You are on page 1of 12

Journal of Materials Processing Technology 145 (2004) 72–83

Evaluation of sheet metal formability by tensile tests


Stefan Holmberg a,∗ , Bertil Enquist b , Per Thilderkvist c
a Volvo Cars Body Components, SE-293 80 Olofström, Sweden
b Department of Structural Mechanics, Chalmers University of Technology, 412 96 Göteborg, Sweden
c Industrial Development Centre, V. Storgatan 20, SE-293 38 Olofström, Sweden

Received 3 July 2002; received in revised form 18 July 2003; accepted 21 July 2003

Abstract

The use of sheet metal forming simulations has reduced lead-times and costs for the development of new car bodies significantly. The
accuracy of the simulations is to a large extent dependent on the quality of the material properties provided as input to the simulations.
Improving the quality of the material properties is the key factor in order to further increase the accuracy of the simulations. This study is
focused on the forming limit properties of sheet metal.
A test procedure for determining the forming limit in plane was developed. The tests are carried out in a tensile testing machine. The
test method enables a fast and easy determination of the forming limit and the results show small scatter. Furthermore, the influence of
friction is eliminated. By modifying the specimen geometry, the test procedure can be applied to determine the complete left-hand side of
the forming limit curve (FLC).
The specimen geometry and the grip arrangements are essential in order to get successful tests. With inappropriate specimen design or
insufficient clamping, fracture will not occur at the desired region of the specimen.
The study involves both numerical and experimental work. Simulations were performed in order to find appropriate specimen design
prior to testing and for performing parameter studies. One mild steel and one high strength steel was included in the study.
© 2003 Elsevier B.V. All rights reserved.

Keywords: Sheet metal; Forming limit; Material testing; Optical deformation measurement

1. Introduction simulations and reality is the experimental determination of


relevant material properties. The accuracy of the simulations
Large efforts are being made in the automotive industry is to a very large extent dependent on the quality of the mate-
in order to shorten lead times and reduce costs when de- rial properties provided as input to the simulations. Improv-
veloping new car models. In order to manage this, physical ing the quality of the material properties is the key factor
testing has to a large extent been replaced with numeri- in order to further increase the accuracy of the simulations.
cal simulations. Simulations are used to verify both the Material properties of main interest for sheet metal forming
properties of the car and the car components as well as are stress–strain relations describing the work-hardening of
their producability. An area of which significant progress the material and forming limits describing how much the
has been made is finite element simulation of sheet metal material can be deformed without cracking. This study is
forming. Simulations have to a large extent replaced try-out focused on the forming limit properties. Improved methods
tools as the process verification method. In this way, both for material characterisation of sheet metal is a field that
lead-times and costs for the development of new car body has gained a larger focus during the last years [2–4].
parts have been reduced considerably [1]. The need for accurate material characterisation methods
Much work has been made during the last years in order is of great importance in order to further strengthen the use
to increase the accuracy of the finite element programs and of the simulation technique for sheet metal forming appli-
in the modelling technique. One area of significant impor- cations. This is becoming of even more important due to
tance in order to further increase the correlation between the increased use of new high strength materials in order
to reduce weight and increase the crash performance. The
problem with these new high strength sheet materials is that
∗ Corresponding author. Tel.: +46-454-265329; fax: +46-454-265788. their increase in strength is compensated by a reduction in
E-mail address: sholmber@volvocars.com (S. Holmberg). formability. Optimal usage of these new materials requires

0924-0136/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2003.07.004
S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83 73

Major strain 1

Plane
Uniaxial
strain
tension
Pure
shear
FLC
Equal
biaxial
tension
Uniaxial
compression

Minor strain 2

Fig. 2. Forming limit diagram indicating safe forming region.


