You are on page 1of 7

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301

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

3D Simulation of magnetic eld distribution in


electromagnetic forming systems with eld-shaper

M.A. Bahmani, K. Niayesh , A. Karimi


School of Electrical and Computer Engineering, University of Tehran, Tehran, Iran

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

Article history: Impulse electromagnetic forming (IEMF) is an effective and powerful technique widely used
Received 13 March 2007 for joining and shaping metals and eld-shaper is a main part of the electromagnetic form-
Received in revised form 6 May 2008 ing which has important effect on the distribution of magnetic eld. In this technique, a
Accepted 11 May 2008 metal work-piece is pushed to a die and formed by a pressure created using an intensive,
transient magnetic eld. This magnetic eld is produced by passing a pulse of electric cur-
rent through a forming coil in a pulsed power circuit. The produced transient magnetic eld
Keywords: induces eddy currents in the surface of work-piece. Induced eddy currents in work-piece
Field-shaper produce a magnetic eld with reverse direction of initial magnetic eld; this results in a
Electromagnetic forming mutual repulsion between coil and work-piece and in this way the work-piece is thrown
Magnetic pressure toward the die. In this process created magnetic forces applied to work-piece are much like
Numerical simulation uniform, but in real applications, some regions of a work-piece have to be more deformed
IEMF and therefore a much greater pressure has to be applied to these regions. The task of concen-
tration of magnetic forces to some desired regions can be accomplished using eld-shapers.
Yu et al. [Yu, H., Li, C., Zhao, Z., Li, Z., 2005. Effects of eld-shaper on magnetic pressure in
electromagnetic forming. J. Mater. Process. Technol. 168, 245249] have recently shown the
effect of eld-shaper on the distribution of the magnetic elds in electromagnetic forming,
but because of the nature of 2D simulations some edge effects in real geometries could not
be taken into consideration. In this paper, a 3D simulation using the FEA software MAXWELL
has been applied to study the magnetic eld distribution during an impulse electromagnetic
forming process. Comparison of the 3D and 2D simulation results indicates that the maxi-
mum magnetic elds achieved in front of nodules of the eld-shaper are about 15% stronger
than those expected by 2D simulations.
By changing the geometry of the eld-shaper, the inuence of the shape of the eld-shaper
on the distribution of the applied forces on the work-piece has been studied. Based on these
simulations, some simple guidelines to design the eld-shaper have been derived.
2008 Elsevier B.V. All rights reserved.

1. Introduction failure in low ductility materials due to strain rate and inertial
stabilization of material failure modes. Impulse electromag-
High-rate forming processes such as impulse electromagnetic netic forming is a powerful and effective high-rate forming
forming (IEMF) can promote signicant increases in strain to technique with a number of advantages such as high clean-


Corresponding author at: School of Electrical and Computer Engineering, University of Tehran, North Karegar Avenue, P.O. Box:
14395/515, IR-14395 Tehran, Iran. Tel.: +98 21 88011247; fax: +98 21 88027766.
E-mail addresses: m.bahmani@ece.ut.ac.ir (M.A. Bahmani), kniayesh@ut.ac.ir (K. Niayesh).
0924-0136/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2008.05.024
2296 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301

