Journal of Materials Processing Technology 103 (2000) 14 22

State-of-the-art of simulation of sheet metal forming

A. Erman Tekkaya
Department of Mechanical Engineering, Middle East Technical University, 06531 Ankara, Turkey

Abstract

The paper starts with a brief historical overview to the attempts of numerical simulation of sheet metal forming. Comparison between
bulk and sheet metal forming processes from the simulation point of view is given. Basic requirements of the applier to the simulation tools
are summarized. Various possible methodologies are brie¯y discussed. Special emphasis is given to the static explicit and dynamic implicit
®nite element procedures. Also different element types are overviewed. Available important commercial ®nite element packages are given.
The current state of the application of the simulation tools is discussed. Some typical industrial applications are reviewed to demonstrate the
current abilities of analysis. Finally, an attempt to prognosticate some future developments is made. # 2000 Elsevier Science S.A. All
rights reserved.

Keywords: Forming; Sheet metal; Numerical simulation

1. Introduction forming process is given in Fig. 1 [25]. Here, a comparison

with experimental results shows quite good agreement up to
As a forming expert in one of the leading car manufactur- the increment of failure due to fracture of the sheet. These
ing companies, Schacher [22] summarized the current pres- methods are, however, restricted to axisymmetrical pro-
sure on this industry as: ``In the past we introduced 3 new blems. Although there were attempts in the 1990s to apply
models every 10 years, now we introduce 10 new models this method to general 3D deep-drawing problems replacing
every 3 years''. This drastic reduction of development the ®nite element method [19], they could not establish
periods as well as the trend to reduce weight of the cars themselves due to the serious drawback of not applying
in order to reduce the fuel consumption leads especially in boundary conditions in a general manner as it could be done
the car manufacturing industry to a rebuilding of the con- elegantly in the ®nite element method.The real break-
ventional design and manufacturing procedures. Sheet metal through of the numerical approximation of sheet metal
forming as an important production process (see for instance processes was possible through the application of the ®nite
[14]) is heavily experience based and involves trial-and- element method. Pioneering studies are made by Gotoh and
error loops. These loops are repeated the more the less the Ishise [8] and Wang and Budiansky [27] and Wi® [30]. The
experience on the part geometry and material is. In the ®rst two studies were utilizing an elasto-plastic material law,
innovative process design procedure, however, the trial-and- whereas the latter one used a rigid plastic law. Furthermore,
error loops are reduced by means of modern numerical the element types were membrane [27] and continuum,
approximation analysis, which is named ironically also as respectively. The ®rst 3D applications are known by Tang
virtual production. The aim of this paper is to discuss the et al. [24] and Toh and Kobayashi [26]. All the mentioned
current state of numerical simulations in sheet metal forming studies are of static implicit, or respectively, static explicit
industry and possible future developments. type. The dynamic explicit methods have their roots in the
Attempts of numerical approximate solution of sheet study of Belytschko and Mullen [3]. Application to defor-
metal forming root back to the 1960s. The very ®rst numer- mation mechanics is given in the work of Benson and
ical solutions of sheet forming processes have been obtained Hallquist [4]. The replacement of draw beads by arti®cial
by ®nite difference methods [31]. A typical example utiliz- forces is, for instance, used by Massoni et al. [15]. The
ing the ®nite difference method for the solution of the stretch incorporation of viscous effects has been ®rstly considered
by Wang and Wenner [28]. Thermomechanically coupled
E mail address: [email protected] (A.E. Tekkaya) modeling has been performed by Xing and Makinouchi [32].

0924 0136/00/$ see front matter # 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 9 2 4 0 1 3 6 ( 0 0 ) 0 0 4 1 3 1
 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22 15

