47

Research
Report

Highly Accurate Numerical Prediction of Springback Shape of

Stamped Thick Metal Sheet
Noritoshi Iwata, Atsunobu Murata, Yasuhiro Yogo, Hideo Tsutamori,
Masatomo Niihara, Hiroshi Ishikura

Abstract
In sheet metal forming simulations, the widely
used shell elements are assumed to be in the plane
stress state, as defined by the Mindlin-Reissner
theory. Unfortunately, numerical prediction with
conventional shell elements is not accurate for
bending radiuses that are small relative to the
sheet thickness. This is mainly because the stress
and strain formulation for a conventional shell
element does not actually reflect reality. So, to
accurately predict the springback of a sheet with a
severe bend, we have proposed a method for
measuring the through-thickness strain. The

Keywords

stress and strain are formulated based on

measured and calculated values for a solid
element, as well as a proposed shell element that
is based on a formulation that has been newly
incorporated into the FEM code. We have
confirmed the accuracy with which this method
can predict the springback shape of two bending
processes. As a result, we found that we can
accurately predict the springback shape even after
severe bending. From the viewpoint of
computation cost, the proposed shell element is
much more effective than a solid element.

Sheet metal forming, Finite element method, Numerical analysis, Bending, Springback,
Shell element

R&D Review of Toyota CRDL Vol. 41 No. 3

48

1. Introduction
Numerical simulations of sheet metal forming
have exhibited huge advances over the past ten years
or so.1-3) They have been applied to the production
of auto body parts and even side panels, which are
the largest parts for which sheet metal is used.
Breakage and wrinkling can be practically evaluated
by sheet metal forming simulations, but the forming
accuracy such as the springback shape is only
partially calculated. In the future, there will be a
strong demand for commercially viable high-level
simulations capable of shortening vehicle
development times and enabling coordination with
overseas production facilities.
For sheet metal forming simulations, the widely
used shell elements are assumed to be in the plane
stress state, as defined by the Mindlin-Reissner
theory.4) Unfortunately, numerical prediction using
conventional shell elements is not accurate when the
bending radius is small relative to the sheet
thickness. Rather, parts such as those with design
break lines and hemming, as well as underbody
parts, are more likely to be thought of as being
stamped under conditions that exceed the application
limits of shell elements. We considered the accuracy
of the springback prediction with conventional shell
elements that are incapable of fully representing the
above-mentioned parts. The main reason for this is
because the stress and strain formulation of
conventional shell elements does not actually reflect
reality.
This paper proposes a new measuring method to
confirm the through-thickness shear strain
distribution. A new shell element is proposed based
on the results obtained from both experiments and
solid FEM simulation. The newly formulated shell
element is introduced into FEM analysis. In
addition, the results of calculations using the new
proposed shell element are compared with both the
results of experiments and calculations using the
solid element on models of U- and hat-shaped
bending.

gauge, mark-off line and indentation were used for

the measurements. The surfaces and edges were
used for the measurement. For the formulation, it is
necessary to measure the strain distribution at the
center where there is no influence from the stress
state at the edge. Then, a solid element FEM is used
to consider the measurement method for the
through-thickness strain distribution. Therefore, we
proposed a new method in which a pin is buried
under a prepared hole, after which the strain is
measured from the deformation (Fig. 1).
2. 1 Measurement method
The pin has a prismatic shape and parallel
rectangular grooves leading up to the surface of the
specimen, which are used for locating after the
deformation (Fig. 2(a)). The grooves were aligned
parallel with the width edge and buried by copper
plating. The pin was lightly press-fitted into the
square hole of the specimen, after which the position
of the pinhead was aligned with the surface of the
test piece. The length of the pin was assumed to be
the initial distance between a neutral plane and the
surface, and was determined by FEM analysis. The
specimen was cut out in a plane parallel to a width
central section, which included the grooves buried
by plating of the pin, as shown in Fig. 2(b). The
coordinates of the angular points in the grooves were
measured, and then the change in the throughthickness direction was determined. Next, an
across-the-width central section including the center
part of the pin but without the grooves was cut out,
and the through-thickness angle distribution between
the side of the pin and the normal to the neutral