Fig. 1. Forming limit curve.

a deep knowledge of the material properties so that it is studying the strain levels. A forming limit diagram is con-
possible to be close to the forming limit. structed from the FLC and is normally used in the design of
A fundamental problem in sheet metal forming is frac- the forming process (Fig. 2).
turing. It is therefore essential to be able to predict the There is no unique standard for determining FLCs, the
risk of fracture with high accuracy. The forming limit curve main routine when determining a FLC is as follows, how-
(FLC) is the most commonly used fracture criterion for ever. By forming a number of sheet specimens with varying
sheet metal forming applications [5–7]. The FLC shows widths, different deformation modes (strain states) are re-
the amount of deformation (strain) a sheet material can re- ceived (Fig. 3). The sheet specimens are equipped with cir-
sist as function of the deformation mode and is a relation cle grids in order to enable strain evaluation. The specimens
between the major and minor strain (Fig. 1). In the fig- are stamped to fracture and the strain state is evaluated just
ure, the different main deformation modes are indicated. outside the fracture zone. The limiting strain levels deter-
The minimum of the FLC is normally at the plane strain mined from these different specimens are then connected
condition. to a curve, by some kind of curve-fitting. The procedure for
In order to avoid fracturing of the material, it is neces- determining FLCs may differ concerning the geometry of
sary that the strain levels everywhere in the stamped part the forming punch used, the specimen dimensions, the num-
be below the FLC. A safety margin is normally introduced, ber of specimen geometries considered, the technique used
resulting in an offset of the FLC. The risk of fracture is de- for the strain states evaluation and the type of curve-fitting
termined by evaluating how close the strain condition is to used. An example of tool and specimen geometries used
the FLC. It is not sufficient only to check the risk of fracture for determination of FLCs is given in Fig. 3. An example
when designing a forming process, however. Other problems of a FLC is shown in Fig. 4. In the diagram, the mea-
such as excessive thinning, wrinkling or insufficient stretch sured strain levels using different specimen geometries are
may occur. These conditions are normally also evaluated by shown.

Fig. 3. Example of tool geometry and stamped specimens used for FLC determination together with fractured specimen after testing.
74 S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83

Fig. 4. Example of FLC, showing measured strain levels using different specimen geometries (indicated by the different colours) and example of curve-fitting
to the measured strain points. For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.

The FLC concept has a number of limitations and uncer- New fracture criterions have been developed but no one has
tainties. One fundamental problem with the FLC concept yet come to general practical use in the automotive industry
is that it is based on the assumption of linear strain paths, [10,11].
i.e. the mode of deformation remains constant throughout The forming limit at the plane strain condition is of par-
the deformation process. This condition is not fulfilled in ticular interest. This is due to several reasons listed below:
most practical cases, however. A lot of studies concern-
ing the effect of non-linear (or broken) strain paths on • It is, normally, the mode of deformation where the ma-
the FLC have been conducted [8,9]. In these studies, the terial can resist the lowest amount of strain, i.e. it is the
FLC has been determined on specimens which have been minimum point of the FLC.
subjected to different types of pre-straining conditions. • It is often a critical strain condition for real stamped parts.
Taking the effect of non-linear strain paths into account in • The plane strain condition is a very unstable condition
practical process design is very complex, however, since with a rapid localisation.
an infinite number of different pre-straining conditions • The forming limit in plane strain is often used when rank-
exist. ing the formability of different materials.
Another problem with the FLC is that the experimental • The plane strain formability limit is a good measure to be
results are influenced by the friction conditions during test- combined with theoretical models of FLCs.
ing. There is also a small deviation from linear strain paths
due to bending effects during forming. A problem in prac-
tical forming process design is that the determination of the 2. Aim and scope of the present study
FLC is a rather lengthy procedure. In the automotive indus-
try, there is often a shortage of time for the process design The main objective of this study was to evaluate a test
phase, which means that there sometimes is not enough time procedure for determination of the formability of sheet
to perform a thorough FLC determination. Therefore, differ- metal through tensile testing. The main focus in the study
ent theoretical and empirical formulas have been developed. was the forming limit in the plane strain condition but
Another drawback of significance is the large scatter in ex- other strain states were also considered. The main expected
perimental results when determining a FLC. This is due to advantages of this test procedure as compared to the con-
the overall testing procedure as well as the strain evaluation ventional forming limit determination would be: a fast and
methodology. simple test with no need of other testing equipment than
Despite the drawbacks with the concept of FLCs, it is, a uniaxial tensile testing machine, a low scatter in experi-
by far, the most commonly used fracture criterion for sheet mental results, and no influence of friction on the results.
metal forming applications. This is due to its, in many cases, Another advantage is that this test procedure simplifies the
good performance, its simplicity and for historical reasons. usage of optical equipment for the strain determination.
S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83 75