bulk metal forming. This requires special precautions in the

analysis. Process-wise, spring-back and residual stresses are
rather critical issues. These phenomena are due to elastic
unloading after elasto-plastic deformation. Therefore, an
extremely accurate computation in the elasto-plastic part
of the deformation is required since the equilibrium state
after unloading depends on the differences in the plastic
strains which are in the order of elastic strains. All these facts
lead to the later start of the industrial application of sheet
metal simulation. On the other hand, today, sheet metal
forming simulation is about 5 years ahead of the industrial
3D bulk metal forming simulation.
The requirements and expectations of the industry from
Fig. 1. Thickness strains in stretch forming as computed by a rigid plastic the simulation techniques will be discussed in Section 2.
®nite difference method [25].
Various different approaches of the ®nite element simulation
in sheet metal forming will be covered in Section 3. An
Developments in the sheet metal forming simulation by overview to the capabilities of the currently available com-
the ®nite element method started roughly 5 6 years later mercial simulation tools is given in Section 4. Finally,
than bulk forming simulation. This is due to the basic expected future developments in the simulation tools will
differences between these processes as depicted in Table be prognosticated before drawing some conclusions in
1. Geometrically, in sheet metal forming the workpiece has a Section 5.
larger surface to volume ratio than in bulk metal forming.
This is the basic reason for utilizing membrane or shell
elements in ®nite element simulation. Furthermore, this 2. Industrial requirements to simulation of sheet metal
allows also the ef®cient utilization of explicit methods forming
(see also Section 3). Material behavior in sheet metal
forming is characterized through an initial anisotropy due The industrial requirements to the numerical simulation
to the manufacturing process (usually rolling) of the blanks. of sheet metal forming processes are highly expressed by
Compared to bulk metal forming, in sheet metal forming Budiansky's dream (Harvard) of the mid-1970s: ``I imagined
pure strains are of moderate magnitudes (say up to 0.7) but a black box a black computation box that was incredibly
displacements can be even larger than in bulk metal forming. powerful and into which we could feed a mathematical
This leads to the fact that in sheet metal forming remeshing description of what the stylists envisioned for a certain sheet
due to excessive element distortions is not the case as in bulk metal shape. Then push a button and the computer spits out
metal forming. Instead, remeshing is necessary due to the the die shape, the blank con®guration needed, the draw
change of the deformation zone in time. The deformation beads and their orientation and con®guration. If it is not
kinematics in sheet metal forming is exhibiting also macro- possible, it tells us that too!''
scopic instabilities leading to necking and wrinkling which Although not openly admitted, this is the general real-life
complicate the numerical treatment substantially. Also, fail- expectation of any sheet metal forming engineer from the
ures are on behalf of the workpiece rather than the tools as in ®nite element simulation of sheet metal forming processes.

Table 1
Comparison of sheet and bulk metal forming features (S: surface, V: volume, t: thickness, a: characteristic dimension)

Sheet metal forming Bulk metal forming

Workpiece geometry Plane structured S/V / 1/t, t<1 resp. t1 Balanced S/V / 1/a, a1
Material behavior Normal (planar) anisotropy Bauschinger effect
Deformation kinematics (Very) large displacements but moderate strains Large displacements and (very) large strain
Instabilities (necking, wrinkling, etc.) Almost no instability mode
Process features Cold Cold/warm/hot
Workpiece failure Die (and workpiece) failure
Spring back critical Spring back uncritical
Residual stresses critical Residual stresses critical
Finite element simulation Always elasto plastic Rigid/visco rigid/elasto plastic
Shell (membrane) elements Continuum elements
Isothermal Thermomechanically coupled
Basically explicit Implicit
Remeshing due to inhomogeneous deformation and Remeshing due to large elemental distortions
geometric description accuracy
16 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22