ness, cost-efciency, productivity. Applying this technique,


the formability of some materials such as aluminum could be
improved substantially. So the most critical limitation of the
application of aluminum in a wide range of industrial appli-
cations like automotive industries; i.e. the low formability of
this material using conventional forming methods with some
undesirable side-effects like wrinkling, tearing and springing
back as demonstrated (Oliveira, 2002; Oliveira and Warwick,
2003), could be solved.
In this technique, a metal work-piece is pushed to a die
and formed by a pressure created using an intensive, transient
magnetic eld. This magnetic eld is produced by passing a
pulse of electric current through a forming coil in a pulsed
power circuit. The produced transient magnetic eld induces
eddy currents in the surface of work-piece. Induced eddy cur-
rents in work-piece produce a magnetic eld with reverse
direction of initial magnetic eld cause to a mutual repul-
sion between coil and work-piece result to throwing the
work-piece toward the die. In this process created magnetic
forces applied to work-piece are much like uniform, but in
real applications, some regions of a work-piece have to be
more deformed and therefore a much greater pressure has
to be applied to these regions. A main part of an IEMF sys-
tem is eld-shaper that concentrates the magnetic elds in
desired points of a metal work-piece during the forming pro-
cess. A eld-shaper transmits the energy produced by inductor
system to the expected points. It must be noted that the ef-
ciency of the system with eld-shaper is reduced because
of energy dissipation occurred in the eld-shaper, but using
this additional part it is possible to shape the force distri- Fig. 1 An IEMF system applied with a eld-shaper
butions in a much more economical way than using special (compression) for (a) at model and (b) cylindrical
coil systems for different applications (Batygin and Daehn, symmetry model.
1999). Although the eld-shaper plays an important role in
impulse electromagnetic forming systems, there is only very
few studies on its inuences on the magnetic eld distribu- 1. Capacitor bank for storing the electrical energy.
tion. 2. Fast switch for connecting capacitor banks to work coil and
In this paper, 3D simulations by FEA Software MAXWELL interrupting it.
are used to calculate the magnetic force distribution applied 3. Work coil for creating magnetic eld in the work zone.
on the work-piece during the impulse electromagnetic form- 4. Work-piece.
ing process. The results indicate a 15% increase in comparison 5. Die or matrix.
to the 2D simulations in the maximum magnetic eld using
eld-shapers. Comparison of the magnetic eld distributions It can be seen that an IEMF system consists of an
with different geometries of the eld-shapers is used to derive electrical pulsed power circuit responsible for generation
simple guidelines for designing the electromagnetic forming of the pulsed current owing through the work coil, and
systems. some parts with pure mechanical functions like matrix,
which is used to determine the desired shape of the
nal formed work-piece. Although eld-shapers seem to
2. Basic working principle be also a mechanical part, they play an important role in
linking the electrical pulsed power system to the mechan-
An impulse electromagnetic system is used in many different ical parts by modifying the magnetic eld distribution
applications such as compression or expansion of metal tubes, generated on the work-piece. This specic role of the eld-
forming of at sheets; e.g. panels in automotive industries, shapes is described in detail in the next parts of this
welding and many other applications, however the main prin- paper.
ciple in all of the applications of this technique is the same.
Depending on the geometry of the work-piece to be modied, 2.1. Physical description of an IEMF process
the geometries of the coils and other parts of the impulse elec-
tromagnetic forming system could be different in shape. In In this process, a pulsed current with signicant amplitude (up
Fig. 1, two different geometries of such systems adapted to to some hundreds of kilo amperes) and high frequency (some
the work-pieces with cylindrical symmetry or at sheets are tens of kilo Hertz) is produced by discharging an energized
sketched schematically. capacitor bank through a coil by closing a switch. This current
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301 2297

the coil to the lower ones in adjacent to the matrix. In the


other words, they intend to pass through the work-piece to the
cavity located between the matrix and work-piece and push
the work-piece toward the region with the lower density of
energy. During this process, the EM waves penetrate into the
work-piece (skin effect). The effective region with consider-
able current ow is limited to the skin depth given in Eq. (3),
where ,  and f are the conductivity and the permeability of
the work-piece and the frequency of current, respectively.


1
= (3)
f

It must be noted that the effective pressure applied on the


surface of the work-piece is higher in case of smaller skin
depths. This magnetic pressure can be expressed as:

0 2
p= H (4)
2

where H is the intensity of magnetic eld on the surface of


Fig. 2 3D geometry of an IEMF system. work-piece. The magnetic eld distribution needed to calcu-
late the applied pressure on the work-piece is simulated by
solving the Eq. (5) with nite element software Maxwell. Tak-
produces an intensive and transient magnetic eld around the ing into consideration that the divergence of the magnetic ux
coil. According to the faraday law in Eq. (1), when a metal piece density is zero, and introducing the magnetic vector potential,
is located in exposure to the magnetic eld, a so-called eddy Eq. (5) is solved with nite element software Maxwell.
current is induced at the surface of this metallic part (work-
piece). 1
2 A = Jtot (5)
0 r
B
E= (1)
t In this equation, A is magnetic vector potential dened as
 = A,
B  where B is the magnetic ux density. Total current
The direction of the induced current is in such a way that
(Jtot ) in the right side of Eq. (5) contains displacement currents,
its magnetic eld opposes the initial magnetic eld produced
eddy currents, conduction currents and convective currents.
by the current owing through the work-coil; therefore two
In the problem considered in this paper, the displacement and
adjacent conductors (xed coil and free work-piece) with cur-
convective currents are negligible because the frequency of the
rents in opposite directions repel each other. As a result of
applied current is not so high that the product of frequency by
the applied pressure on the work-piece, it is thrown to the
permittivity is still much lower than the conductivity and the
die (matrix). In the majority of applications of IEMF technique,
induced voltages are relatively small, therefore only conduc-
the magnetic pressure on work-piece is not exerted directly by
tion and eddy currents are taken into account.
the coil and an additional metallic part (eld-shaper) is used
Inserting a eld-shaper between the work-coil and the
in between as demonstrated (Batygin and Daehn, 1999; Davis
work-piece, does not change the mathematical formulation
and Semiatin, 1993). In this case, for acquiring the equations
of the problem. It must only be taken into consideration that
for the exerted pressure on work-piece surface and analyz-
the eld-shaper is another metallic body and therefore some
ing it, we consider an IEMF system with cylindrical structure,
eddy currents are induced on its surfaces as well. The interac-
shown in Fig. 2.
tion of the induced currents on the surface of the eld-shaper
Fig. 2 shows the 3D geometry of an IEMF system with
with those induced on the surface of the work-piece results
eld-shaper and a 3D coil used to study the magnetic eld dis-
in a change in distribution of applied pressure exerted on the
tribution. This geometry can be applied to join to cylindrical
work-piece. This can be used to modify the eld distribution
tubes.
on the work-piece; e.g. to increase the applied pressure in the
According to the Eq. (1), time variations of the magnetic
desired region.
eld produce electric eld. In a metallic body with given elec-
trical conductivity, this leads to current ow as:
3. Numerical simulation
Jeddy = E (2)
To be able to simulate the magnetic eld distributions, it is
Electromagnetic elds generated by eddy currents owing assumed that the pulsed current owing through the work-
on the surface of the work-piece exert a force on the sur- coil of the geometry shown in Fig. 2 is given as:
face of work-piece. Electromagnetic elds intend to diffuse
from the regions with higher density of energy in vicinity of I(t) = 50, 000 e5000t sin(60, 000t)
2298 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301

Fig. 4 Displacement and velocity of work-piece during the


electromagnetic forming process.

Fig. 3 Mesh operation in a cross-section. is solved only within the work-piece, the radial coordinate is
always nonzero in the considered conguration and the Eq. (6)
is well dened.
This is the response of a lossy LC resonance circuit. As it will Fig. 4 shows typical displacement and velocity increase of
be seen in the following sections, the magnetic eld distribu- the work-piece during the electromagnetic forming process.
tion is proportional to the current amplitude, and therefore it Due to the displacement and deformation of the work-piece,
is possible to use the results of these simulations to predict the the air gaps in the equivalent magnetic circuit of the congu-
magnetic eld distributions with different current amplitudes. ration considered here increase with time. This leads to lower
Moreover, it is assumed that the work-coil is made of cop- magnetic ux densities compared with the case where the
per, the eld-shaper of copperberyllium and the work-piece displacement of the work-piece is not considered.
of aluminum. The gas lling the region between solids is air.
3.1.1. Boundary conditions
3.1. Model and boundary conditions In this paper, two types of boundary conditions are used:

In Fig. 3, a cross-section of a typical mesh generated during the 1. Natural boundaries: they are assigned to the surfaces
simulation is shown. This consists of 1037 elements for coil, between objects. The natural boundary condition reects
36,289 elements for eld-shaper, 11,651 elements for work- the particular local form of Maxwells equations at the
piece and 61,208 elements for air region. In solids, tetrahedral interface between objects where there is a sudden change
cells and in air region, hexahedral cells have been used as in the material properties and the normal component of B
elements. The hexahedral grid system has been connected is continuous (Manual of the FEA Software Maxwell, 2006).
through Pyramid cells to the tetrahedral grid system at inter- 2. Neumann boundaries: they are assigned to the outside
faces. The quadrilateral cell face of the hexahedral cell system edges of the problem region. The homogeneous Neumann
forms the base of ve sided pyramid cell. The four triangular boundary condition species that the normal component
sides of the pyramid cell are used as boundaries for the tetra- of the H eld is zero (Manual of the FEA Software Maxwell,
hedral grid system, so the interface between the hexahedral 2006).
and tetrahedral grid system is a true interface and the arbitrary
interface feature is not needed. 3.2. Simulation results
During the electromagnetic forming process, the work-
piece is deformed and accelerated towards the die. The change Fig. 5 shows the magnetic ux density distribution in the
in the geometry resulted during the mechanical deformation whole region without the eld-shaper. In Fig. 6, for the same
of the work-piece, has an impact on the magnetic ux den- geometry of the work-coil and work-piece the magnetic eld
sity distribution. To be able to take this phenomenon into distribution with eld-shaper is shown.
account, the deformation and displacement of the work-piece As it can be seen, the magnetic ux density near the nod-
is simulated using the equivalent surface forces method as ule of eld-shaper has much greater values compared to the
demonstrated (Motoasca, 2003). For this purpose, the following situation without eld-shaper. To be able to have a quantita-
equation has to be solved: tive comparison between these two situations, the magnetic
ux density along the Z-axis near the work-piece is shown
1 1 1 in Fig. 7. The results indicate that the magnetic ux densities
r2 ur + r ur 2 ur 2 r2 ur = 0 (6)
r r CP in the desired region (here on the nodules) are increased by
4050%. So, according to Eq. (4), the magnetic pressure with
where r is the radial coordinate, ur is the radial displacement of eld-shaper is more than two times stronger than the situa-
the work-piece and CP is the compressional or p-wave speed tion without eld-shaper. In other regions, the magnetic ux
that is typically between 1000 and 14,000 m/s. As the Eq. (6) density is decreased inserting the eld-shaper.
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301 2299

Fig. 5 Magnetic ux density in a cross-section near the


work-piece without eld-shaper.
Fig. 8 Magnetic ux density distribution of the IEMF
system of Fig. 2 with eld-shaper in a 2D simulation.

Fig. 9 The magnetic ux density in a line along Z-axis in


air gap near work-piece with eld-shaper in 2D and 3D
Fig. 6 Magnetic ux density around the work-piece with simulations.
eld-shaper.
If the edge effects of the 3D geometry of Fig. 2 are neglected,
this geometry can be simplied to a 2D geometry. Fig. 8 shows
the results of the 2D simulations.
It can be seen that the magnetic ux density has higher
values on the nodule, this is in accordance to the previously
shown results of the 3D simulations, but the comparison
between the amplitude of the magnetic ux densities simu-
lated using 2D and 3D geometries indicates that the value of
B in 3D model is about 0.45 Tesla greater than the value of B
in 2D model (see Fig. 9). So, neglecting the edge effects and
reducing the real 3D geometry to the simplied 2D geometry
would result in an error of about 15% in the amplitude of the
magnetic ux density on the nodule, the simulation results
are in good agreement with the experimental measurements
that Yu et al. (2005) have recently shown. This corresponds
to a 30% lower magnetic pressure in this region. It must be
taken into consideration that the reduction of the length of
the coil would increase the edge effects and therefore some
higher errors would be resulted.
Because of the signicant increase in magnetic ux density
Fig. 7 The magnetic ux density in a line along Z-axis in and magnetic pressure in an IEMF system with eld-shaper,
air gap near the work-piece with and without eld-shaper. in designing a eld-shaper, it has to be considered that the
2300 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301

Fig. 11 The magnetic ux density in a line along Z-axis in


air gap between eld-shaper and work-piece in three
conditions.