Table 1
Material characteristics for the two materials included in the study
Material t (mm) Rp0.2 (MPa) Rm (MPa) r0 r45 r90 n E (GPa) ν

Mild steel 1.0 137 284 1.99 2.61 2.13 0.24 210 0.3
High strength steel 1.0 278 469 0.98 0.66 1.22 0.21 210 0.3

Another aim with the investigation was to get a better properties of these materials are given in Table 1. In the
understanding of the material behaviour prior to and during table, t is the initial sheet thickness, Rp0.2 the yield strength,
localisation and fracturing. Rm the ultimate tensile strength, r0 , r45 , r90 the anisotropy
The study involves both numerical simulations and ex- coefficients, n the strain hardening exponent, E the Young’s
perimental work. The simulations were performed in order modulus and n the Poisson’s ratio. Stress–strain relations
to find appropriate design of the specimens prior to testing describing the work-hardening behaviour of the materials
and for performing various parameter studies. Two different are shown in Fig. 6. These data were determined through
materials were included in the study, a mild steel and a high standard uniaxial tensile tests. FLCs for the materials are
strength steel. given in Fig. 7. The curves are determined through the con-
ventional method by stamping specimens of different widths
to fracture and determine the strain levels by circle grid
3. Specimen geometries evaluation. The curves are fourth order polynomial fits of
the measured strain data points. As can be seen, the material
Several different specimen geometries were evaluated, properties differ significantly between the grades. The high
both numerically by finite element simulations and experi- strength steel grade has yield stresses and tensile strengths
mentally. A schematic drawing of a specimen is shown in of about two to three times as high as the mild steel. The
Fig. 5. A large part of the specimen length is used for the anisotropy coefficients differ significantly. The FLCs also
clamping. The total length of the specimen was approx- differ, the high strength steel generally, show lower forma-
imately 120. The specimens were provided with a notch bility. The forming limit in plane strain is about 40% for
in order to receive a well-defined localisation and fracture the mild steel and about 30% for the high strength steel.
region in the middle part of the specimen. Specimens with
different widths (w1 and w2 ), initial free lengths (h0 ) and
notch sizes were evaluated. In this study, specimens widths 5. Experimental procedure
(w1 and w2 ) ranging from 16 up to 76 mm and free lengths
ranging from 6 up to 12 mm were considered. The experiments were performed in an uniaxial material
testing machine (Fig. 8). The testing machine is built up with
a load frame of a solid T-slot table, columns and a hydrauli-
4. Materials cally manoeuvrable crosshead to which the servo-hydraulic
actuator with capacity of ±100 kN is mounted. Control and
Two different materials were included in the study, a data acquisition is performed by computer boards and spe-
mild steel and a high strength steel. The basic mechanical cially designed programs. To be able to clamp the specimen
in the desirable manner, special attention was put on the
shaping of the wedges, specially their surface towards the
specimen. Sliding between the specimen and the wedges
should be avoided during loading (or kept to a minimum).
Measuring equipment, consisting of two cameras and a
computer for picture grabbing and analysis [12], was used
in order to measure deformations and strain distributions
in the specimens during testing. Specimens used for opti-
cal deformation measurements were sprayed with a dark
colour forming a random pattern. The other specimens
were equipped with circle grids in order to facilitate strain
measurements after testing.
The test was run in deformation control with the initial
piston rate of 0.2 mm/min and was, after the elastic limit,
successively increased to about 0.5 mm/min. After the crack
was opened to at least half the specimen width and the ap-
plied load was decreased to about one third of the maximum,
Fig. 5. Schematic drawing of specimen geometry used in the tests. the test was stopped.
76 S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83

500

450 Mild steel

400 High strength steel


Tensile stress [MPa]

350

300

250

200

150

100

50

0
0,0 5,0 10,0 15,0 20,0 25,0 30,0 35,0 40,0 45,0 50,0 55,0

Engineering strain [%]

Fig. 6. Tensile stress–strain curves for the materials included in the study.