Currently, the industrial goals of the forming simulation pressure to have response times less than 2 h is
can be summarized in three main groups. increasing because of the aim to optimize the process.
5. Simulation output. The industrial engineer want to
1. Time reduction: evaluate the following results computed numerically:
 Early checking of producibility of workpieces,  Strain and thickness distribution in the sheet,
 Reduction of the development times,  Time tracing of certain material zones such as the
 Reduction of the try-out-times, welding seams in tailored blanks,
 Quick response to modification wishes.  Failure indicators for wrinkling, fracture, loose metal
2. Cost reduction: zones, low spots, rabbit ears, orange skin, flow figures,
 Cheaper products, etc. (also using for instance forming limit diagrams),
 Reduction of die costs,  Post-failure data such as number and amplitude of
 Press down sizing, wrinkles,
 Increase of reliability.  Optimum shape of the blank,
3. Increase of product quality:  Location and position of the draw beads or draw stops,
 Optimal selection of the workpiece material,  Spring-back value (especially after trimming and flan-
 Production of more complicated parts, ging operations),
 Know-how accumulation for new materials,  Residual stresses in the finished product,
 Process repeatability,  Stiffness and/or pressure of the blank-holder (position
 Optimization by variants. dependent),
 Lubrication conditions,
The requirements corresponding to these goals are  Punch loads.
diverse. 6. Material behavior. Initial anisotropy due to the rolling
process of the sheet is a standard feature of the material
1. Processes and operations. It is expected that the model required for sheet metal simulation. Recently, due
simulation tool is able to model various processes and to the increase in the desired accuracy, induced
operations of sheet metal forming one by one or even in anisotropy in form of kinematic hardening models is
combination: required. Furthermore, nonquadratic ¯ow loci for
 Binder closure, aluminum based materials is looked for.
 Deep-drawing process (with draw beads, etc.), 7. Accuracy. Generally the accuracy requirements have
 Stretch forming processes, increased with the usage of the simulation tool.
 Trimming operation, Specially high accuracy is expected for failure determi-
 Flanging operation, nation (better than 20%) and spring-back. These require
 Restrike operations, an increased precision in material and friction modeling,
 Hydroforming process. but also in numerical solution techniques. It is interest-
2. User-friendliness. This is a critical issue for the ing to notice that even the Young's modulus change with
application of the simulation tool in industry. The straining, which is in the order of roughly 3% for each
industry expects that the usage of the code should be 10% strain for aluminum alloys for instance, is being
such simple, that there is no need for an extra ®nite considered seriously by some industrial companies in
element expert. Furthermore, the simulation tool should the numerical approximation.
be available there where it is needed, that is, it must be
usable in the design of®ce and not only in the In the next section, the available methodologies for
computational department of the company. numerical simulation of sheet metal forming processes will
3. CAD FEM CAD interface. The ideal requirement is be discussed.
that the CAD model of the tools can be easily
transferred to the ®nite element code and that any
geometrical modi®cations performed during the 3. Various approaches of approximate analysis
®nite element simulation process can be transferred
back to the CAD model and from there to the CAM In this section, only the various ®nite element methodol-
system. ogies will be brie¯y reviewed and discussed. Numerical
4. Ef®ciency of simulation. Depending on the design and approximation methods based on the ®nite difference and
development stage at which the simulation tool is used upper bound technique as well as the slip-line ®eld solution
different computational times are desired. Whereas techniques will remain hence untouched. The next section
response times less than 1 h are required in the discusses brie¯y the different element types used in sheet
preliminary design stages, over-night responses are forming modeling. Then, the basic approaches such as static
tolerated in the main design stage or problem solving implicit, dynamic explicit and inverse (one-step) methods
stages. However, even in the main design stages the will be summarized.
 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22 17

Table 2
Different element types used in sheet metal simulations

Element type Membrane stiffness Bending stiffness Transverse shear stiffness Transverse strain variation

Kirchhoff Mindlin

Membrane Yes No No No No
Plate No Yes No Yes No
Shell (thin) Yes Yes No Yes No
Shell (thick) Yes Yes No Yes (Yes)
Continuum Yes Yes Yes Yes Yes