Neutral plane
Side wall

Pin
: Measurand
Tangent of
measurement line

2. Formulation of strain
The through-thickness strain distribution in a bend
was measured to formulate the strain. A strain

R&D Review of Toyota CRDL Vol. 41 No. 3

Measurement line

Normal of
neutral plane

Across-the-width center section

Fig. 1

Measurement method. (Measurement line is a

location where is the most inhomogeneous)

49

plane was calculated based on the above-mentioned

change in the through-thickness direction (Fig. 2(c)).
Figure 3 shows the measurement results. If the
bend radius R is 7 or less, then the angle is large at
the center and contradicts the assumption for a
conventional shell element, where the shear strain is
constant in the through-thickness direction.
2. 2 Formulation
We compared the solid FEM with the results
shown in Fig. 3, and confirmed that the actual
through-thickness strain distribution could be
estimated using the FEM. Figure 4 shows the
through-thickness shear strain distribution of the
FEM analysis. The results were approximated by a
quadratic function, and the curves were also plotted.
Based on these results, we can assume the following.
The shear strain actually grows near the center of
thickness, and plastic deformation is caused at the
center even though it does not occur with a
conventional shell element. In addition, the
unbending force increases because of plastic
Grooves are buried with copper plating

w
4
yz =
x 1 2 z 2

y
h

(1)

w
zx =
+ y 1 2 z 2
x
h

(2)

3. Formulation of through-thickness stress

The through-thickness stress is thought to
originate in the contact pressure because the work
bends as a result of continuous contact with the dies
in a common bending process. On the other hand,
the equilibrium of stress in the work can be

Line on measured points

Z
Z

0.09
0.18

deformation near the center of the thickness.

Therefore, we assumed that the formulation whereby
the shear strain rises near the center of thickness was
adequate, and that it is appropriate to assume the
equation quadratic polynomial, which does not need
additional parameters. We decided to consider the
nonlinearity of the through-thickness shear strain by
applying the following expression.

Z
Z

Section profile
X
X

Z1
Z2

0.1

X
X

(X1, Z1)
(X2, Z2)

1.7 ~ 2.1
Neutral plane
(unit: mm)

Pin
Pin

Grinding to center section

Bending process
Measurement X-coordinate of at measured
Measurement Z-coordinate of
Z-coordinate
angular point
: Angle of line on measured points to
normal to neutral plane

0.0

(a) Initial pin shape

(b) Pin shape after bending

(d) Photograph in center section

including pin

(c) Center section of pin

( 1 1 (1.7 ~ 2.1))

Measurement method for through-thickness shear strain.

R Bending radius/mm
R=5
R=5
R=6
R=7
R=11
R=18
R=29

10
8
6
4
2
0
-2
0

0.5

1.5

Initial distance from inner surface /mm

Fig. 3

Measurement values of .

Angle between initial through -thickness direction

and normal to neutral plane

Angle between initial through-thickness

direction
and normal to neutral plane /

Fig. 2

Fig. 4

R: Bending radius/mm
14

R=5
R=5
R=6
R=6
R=7
R=7
approximation:R=5
approximation:R=5
approximation:R=6
approximation:R=6
approximation:R=7
approximation:R=7

12
10
8
6
4
2
0

Initial distance from inner surface /mm

Approximation of numerical results of .

R&D Review of Toyota CRDL Vol. 41 No. 3

50

illustrated by the following expression.

rr 1
1 r zr
+ ( rr ) +
+
+ Fr = 0
r
z
r
r

(3)

The through-thickness stress relates to the bending

stress when the shear stress is disregarded in the
above-mentioned expression. The through-thickness
stress is believed to be considerable when the
bending stress is large and the bending radius is
small. These indicate that the through-thickness
stress originates chiefly from the bending
deformation and the contact pressure with the dies.
Hence, the stress can be formulated using the
following expressions.
3. 1 Through-thickness stress from bending
deformation
The next expression has equality based on the
equilibrium of stress in the case of bending
deformation. Here, zb is the through-thickness
stress and x is the hoop stress.
zb 1
= ( x z )
r
z

(4)

We assumed that the distribution of zb was shown

by using this expression and the least square.
3. 2 Through-thickness stress from contact
We assumed that the distribution of the throughthickness stress caused by the contact ( zc ),
according to Reissner's theory, when the sheet is
subject to contact pressure (p) from the tool can be
illustrated by the following expression (h:
thickness).
3
3 2 2 z 1 2 z
zc = p
+
4 3 h 3 h