0,80

0,70

0,60
Major strain

0,50

0,40

0,30

0,20

0,10

0,00
-0,40 -0,30 -0,20 -0,10 0,00 0,10 0,20 0,30 0,40
Minor strain

0,80

0,70

0,60
Major strain

0,50

0,40

0,30

0,20

0,10

0,00
-0,40 -0,30 -0,20 -0,10 0,00 0,10 0,20 0,30 0,40
Minor strain

Fig. 7. FLCs for the materials included in the study: mild steel (above); high strength steel (below).
S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83 77

placement controlled loading in the y-direction at the nodes


on the upper boundary.
The yield criterion of Hill was adopted [14]. Hill’s po-
tential function is an extension from the Mises function and
can be expressed as

F(σ22 − σ33 )2 + G(σ33 − σ11 )2 + H(σ11 − σ22 )2
f(σ) =
+ 2Lσ23 2 + 2Mσ 2 + 2Nσ 2
31 12

where σ ij denote the stress components and F, G, H, L, M


and N are material constants obtained by tests of the material
in different directions. These constants can be expressed in
Fig. 8. Testing equipment used for the experiments.
terms of six yield stress ratios R11 , R22 , R33 , R12 , R13 and
R23 according to
 
6. Numerical procedure 1 1 1 1
F= + 2 − 2 ,
2 R222 R33 R11
The analyses were carried out in Abaqus/Standard [13]  
with static conditions assumed and shell elements employed. 1 1 1 1
A simplified modelling approach was used, only the cen- G= + 2 − 2 ,
2 R233 R22 R22
tral part of the specimens was modelled (Fig. 9). The grip  
arrangements of the testing machine were not taken into ac- 1 1 1 1
H= + 2 − 2 ,
count. This means that the strain distribution predicted by 2 R211 R22 R33
the simulations will not be exactly correct if there is slip
3 3 3
in the grip conditions. Therefore, there should be a ‘safety’ L= , M= , N=
margin in the simulations in terms of the strain distribution 2R223 2R213 2R212
in order to compensate for slip in the grips. Slip will in- In sheet metal forming applications, it is common to use
crease the strains at the edges and decrease the strains in anisotropic material data in terms of ratios of width strain
the middle part of the specimen and thereby might trig a to thickness strain. The stress ratios can then be defined as
fracture at the edges of the specimen and not in the middle 
region. r90 (r0 + 1)
R11 = R13 = R23 = 1, R22 = ,
The nodes at the upper boundary of the modelled spec- r0 (r90 + 1)
imen were constrained to move in the x- and z-directions  
and the nodes at the lower boundary were constrained in r90 (r0 + 1) 3r90 (r0 + 1)
R33 = , R12 =
the x-, y- and z-directions. The loading is applied as a dis- r0 + r90 (2r45 + 1)(r0 + r90 )

Fig. 9. Only the central part of the specimen (specimen free length) was modelled in the numerical analyses.
78 S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83

where r0 , r45 and r90 are the anisotropy coefficient in the FLC as used normally in sheet metal forming simula-
terms of width strain to thickness strain ratios obtained tions.
from uniaxial tensile tests in different directions, see In order to capture the strain localisation, the mesh size
Table 1. The material hardening behaviour is modelled must be sufficiently small. The model used consists of about
as a piecewise linear true stress curve as function of the 1200 element, with typical element lengths of about 0.4 mm.
effective plastic strain, based on the tensile stress–strain The mesh size chosen is judged to be sufficiently small in
curves shown in Fig. 6. A linear extrapolation of the true order to capture the accuracy necessary for this investiga-
stress–strain curve is also made in order to be able to tion. Test simulations with a finer mesh (element lengths of
capture strain levels beyond the maximum levels obtained about 0.20 mm) were performed showing very similar re-
from the uniaxial tensile tests. The risk of fracture is eval- sults. The coarser mesh was then chosen in order to reduce
uated by comparing the strain state in the specimen with the computation time.