3.1. Element types This equation can be divided by a virtual time yielding a rate
form. The corresponding constitutive equation is then also of
The high surface to volume ratio of the workpiece in sheet rate type,
metal forming gives the opportunity of using structural 
elements which make use of the plane shape of the sheet. T ij ˆ Cijkl Dkl : (3)
Hence, there are four different types of elements available 
Here, T ij is, for instance, the Jaumann derivative of the
for the analysis of sheet metal forming processes (see also
Cauchy stress, Cijkl the fourth-order constitutive tensor and
Table 2). In sheet metal forming processes for which bend-
Dkl the rate of deformation tensor. Discretization of Eq. (2)
ing occurs over a bending radius which is larger than roughly
yields a stiffness equation of the type,
ten times the sheet thickness [20], membrane elements can
be used for modeling. Generally, however, if deep-drawing ‰K…u†Šfug ˆ fFg: (4)
type of deformation dominates, experience has shown that
This equation is nonlinear in the displacements and can be
membrane elements are not appropriate for modeling.
solved with one of the standard numerical methods [20]. An
Therefore, the most common element used for deep-drawing
Euler type solution of this system of equations leads to the
dominated forming processes is the shell element. These
so-called static explicit methods [29].
elements can be divided into two groups: shell elements
Hence, in the static implicit methods, which were the very
which can (thick) and cannot (thin) consider a transverse
®rst methods used in simulation of metal forming processes,
pressure appropriately. Transverse pressure is essential if
static equilibrium is satis®ed in the unknown ®nal con®g-
blank holder actions on the general deformation behavior
uration of a time increment. This method enables a full static
shall be considered accurately. Continuum elements are
solution of the deformation problem with convergence
seldom used for sheet metal modeling (see for instance
control. Theoretically the increment sizes can be very large;
[9]). On the other hand, in sheet forming processes such
practically, however, it is limited by the contact conditions.
as hydroforming or blanking, continuum elements can be
Computational times increase almost quadratically with
appropriate.
increasing element number. Because of the matrix inversion
step and accurate integration schemes, memory require-
3.2. Quasi-static implicit approach
ments are also high. Another disadvantage of the implicit
methods is said to be the problem of divergence of the
The principle of virtual work after neglecting body forces
solution. This is basically initiated due to the large number
can be written as
of contact nodes which can overload many implicit contact
Z Z
algorithms. Of course, it is questionable whether to receive
Tij dui;j dV ˆ ti dui dA; (1)
V A
the feedback that the solution is wrong is a real drawback or
not. Another disadvantage of the implicit methods is the
where Tij is the Cauchy stress tensor, ui,j the gradient of the singularity of the stiffness matrix at bifurcation points, such
displacements, ti the traction vector and d the variational as instabilities at wrinkling initiation.
operator. The Euler equations of this statement are simply
the static equilibrium equations. In classical sheet metal
forming processes the work done by body (dynamic) 3.3. Dynamic explicit approach
forces is roughly four orders of magnitude less than the
plastic deformation energy. Hence, the natural equations The virtual work equation (1) can be modi®ed by an
of analysis are the static equilibrium equations. Eq. (1) inertia term to
can be linearized about a known con®guration in time Z Z Z
yielding
Tij dui;j dV ˆ ti dui dA ÿ rui dui dV; (5)
Z Z V A V
‰dTij ÿ Tkj dui;k ‡ Tij duk;k Šdui;j dV ˆ ti duj dA: (2)
V A where r is the density and uÈi the acceleration of material
18 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22

particles. Discretization of this equation yields The additional arti®cial body forces are taken over by the
rigid dies, since in deep-drawing most of the workpiece is
‰MŠ ut ˆ fF t g ÿ fI t g: (6)
supported by the dies thanks to the high surface to volume
In this equation [M] is the lumped mass matrix, {F} the ratio. Therefore, the dynamic explicit method can be applied
external force and {I} the internal force vector at a given successfully in sheet metal forming from this point of view.
time t. Now, an explicit central difference scheme for the The most important advantage of the dynamic explicit
time marching [18] is utilized in form of method is its robustness. There is no check of unbalanced
forces and hence no convergency control. Currently, the
ut‡Dt ˆ ut ‡ Dtt‡Dt u_ t‡Dt=2 (7) computational speed is higher and the memory requirement
with is less than the static implicit methods. Furthermore, the
numerical procedure can be easily parallelized resulting
Dtt‡Dt ‡ Dtt t even bigger speed advantages. A by-product of the explicit
u_ t‡Dt=2 ˆ u_ t Dt=2
‡ 
u: (8)
2 scheme is the determination of the wrinkles. The wrinkles
Eq. (6) must be modi®ed by an arti®cial damping term in are initiated through numerical inaccuracies. Usually, the
order to compensate for possible density manipulations in region of the wrinkles is quite accurately determined; but the
the structure to [23], number of wrinkles and the amplitude of the wrinkles may
be inaccurate due to the numerically driven initiation pro-
‰MŠ ut ‡ ‰CŠu_ t ˆ fF t g ÿ fI t g (9) cess. A ®nal advantage is the simpleness of programming.
The central difference algorithm is conditionally stable. On the other hand, there are several disadvantages. The
Therefore, the time step has to be less than explicit character of the numerical scheme is ful®lled if and
only if the mass matrix is lumped, i.e. is diagonal. Further-
L 2 L more, the speed advantages can be hold only if the element
Dt   p (10)
Cd o E=r computations are as few as possible. This is maintained by
using single-quadrature elements, which deliver rather poor
where L is a characteristic length of the element, Cd the
stress and strain accuracy. It is claimed that the error
speed of sound in the workpiece material, o the largest
introduced by a lumped mass matrix is compensated by
eigenvalue of the system and E the Young's modulus of the
the reduced integration schemes of the elements. However,
material. Consider a typical deep-drawing operation for
this leads to the disadvantages that local stresses and spring-
which Cd is 5000 m/s for steel, L is in the order of mm's,
back is not reliably computed. Besides, the absence of
say 1 mm (this corresponds to the smallest element in the
convergence control is a critical issue. Finally, remeshing
mesh!), the punch speed is about 0.5 m/s and the punch
leads a reduction of the time step.
traveling distance is roughly 300 mm. The time for deep-
drawing can found as 0.6 s. The minimum time step is given
3.4. Inverse approach (one-step methods)
by the above equation as 210 7 s. Dividing the total
process time by the minimum time step yields the number
The essence of one-step methods is the simpli®cation of
of time increments as 3106. This is an unacceptable high
the physics of the forming operation. Basic simpli®cations
number of increments, which would make the dynamic
are the assumption of a linear strain path, neglecting the
explicit methods unfeasible. For this reason, two different
history of contact, ignoring friction, etc. Utilizing these
numerical tricks are applied:
assumptions it is possible to perform the computations
1. The punch speed is increased as compared to the real starting from the ®nal deformed con®guration to the initial
process speed. In order to reduce the undesired effects of undeformed con®guration by a single step in a very short
the arti®cial mass forces numerical damping is intro- time (usually one order less than the previously described
duced. The damping matrix is taken proportional to the two other methods) [2]. Therefore, in conjunction with a
mass matrix usually. This precaution, however, does not CAD software and an automatic meshing ability, these
work if the material is strain rate sensitive. Also, if methods may supply valuable information about the feasi-
thermomechanical phenomena is involved, increasing bility of the concept at the very ®rst stages of the design
the process speed is not allowed. In such cases the process. On the other hand, the accuracy of the results is in
second treatment is used. certain cases quite low. The problem is to identify for which
2. Increasing the density of the material leads to a cases the accuracy is in an acceptable range.
reduction of the speed of sound in the material and
hence an increase in the allowable time increment.
Increasing the density by a factor of 100 reduces the 4. Current state
speed of sound by a factor of 10 and so increasing
the time increment by a factor of 10. In this case, Currently a clear tendency to the application of dynamic
however, introducing numerical arti®cial damping is not explicit methods can be observed. This is demonstrated, for
possible. instance, by the increased use of dynamic explicit codes in
 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22 19

