily concentrated than with the one-dimensional case.
time goes by after the workpieces are charged into the induction Because of the same reason, the temperature differences inside the
furnace. Also, as depicted at the surface, we know that the strength workpiece are higher than those of the one-dimensional case. Also,

Fig. 7. Predicted isothermal lines of each position with residence time: (a) at t ¼ 30 s; (b) at t ¼ 4 min; (c) at t ¼ 8 min; (d) at t ¼ 12 min.
 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204 203

Fig. 9. Predicted temperature history along the radial direction with residence time.

Finally, Fig. 10 presents a comparison between the results of

surface temperature measurements obtained with two-wavelength
pyrometers (Williamson Pro 220) and the predicted results shown
in Figs. 5 and 8. For the two measured data, the upper curve
represents the temperature distribution measured at just the ends
of the No. 3 and the lower part represents the temperature distri-
bution measured at just the ends of the No. 2. As shown in Fig. 10,
discrepancies are also found, most likely due to the measurement
uncertainty of pyrometers, an idealization of the geometric reality
(axi-symmetric geometry), the problem of workpiece alignment
within the coil bore, and so on. However, generally good agreement
Fig. 8. Predicted temperature history of each position with residence time for 2-D case is observed for three cases (1-D analysis, 2-D analysis and experi-
(non-stationary): (a) location (1)e(6); (b) location (7)e(9). mental results), though the measured positions for compared
temperature are fixed.
the soaking time is about 65 s if the interested positions are
confined to points (7), (8), and (9) because they start from the inlet
at t ¼ 0 s, as shown in Fig. 8. A close scrutiny of Fig. 8 also reveals
1400
that the temperature distribution along the radial direction of the
workpiece becomes nearly uniform from the No. 2 coil as in the
2-D Predicted (No.3)
result of the one-dimensional case (see Fig. 5). In addition, it is
observable that the temperature difference within the workpiece is 1200
somewhat higher than of the one-dimensional case; this may be 2-D Predicted (No.2) 1-D Predicted (No.3)

caused by the uneven magnetic fields due to the dynamic effects

Temperature ( C)

1-D Predicted (No.2)

o

presented in this calculation, unlike the 1-D case. Furthermore,

interestingly, in comparing the temperature history patterns of the 1000

1-D case (see Fig. 5) and the 2-D case along the positions (7)
through (9) (Fig. 8b), it was found that similar patterns were Temp. at No. 2 inductor
Temp. at No. 3 inductor
obtained, as shown in Figs. 5 and 8b. 800
Fig. 9 shows the temperature gradient distribution from the
inner surface of the workpiece along the radial direction (positions
(4) through (6) in Fig. 8) during the residence time. As shown in
Fig. 9, we know that there are some clear temperature differences 600
8:16:37 8:18:17 8:19:57 8:21:37 8:23:17 8:24:57
along the radial direction inside workpiece until t ¼ 2 min (position
Operating time(hour:min.:sec)
(1)) after being charged into furnace but become nearly uniform
after then (t ¼ 4 min) i.e. when the workpiece goes through the end Fig. 10. Temperature variation on surface of workpiece at the ends of the No. 2 and the
of the No. 1 inductor. No. 3 inductor.
204 K.-H. Cho / International Journal of Thermal Sciences 60 (2012) 195e204