(1) Calculate the contact pressure at the contact

location and then compute the through-thickness
stress from contact (zc) using Eq. (5).
(2) Calculate the through-thickness stress from the
bending deformation (zb) based on the curvature
radius and the stresses using Eq. (4).
(3) Calculate stresses that satisfy the yield criteria
when z is given.
The method for introducing the proposed shell
element changes the loop for updating the stress
from displacement, while the rigidity matrix remains
unchanged. Therefore, it is easy to introduce the
proposed shell element into all the FEM codes.
5. Results and discussion
The proposed shell element was verified for Uand hat-shaped bending processes, both of which are
basic bending processes.
5. 1 U-shaped bending process
In this processing, the influence on the springback
angle of the punch shoulder radius was compared.
The specimen was 590 MPa class hot-rolled steel
with a thickness of 4.5 mm. The work was stamped
by applying back pressure with a pad. The
springback calculation used a method that considers
the plastic strain dependency of the pseudo-elastic
modulus and Bauschinger effect during the unloading
Start
Calculate incremental in-plane and out- of-plane strain
Calculate contact pressure

(5)

at contact location

Calculate curvature radius of element

Here, the contact pressure (p) is assumed to be the

value at which the contact force is divided in the
effective contact area.

Calculate
Calculate

4. Implementation of proposed formulation

in FEM

using least-square method

using Equation F(

Calculate trial stress and plastic evaluation

Update stress

The main area where the proposed shell element is

introduced to FEM is the flow for calculating the
reaction force from the displacement increment of
the main FEM program. Figure 5 shows the flow.
It differs from the calculation flow for a conventional
shell element in the following points.

R&D Review of Toyota CRDL Vol. 41 No. 3

Calculate reaction force

Return

Fig. 5

Flowchart of numerical algorithm for implicit

integration using proposed shell element.

51

process, as proposed by several authors. The results

obtained by calculation were compared with those of
the experiments and are shown in Fig. 6. The results
were calculated using the proposed shell element,
solid element, and conventional shell element. It
was found that the accuracy of the proposed shell
element is equal to the solid results without
depending on the punch shoulder radius. There are
large differences between the results obtained with
the conventional shell element and those with the
solid element when the ratio of the bend radius and
thickness is less than 5. We confirmed that the
proposed shell element was effective for severe
bending where the ratio of the bend radius and
thickness is less than 5.
5. 2 Hat-shaped bending process
The shoulder radius of the punch and die are some
of the factors that determine the springback shape in
hat-shaped bending. The radius of the die shoulder
is discussed here. The specimen was 440 MPa class

hot-rolled steel with a thickness of 2.3 mm. Figure 7

shows the springback shape when the radius of the
die shoulder is 3 mm as an example of our
calculation results. The result of our experiments
and the results obtained using the solid element had
little tendency to springback, while the result
obtained with the conventional shell element showed
a spring-go phenomenon different from the others.
On the other hand, the result obtained using the
proposed shell element closely corresponds to the
results obtained with the solid element. The values
used for the springback evaluation were the opening
distance and the curvature, and are shown in Fig. 8.
The values are compared for three radii of the die
shoulder. Figure 9 shows the results. It was found
that the two values obtained using the conventional
shell element differ greatly from the others when
r/t = 1.3. With the conventional shell element, the
opening distance and curvature are negative and
large, and indicate the spring-go phenomenon. On
Opening distance

Proposed
shell element
Conventional
shell element

CC
uruvravt
autruerera
rdaiduisu
s
80

29
R29

Springback angle in U-shaped

bending.

Fig. 7

Springback shape in
hat-shaped bending.

14
12

Proposed
10
shell element
8
Solid
6 element
4
2
0
-2
Experimental
-4
Conventional shell element
-6
1.3
2.2
4.3
Bending ratio r/t

(a) Opening distance

Fig. 9

Punch

Experimental
Solid element

Curvature /mm-1

Fig. 6

Opening distance /mm

Springback angle /

Punch width
10
t = 4.5
9
8
8
Proposed
7 Solid element
shell element
66
Experimental
Conventional
shell element
5
44
3
22
1
00
5
7
11
R5
R7
R11
Punch radius /mm

Fig. 8

Opening distance and

curvature radius.