deep-drawing steels can be considered as satisfactory,

whereas for new high-strength steels as well as for alumi-
num alloys available models are still unsatisfactory. 3D
deep-drawing processes can be handled with element num-
bers up to 100 000 on single-processor workstations with
computational times of several days. Simulations of more
complex problems which need many hundred thousand of
elements are currently not feasible on single processor
workstations. The user-friendliness of the codes is good
to very good and improves drastically through specialization
of the programmes steadily. The simulations can be per-
Fig. 2. Proportion of the dynamic explicit codes in various benchmark formed for plastic deformation analyzes robustly with
tests [29]. dynamic explicit codes, whereas static implicit codes may
diverge for some problems under certain conditions. Purely
benchmarks, (Fig. 2) [29]. But also one-step codes ®nd an explicit codes are, however, not able to conduct the elastic
increased use in the preliminary design stages. Static impli- unloading analysis reliable; therefore, after a dynamic expli-
cit codes, on the other hand, are used in the spring-back cit deformation analysis a static implicit unloading analysis
computations in combination with the dynamic explicit is performed. The mesh generation from CAD models is
codes, but also in the simulation of certain hydroforming possible. However, the regeneration of CAD models from
processes. In some cases, where the springback and residual ®nite element models is currently not realizable. Remeshing
stress states are important, static implicit codes are even is a critical issue in the simulation of sheet metal forming
used for the complete simulation of the sheet forming processes in several aspects. First of all, the mapping of the
process. CAD representation on a ®nite element mesh (almost exclu-
Table 3 lists some of the important commercial software sively linear quadrilateral elements are used) can be per-
packages used by the industry in the simulation of sheet formed without dif®culties. However, to rebuild the ®nite
metal forming processes (see also [10,21]). In this table, element mesh from an existing mesh data without introdu-
Autoform has a special static implicit formulation in which cing geometric inaccuracies is a dif®cult task. One common
the membrane and bending behaviors of deformation are method to generate new meshes with least amount of
decoupled. This allows a considerable speed-up of the geometrical inaccuracies is to re®ne existing meshes
solution procedure. through incompatible meshes, in other words, meshes which
The capabilities of today's simulation tools can be sum- are obtained through dividing existing elements. This
marized as follows: The material modeling is basically requires additional multi-point-constraints for the incompa-
established on an elasto-plastic material law with initial tible nodes. Multi-point-constraints, however, increase the
anisotropy incorporated by the classical Hill's model [11] stiffness of the solution and may introduce numerical inac-
for steel and the modi®ed Hill's model [12] or Barlat's curacies, which must be evaluated carefully. Recently, also
model [1] for aluminum type materials. The experimental compatible adaptive meshing techniques are offered by
determination of the respective yield loci is a critical issue some software vendors. Multistep processes, consisting of
and is described in [16]. Induced anisotropy, i.e. the aniso- binder closure, deep-drawing, trimming, ¯anging, etc. can
tropy which develops in the course of deformation, can be be also handled more or less. Element removal for cutting
handled macroscopically with kinematic hardening models operations is a standard utility of all commercial codes.
such as handled in [13] or with microstructural models such Draw beads are handled by pre-compiled models (see for
as given by [6] or [17]. Constitutive modeling for classical instance [5]). The analysis of tailored blanks is conveniently
possible only in some codes. The output of a ®nite element
Table 3
Various commercial code for sheet metal simulations
analysis includes suf®ciently accurate strain and thickness
distributions, drawing forces and wrinkles in free zones.
Software Origin and company Code type Less accurate outputs are obtained for tearing type and other
ABAQUS Standard HKS, USA Static implicit surface related failure predictions as well as to wrinkling
MARC MSC, USA under contact conditions.
NIKE3D LSTC, USA To demonstrate the effectiveness of the available com-
Autoform, Switzerland Static implicit (special)
AUTOFORM
mercial simulation codes two examples of application will
ABAQUS Explicit HKS, USA Dynamic explicit
DYNA3D LSTC, USA be given. The ®rst one is the simulation of the ®rst stage of
PAM STAMP ESI, France the deep-drawing of a wheel cover. The simulation has been
OPTRIS Dynamic Software, France performed with the dynamic explicit code OPTRIS by Roll
AUTOFORM one step Autoform, Switzerland Inverse methods [21]. Fig. 3 shows the thickness distribution at the end of the
Sollac, France
ISO PUNCH
drawing operation. The evolution of the wrinkles in the free
SIMEX2 SimTech, France
zones can be seen clearly.
20 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22