Fig. 10 Geometry of an IEMF system with eld-shaper


with two nodules. factor, multiplied by the amplitude of the pulsed current (I)
and a constant (K) determined by other geometrical parame-
ters, material properties and characteristics of the electrical
eld-shaper has the enough mechanical strength to stand circuitry.
the exerted magnetic pressure. Its mechanical strength is a
 = KIf (Aeffective )
B (7)
function of its geometry and material. For this purpose, the
eld-shaper is normally made of materials with high mechan-
ical strength like copper-beryllium. Fig. 13 shows the enhancement factor of the magnetic ux
density (f) as a function of the ratio of width of nodule to the
3.2.1. Other shapes of eld-shaper whole length of the work-coil for different radii of the IEMF
In this part, simulations with different geometries of the system. As it can be seen, the enhancement factor is almost
eld-shapers are carried out. Comparing the results, an independent of the radius of the system.
approximate guideline to design the eld-shaper is derived. These two simulations show that with expansion of the
Two different parameters of the eld-shaper of Fig. 2 are area of the nodules in eld-shaper, the maximum value of
varied; these are the effective area of the nodule of the eld- B decreases, so to calculate the appropriate parameters for
shaper and the number of nodules. In Fig. 7 width of nodule the electromagnetic forming system, it would be enough to
was 15 mm, but second type of eld-shaper that is shown in carry out one simulation to determine the proportionality fac-
Fig. 10 has two nodules and width of each nodule is 10 mm tor in Eq. (7) and according to the relation given in Fig. 13, it
and other dimensions are similar to Fig. 7. We expect that the
maximum value of magnetic eld decrease with the increase
of the effective area of nodule.
The boundary conditions are similar to previous simula-
tion, but the quality of mesh operation is more complicated
than Fig. 3. This geometry consists of 1086 elements for coil,
43,265 elements for eld-shaper, 12,439 elements for work-
piece and 65,727 elements for air region.
In Fig. 11, the comparison between the values of magnetic
ux densities along Z-direction near work-piece is shown for
three geometries, with one nodule eld-shaper, with two
nodules eld-shaper, and without eld-shaper.
Fig. 12 shows the magnetic ux density along the Z-
direction near the work-piece for three different eld-shapers
with different widths of 10 mm, 15 mm and 20 mm. It can be
seen that the magnetic ux density in the region on the nodule
decreases as the effective area of the nodule increases.
Based on the simulation results shown in Figs. 11 and 12,
the magnetic ux density on the nodule (B) can be expressed Fig. 12 The magnetic ux density in a line along Z-axis
as a function of the effective area (Aeffec ); called enhancement near work-piece with three types of eld-shaper.
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 22952301 2301

the magnetic ux density plays a central role. This paper intro-


duces a 3D simulation of the magnetic ux densities during
the impulse electromagnetic forming process with emphasis
on the inuence of the eld-shaper on its distribution. The
comparison of the 2D and 3D simulation results indicates an
about 15% increase in the magnetic ux densities. This dis-
crepancy can be caused by the neglected edge effects in 2D
simulations.
Different geometries of the eld-shapers have been stud-
ied, and based on the simulation results a simple guideline
to estimate the maximum ux density on the region on the
nodule of the eld-shaper is derived.

references
Fig. 13 Enhancement factor of B for various eld-shapers
with three different radius of IEMF system.
Batygin, Y.V., Daehn, G.S., 1999. The Pulse Magnetic Fields for
Progressive Technologies. Kharkov, Columbus.
would be possible to determine the necessary current ampli- Davis, J.R., Semiatin, S.L., 1993, ASM Metal Forming Handbook, p.
tude to achieve the desired magnetic pressure at the desired 14.
region. Yu, H., Li, C., Zhao, Z., Li, Z., 2005. Effects of eld-shaper on
magnetic pressure in electromagnetic forming. J. Mater.
Process. Technol. 168, 245249.
4. Conclusions Motoasca, T.E., 2003. Electrodynamics in Deformable Solids for
Electromagnetic Forming, Ridderprint Offsetdrukkerij B.V.
Impulse electromagnetic forming (IEMF) is a powerful and Manual of the FEA Software Maxwell, V. 11.1, 2006.
Oliveira, D.A., 2002. Electromagnetic forming of aluminium alloy
effective high-rate forming technique with a number of advan-
sheet experiment and model. Masters Thesis, Department of
tages such as high cleanness, cost-efciency, productivity. To
Mechanical Engineering University of Waterloo, Ontario,
be able to apply the desired pressure to different regions of Canada.
the work-piece to achieve the ideal forming, in this technique Oliveira, D.A., Warwick, M.J., 2003. Electromagnetic forming of
a so-called eld-shaper is used to transfer the stored electrical aluminium alloy sheet. J. Phys. 110, 293298.
energy to the desired points. In this process, the distribution of