4. Conclusions [6] M. Enokizono, H. Tanabe, Numerical analysis of high-frequency induction

heating temperature dependence of material characteristic, IEEE Trans. Magn.
31 (1995) 2438e2444.
In this paper, we have presented an axi-symmetric dynamic [7] K. Sadeghipour, J.A. Dopkin, K. Li, A computer aided finite element/experi-
model applicable to the workpieces moving relative to the induc- mental analysis of induction heating process of steel, Comput. Ind. 28 (1996)
tors. The advantage of this model is that it can increase the accuracy 195e205.
[8] R. Ernst, P. Grosse, L. Philippe, An electromagnetic-thermal coupling example
for predicting the eddy current loss and can be extended to the solved on a high frequency inductive system with the Matlab finite elements
problem with a variable flux field, where the interested workpieces solver, Int. J. Therm. Sci. 38 (1999) 481e491.
undergo an arbitrary change in the electro-magnetic fields. [9] H. Arita, T. Todaka, M. Enokizono, Thermal magnetic characteristic for high
frequency induction heating analysis, J. Appl. Phys. 91 (2002) 8317e8318.
The contribution of this paper is two-fold. First, it offers [10] F. Cajner, B. Smoljan, D. Landek, Computer simulation of induction hardening,
a coupled electro-magnetic and thermal model for numerical J. Mater. Process. Technol. 157e158 (2004) 55e60.
analysis of an induction heating system, highlighting the dynamic [11] H. Kawaguchi, M. Enokizono, T. Todaka, Thermal and magnetic field analysis of
induction heating problems, J. Mater. Process. Technol. 161 (2005) 193e198.
effect of the workpieces moving relative to the inductors. Second, it [12] H. Jiang, T.H. Nguyen, M. Prud’homme, Optimal control of induction heating
provides experimental evaluation of the presented model for for semi-solid aluminum alloy forming, J. Mater. Process. Technol. 189 (2007)
induction heating of a circular billet. 182e191.
[13] A. Tibouche, A. Bourouina, M.R. Mekideche, A. Djerdir, Solution of the tran-
The electro-magneto-thermal coupling problems that charac- sient electro-magneto-thermal problems fed by voltage in induction heating
terize the induction heating problems are analyzed and discussed. applications, Int. J. Electr. Power Eng. 3 (2009) 249e254.
The validity of our approach is also carried out, comparing [14] M. Kranjc, A. Zupanic, D. Miklavcic, T. Jarm, Numerical analysis and thermo-
graphic investigation of induction heating, Int. J. Heat Mass Transfer 53 (2010)
the results to those obtained from the stationary 1-D analysis and
3585e3591.
the measurements. Compared to the results of measurements, the [15] Barglik, Induction heating of thin strips in transverse flux magnetic field,
computational results of the presented model show more accurate in: S. Grundas (Ed.), Advances of Induction & Microwave Heating of Mineral
than that of the conventional or stationary model. and Organic Materials, InTech, Rijeka, 2011, pp. 181e206.
[16] ANSYS, 11.0 Documentation, SAS IP, Canonsburg, 2007.
[17] J. Nerg, J. Partanen, Numerical solution of 2D and 3D induction heating
problems with non-linear material properties taken into account, IEEE Trans.
Magn. 36 (2000) 3119e3121.
References [18] M.G. Pantelyat, M. Feliachi, Magneto-thermo-elastic-plastic simulation of
inductive heating of metals, Eur. Phys. J. Appl. Phys. (2002) 29e33.
[1] L.R. Egan, E.P. Furlani, A computer simulation of an induction heating system, [19] A. Bermúdez, D. Gómez, M.C. Muñiz, P. Salgado, Transient numerical simu-
IEEE Trans. Magn. 27 (1991) 4343e4354. lation of a thermoelectrical problem in cylindrical induction heating furnaces,
[2] H.H.J.M. Janssen, E.J.W. ter Maten, D. van Houwelingen, Simulation of coupled Adv. Comput. Math. 26 (2007) 39e62.
electromagnetic and heat dissipation problems, IEEE Trans. Magn. 30 (1994) [20] J.Y. Jang, Y.W. Chiu, Numerical and experimental thermal analysis for
3331e3334. a metallic hollow cylinder subjected to step-wise electro-magnetic induction
[3] A.C. Metaxas, Foundations of Elctroheat: a Unified Approach, John Wiley heating, Appl. Therm. Eng. 27 (2007) 1883e1894.
& Sons Ltd., Chichester, 1996. [21] C. Chaboudez, S. Clain, R. Glardon, D. Mari, J. Rappaz, M. Swierkosz, Numerical
[4] J.D. Lavers, Numerical solution methods for electroheat problems, IEEE Trans. modeling in induction heating for axisymmetric geometries, IEEE Trans.
Magn. 19 (1983) 2566e2572. Magn. 33 (1997) 739e745.
[5] E.J. Davies, P.G. Simpson, Induction Heating Handbook, first ed., McGraw-Hill, [22] E.J. Davies, Conduction and Induction Heating, first ed., Peter Peregrinus Ltd.,
London, 1990. London, 1990.