Fig. 3. Simulation of a wheel cover with OPTRIS [21] (Courtesy of Daimler Benz, Stuttgart).

Another example of application is given in Fig. 4. Here, believes that in the long-term static-implicit methods will
simulation of the deep-drawing of an S-Rail, which has been get the standard tool if and only if, of course, computa-
introduced as a benchmark on the NUMISHEET'96 con- tional speed and robustness can be increased.
ference will be described. In the analysis an elasto-plastic  Increasing analysis capabilities. With the increasing
material law with initial planar anisotropy has been utilized. utilization of the simulation by the die makers, the
The computations has been conducted with the static impli- demand for more complex analysis is also increasing.
cit code MARC. Totally 3000 shell elements with 18 666 Hence, it is expected that in the near future not only the
degrees of freedom has been used. The solution has been forming process with rigid dies will be analyzed, but in
obtained with 80 increments in 12 h and 26 min on an SGI addition, also the deformation of the dies and even the
R8000 Power Challenge with four processors. This time forming presses will be included. In this way, also the
includes the spring-back calculation. quality of the wrinkling prediction in contact zones will
be enhanced. These extensions will require that the
existing codes must get faster several folds, which can
5. Expected future developments be achieved apparently only by parallelization of the
codes. Furthermore, an improved bi-directional imbed-
It is expected that developments in the ®eld of ®nite ding of the simulation into CAD environments will be
element simulation of sheet metal forming processes will inevitable.
cover the following topics:  New visualization techniques. The increasing complexity
of the analyzed problems leads to excessive amounts of
 Increasing the analysis accuracy. Here, the development data which requires new post-processing techniques. The
and application of new constitutive equations, failure utilization of virtual reality techniques such as stereo-
criteria and friction models is expected. Especially for scopic viewing is getting a standard post-processing
high-strength steel and aluminum alloys better models practice. By this technique, 3D visualization is enabled
are required. Also the material data needed for such that allows the user to interact with the forming process
sophisticated models must be supplied. Finally, failure dynamically in a virtual environment.
criteria (and respective material data) for tearing, flow  Optimization abilities (synthesis). Today's practice of
figures, orange skin, etc. must be improved for better forming simulations is that they are used as virtual
utilization of the simulation tools. Spring-back and resi- experimentation tools replacing the physical trial-and-
dual stress computations must be improved. The author error experiments. This means however, that only a small
 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22 21