10 -3
Solid element
Proposed
shell element

1
0
-1

Experimental

-2
-3

Conventional shell element

1.3

2.2
Bending ratio r/t

4.3

(b) Curvature

Relationship between bending ratio r/t and springback evaluation values.

R&D Review of Toyota CRDL Vol. 41 No. 3

52

the other hand, the results obtained with the

proposed shell element correspond both to those
obtained with the solid element and those obtained
by experimental. Moreover, the calculation time for
the proposed shell element was about twice that for a
conventional shell element. On the other hand,
when this object is modeled using a solid element
for high prediction accuracy, the calculation time is
ten times longer than with a conventional shell
element, and it is expected to become several
hundred times longer in the case of actual auto body
parts. The above-mentioned results suggest that the
accuracy of the springback prediction can be
improved and the increase in calculation time
suppressed when the proposed shell element is
applied to FEM.
6. Conclusion
To accurately predict the springback of a bent
sheet with a severe bend, a method of measuring the
through-thickness strain has been proposed. The
stress and strain were formulated based on the
results obtained by measurement and calculation
using a solid element, as well as the proposed shell
element that was based on a formulation that is
newly introduced to the FEM code. The accuracy of
this method's prediction of the springback shape of
two bent processes has been confirmed. As a result,
we found that the springback shape can be predicted
with high accuracy even under severe bending.
Moreover, the calculation time with the proposed
shell element is about twice that for a conventional
shell element, and has been shortened to about 1/20
that of a solid element. In addition, further
shortening to 1/100 or less compared to the time
needed for a solid element is considered possible for
actual auto body parts. Therefore, the springback
prediction method using the proposed shell element
is useful in die design from the standpoints of
efficiency and accuracy.

Simulation Using Static Explicit FEM Program

Coupled with Elastic Deformation of Tools",
J. Jpn. Soc. Tech. Plast., 47-540(2006), 64-68
3) Tsutamori, H. and Yoshida, F. : "Analysis of
Springback in Sheet Metal Stamping Considering
Elastic Deformation of Dies", J. Jpn. Soc. Tech.
Plast., 46-532(2005), 407-411
4) Reissner, E. : " The Effect of Transverse Shear
Deformation on the Bending of Elastic Plates",
J. Appl. Mech., 67(1945), A69-A77
(Report received on Jun. 21, 2006)
Noritoshi Iwata
Research fields : Metal forming
Academic degree : Dr. Eng.
Academic society : Jpn. Soc. Technol.
Plast.
Awards : Jpn. Soc. Technol. Plast. Tokai
Chapter Award for Outstanding
Research, 1998
Atsunobu Murata
Research fields : Metal forming
Academic society : Jpn. Soc. Technol.
Plast.
Awards : Jpn. Soc. Technol. Plast. Tokai
Chapter Award for Outstanding
Research, 2002
Yasuhiro Yogo
Research fields : Metal forming
Academic society : Jpn. Soc. Technol.
Plast., Iron Steel Inst. Jpn., Jpn.
Welding Soc.
Hideo Tsutamori*
Research fields : Sheet metal forming
simulation
Academic degree : Dr. Eng.
Academic society : Jpn. Soc. Technol.
Plast.
Awards : Paper Award of Jpn. Soc.
Technol. Plast., 2006

Masatomo Niihara*
Research fields : Sheet metal forming
simulation
Academic society : Jpn. Soc. Technol.
Plast.

References
1) Tsutamori, H., Iwata, N. and Suzuki, N. : "Prediction
of the Geometrical Defects of Sheet Metals in 3
Dimensional Shape", J. Jpn. Soc. Tech. Plast.,
44-513(2003), 1024-1028
2) Takamura, M., Ohura, K., Sunaga, H., Kuwabara, T.,
Makinouchi, A. and Teodosiu, C. : "Sheet Forming

R&D Review of Toyota CRDL Vol. 41 No. 3

Hiroshi Ishikura*
Research fields : Development of
instrumentation engineering
Academic society : Jpn. Soc. Technol.
Plast.
* Toyota Motor Corp.