Fig. 10. Deformation process for plane strain test, the pictures are linked to the recorded stress–strain curve below. The results show the behaviour of
the mild steel material.
S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83 79

7. Results and discussion pictures are linked to the recorded stress–strain curve also
shown. The results shown are for the mild steel grade. As
The deformation process of a specimen for plane strain can be seen, fracture occurs in the middle region of the
forming limit evaluation is shown in Fig. 10. The speci- specimen. The crack does not become visible until the
men is shown at six different levels of deformation, from force has decreased significantly. It should be noted that
the undeformed state to a state with an open crack. The the complete stress–strain curve, including the softening

Fig. 11. Strain distributions for plane strain test at three different levels of deformation, see link to the recorded stress–strain curve: (left) longitudinal
strain; (right) transverse strain. The results show the behaviour of the mild steel material. The region where an open crack is developed is removed from
the strain evaluation and is shown in black.
80 S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83

branch, is captured. This is enabled by the short free length case. A lower free length complicates the strain evaluation,
of the specimen in combination with the high stiffness of however.
the testing machine. The measured strain distributions in For high strength steel, the strain condition in the middle
the specimen at different levels of deformation are shown part of the specimen was closer to an exact plane strain con-
in Fig. 11. The region where an open crack is developed is dition than for the mild steel. The transverse strain was about
removed from the strain evaluation and is shown in black. 1–2% units lower for the high strength steel compared with
As can be seen, there is a clear localisation in the middle the mild steel. This was seen both in the numerical simula-
of the specimen. An exact state of plane strain in the frac- tions and in the practical testing. The reason for this is the
ture area was not reached. The transverse strain in and just difference in anisotropy between the high strength steel and
outside the fracture zone is very low, however. the mild steel. The mild steel shows a significant anisotropy,
In order to get a condition as close to plane strain as with r-values of about 2. A high r-value results in that the
possible in the middle region of the specimen, the specimen material has a tendency of compensating the axial strain-
width should be as large as possible. The maximal width of ing by transverse compressive straining rather than through
the specimen is restricted by the test equipment, however. thinning. For materials with a high degree of anisotropy, it
The wider the specimen, the higher the necessary maxi- is therefore somewhat more difficult to get a condition of
mal tensile force and the necessary clamping force. If the plane strain in the middle of the specimen than for material
clamping force is too low, this will result in a diffuse grip with a low degree of anisotropy.
condition with significant slip between the specimen and A comparison between simulated response and experi-
the grips. Also, a wide specimen requires a corresponding mental results are given in Fig. 12. The pictures show the
width of the wedges. In this study, the maximum specimen strain distributions in the specimen just before cracking.
width was 76 mm. The transverse strain in the middle re- The colour scales used in the simulation are similar but
gion of the specimen was in the region of 2–5%. Another not identical to those used for the experimental results. It
way of reducing the transverse strain is to decrease the can be concluded that the results from the simulations are
specimen free length. The ratio between the width and the close to the experimental results. The simulated response,
free length of the specimen should be as high as possible. generally, indicates a somewhat more distinct localisation
By reducing the free length from 8 to 6 mm, the transverse and slightly lower transverse strains than the experimental
strains decrease from 4 to about 1% for the mild steel results. This difference is mainly due to the grip conditions

Fig. 12. Comparison of results from numerical simulation and experimental results. Strain distributions just before cracking: (left) simulation; (right)
experiments.
S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83 81

Fig. 13. Influence of notch size on strain distribution in specimen for plane strain fracture limit determination. Strain plots for a central cut through the
specimen for specimens with different notch sizes.