may be still not achieved. This leads to the new concept of

production oriented part design, which requires in the part
design stage the information about the producibility of that
part. With the conventional production technology this early
interaction is very dif®cult not to say impossible. The ®nite
element simulation of production processes, however, can
make this early interaction feasible. The usage of the
production simulation in the part design stage will be the
third phase of sheet metal simulation. In this phase, the
simulation tool may not be used only by ®nite element
laymen but also by a metal forming laymen. This will
require new simulation tool features obviously. The ultimate
®nal phase of sheet metal forming simulation imaginable, is
the utilization of this tool together with other computer
based engineering tools in the of®ce of the stylist of various
products, as pointed already by Budiansky in his dream of
the seventies. Some decades will be necessary to reach this
®nal phase of simulation in the world of virtual reality.

References

[1] F. Barlat, D.J. Lege, J.C. Brenz, A 6 component yield function for
anisotropic materials, Int. J. Plast. 7 (1991) 693 712.
[2] J.L. Batoz, Y.Q. Guo, F. Mercier, The inverse approach including
bending effects for the analysis and design of sheet metal forming
parts, in: NUMIFORM'95, Balkema, Rotterdam, 1995, pp. 661 667.
[3] T. Belytschko, R. Mullen, Explicit integration of structural problems,
in: P. Bergan, et al. (Eds.), Finite Elements in Nonlinear Mechanics,
1977, Tapir, Trondheim, Norway, pp. 672 720.
[4] D.J. Benson, J.O. Hallquist, A simple rigid body algorithm for
Fig. 4. Simulation of the S Rail forming using static implicit codes structural dynamics programs, Int. J. Num. Meth. Eng. 22 (1986)
(Courtesy of MARC Analysis, Palo Alto, CA, USA, 1997). 723 749.
[5] B.D. Carleer, M.F.M. Louwes, J. Huetink, P.T. Vreede, Modelling
drawbeads in 3D ®nite element simulations of the deep drawing
portion of the mathematical capabilities are utilized, process, in: NUMIFORM'95, Balkema, Rotterdam, 1995, pp. 681
since current mathematical methods allow optimization 685.
based on a sensitivity analysis for many parameters. [6] P.R. Dawson, A.J. Beaudoin, Simulating deformation induced texture
Impressive examples are known from bulk metal forming in metal forming, in: NUMIFORM'92, Balkema, Rotterdam, 1992,
pp. 25 33.
praxis [7]. This rigorous approach enables optimization
[7] L. Fourment, T. Balan, L. Chenot, Optimal design for non steady
of the blank geometry, blank-holder pressure, seam posi- state metal forming processes II. Application of shape optimization
tion for tailored blanks, draw bead positions, etc. in a in forging, Int. J. Num. Meth. Engg. 39 (1996) 51 65.
mathematically deterministic way. [8] M. Gotoh, F. Ishise, A ®nite element analysis of rigid plastic
deformation of the ¯ange in a deep drawing process based on a
fourth degree yield function, Int. J. Mech. Sci. 20 (1978) 423 435.
[9] M. Herrmann, Beitrag zur Berechnung von VorgaÈngen der Blechum
6. Conclusions formung mit der Methode der ®niten Elemente, Dr. Ing. Dissertation,
UniversitaÈt Stuttgart, 1990.
The simulation of sheet metal forming processes is not [10] M. Herrmann, Was erwarten Kunden von einem Dienstleister im
any more an academic exercise. If the academic works on Bereich der Umformsimulation. In: Workshop Umformsimulation,
simulation can be regarded as the ®rst phase of sheet metal Haus LaÈmmerbuckel, Wiesensteig, Germany, 28 29 April, 1997.
[11] R. Hill, The Mathematical Theory of Plasticity, Clarendon Press,
forming simulation, now the second phase has started, which Oxford, 1950.
is the industrial application of the simulation techniques by [12] R. Hill, A user friendly theory of orthotropic plasticity in sheet
the die makers and process designers. Now, given a part metals, Int. J. Mech. Sci. 35 (1993) 19 25.
design, the production process is developed and/or improved [13] J. Huetink, A.H. Streppel, P.T. Vreede, Development and experi
by systematic numerical analyzes in the industry. In the mental veri®cation of constitutive equations for anisotropic sheet
metal, in: COMPLAS'95, Pineridge Press, Swansea, 1995, pp. 2271
production stage, unfortunately, there is almost no chance of 2282.
changing the part geometry in case of dif®culties in the [14] K. Lange (Ed.), Lehrbuch der Umformtechnik. Band 3: Blechum
production of that part. Therefore, the real optimum product formung, Springer, Berlin, 1975.
22 A.E. Tekkaya / Journal of Materials Processing Technology 103 (2000) 14 22

[15] E. Massoni, et al., A ®nite element modelling for deep drawing of [24] S.C. Tang, E. Chu, S.K. Samanta, Finite element prediction of the
thin sheet in automotive industry, in: K. Lange (Ed.), Advanced deformed shape of an automotive body panel during preformed stage,
Technology of Plasticity 1987, Vol. II, Berlin, 1987, pp. 719 725, in: NUMIFORM'82, Pineridge Press, Swansea, 1982, pp. 629 640.
ISBN 3 540 17915 1. [25] A.E. Tekkaya, Complete numerical solution of the axisymmetrical
[16] W. MuÈller, K. PoÈhlandt, New experiments for determining yield loci deep drawing problem, M.S. Thesis, Middle East Technical Uni
of sheet metal, J. Mater. Process Technol. 60 (1996) 643 648. versity, Ankara, 1980.
[17] E. Nakamachi, Elastic/Crystalline viscoplastic ®nite element analysis [26] C.H. Toh, S. Kobayashi, Finite element process modelling of sheet
and optimum fabrication design of sheet metal, in: COMPLAS'97, metal forming of general shapes, In: Grundlagen der Umformtechnik
Pineridge Press, Swansea, 1997, pp. 101 118. I, Berlin, 1983, pp. 39 56, ISBN 3 540 13039 X.
[18] N. Rebelo, J.C. Nagtegaal, L.M. Taylor, Comparison of implicit and [27] N. M. Wang, B. Budiansky, Analysis of sheet metal stamping by a
explicit ®nite element methods in the simulation of forming ®nite element method, Trans. ASME. J. Appl. Mech. 45 (1978) 73
processes, in: NUMIFORM'92, Balkema, Rotterdam, 1992, pp. 99 82.
108. [28] N. M. Wang, M.L. Wenner, Elastic viscoplastic analysis of simple
[19] K. Roll, Private Communication, Stuttgart, 1990. stretch forming problems, in: D.P. Koistinen, N. M. Wang (Eds.),
[20] K. Roll, A.E. Tekkaya, Numerische Verfahren der Prozeûsimulation Mechanics of Sheet Metal Forming, 1978, pp. 367 402.
in der Umformtechnik, in: K. Lange (Ed.), Umformtechnik, [29] M.L. Wenner, State of the art of mathematical modelling of sheet
Handbuch fuÈr Industrie und Wissenschaft, Band 4, Springer, Berlin, metal forming of automotive body panels, in: Sheet Metal Stamping:
1993, pp. 330 435. Development Applications, SP 1221, Society of Automotive En
[21] K. Roll, Simulationssysteme im Vergleich Handlungsschwerpunkte gineering (SAE), 1997, pp. 1 6.
fuÈr die Zukunft, in: 4. SaÈchsiche Fachtagung Umformtechnik, [30] A.S. Wi®, An incremental complete solution of the stretch forming
November 5 6, 1997, Chemnitz, pp. 315 334. and deep drawing of a circular blank using a hemispherical punch,
[22] H. D. Schacher, Entwicklungstendenzen in der Massivumformung Int. J. Mech. Sci. 18 (1976) 23 31.
fuÈr die Automobilindustrie, in: Seminarband Neuere Entwicklungen [31] D.M. Woo, On the complete solution of the deep drawing problem,
in der Massivumformung, Stuttgart, Germany, June 3 4, 1997. Int. J. Mech. Sci. 10 (1968) 83 94.
[23] K. Schweizerhof, J.O. Hallquist, Explicit integration schemes and [32] H. L. Xing, A. Makinouchi, 3D thermal elastic plastic FEM in ®nite
contact formulations for thin sheet metal forming, in: FE Simulation deformation and its application to non isothermal sheet forming, in:
of 3D Sheet Metal Forming Processes in Automotive Industry, Vol. D.J. Owen, et al. (Eds.), Computational Plasticity, 1997, Pineridge
894, VDI Berichte, 1991, pp. 405 439. Press, Swansea, UK, pp. 1445 1452.

View publication stats