assumed being perfect in the simulations. Large efforts were results obtained by the conventional forming limit determi-
made in order to minimise sliding between the specimen nation. A number of similar specimens were tested (nine
and the grip wedges but ideal conditions were very difficult mild steel specimens and eight high strength steel speci-
to achieve in reality. mens). The specimens were equipped with circle grids and
The influence of the notch size on the strain distribu- the strains just outside the fracture zone were evaluated.
tion was investigated. In Fig. 13, results from numerical For each specimen, the strains were evaluated at three lo-
simulations showing the influence of the notch size on the cations. An average forming limit strain of 39 and 31%
strain distribution is given. The strain distribution for a was received for the mild steel and the high strength steel,
central cut through the specimen for specimens with four respectively. These values show good agreement with the
different notch sizes are shown. The results show the strain conventionally determined forming limits (Fig. 7). It can
distribution when the strain in the middle of the specimen be concluded that the results show small scatter both con-
approaches 40%. As can be seen, the notch size clearly in- cerning the measurements of the individual specimens and
fluence the strain distribution at the edges of the specimen. when comparing the results from the different specimens.
The strain mode in the middle of the specimen is not af- By changing the specimen geometry, the strain state is
fected, however. A large notch give rise to high strain levels changed and fracture limits in other strain modes than plane
and thereby a higher risk of fracture at the edges of the spec- strain may be determined. This is illustrated in Fig. 14.
imen. A large notch might thus induce fracture at the edges In the figure, results from numerical simulations of three
of the specimen and not, as desired, in the middle region of different specimen widths are given. The diagram show the
the specimen where plane strain conditions apply. This risk predicted strain state at fracture for the different specimen
of unwanted fracture at the edges of the specimen is further configurations. The results shown are for the mild steel.
increased in practical testing, since perfect grip conditions By decreasing the width of the specimen, the strain state
are not possible to achieve. In the practical testing, a large changes from plane strain and approaches that of uniaxial
number of the specimens with the large notch size were tension. Another alternative is to increase the free length of
unsuccessful with fracture coming from the edges of the the specimen or change both the width and the free length.
specimen. It is advantageous to have a rather wide specimen in order
The fracture limits in the plane strain condition deter- to get as small boundary effects as possible and to have a
mined by this procedure are in good agreement with the wide fracture area in order to ease the strain evaluation.
82 S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83

0,80
1: w1=34, h0=8
0,70 2: w1=22, h0=8
3: w1=22, h0=12
0,60
4 3 4: w1=16, h0=8
2
Major strain

0,50
1
0,40

Uni-axial 0,30
tension
0,20

0,10

0,00
-0,40 -0,30 -0,20 -0,10 0,00 0,10 0,20 0,30 0,40
Minor strain

Fig. 14. Results from numerical simulations showing the influence of specimen width and specimen free length on the strain state at fracture. w1 denote
the maximal width and h0 the free length of the specimen (Fig. 5).

8. Conclusions With a too large notch, the strains at the edges of the
specimen become large and fracture occur at the edges of
The following conclusions apply: the specimen and not in the desired middle region of the
specimen.
• A test procedure for determining the forming limit in plane • A test procedure for determining the complete left-hand
strain for sheet metal was developed. The main advan- side of the FLC by tensile tests was outlined. By changing
tages with the method are that it enables a fast and easy the geometry of the specimen, different strain conditions
determination of the formability. The results show small are achieved. The right-hand side of the FLC cannot be
scatter as compared to the conventional methods and it determined by this procedure, however.
can be carried out in a tensile testing machine. It does not • The short free length of the specimens and the high
require any forming tools or press. Furthermore, the re- stiffness of the testing machine enabled stable test
sults are not influenced by the friction. Exact plane strain performance and thereby recording of the complete
conditions were difficult to achieve, a minor compressive stress–strain curve, including the softening branch. The
strain of about 2–4% occurred. Successful tests were car- test procedure can therefore be used in development
ried out for both a mild steel and for two different high and verification of fracture mechanics models describ-
strength steels. ing the localisation and softening behaviour of the
• The grip arrangements are essential in order to get suc- material.
cessful tests. With insufficient clamping of the specimen,
localisation and fracture will not occur at the desired mid-
dle region of the specimen where plane strain conditions
apply. References
• The specimen width influences the strain state signifi-
cantly. The specimens width should be as large as possible [1] A. Andersson, Use of FE-analysis in predicting and verifying
the design of an automotive component forming process, Report
in order to get a condition as close to plane strain as pos-
TMMV-1044, Ph.D. Thesis, Division of Production and Materials
sible in the middle region of the specimen. The maximum Engineering, Lund University, Sweden, 2001.
possible specimen width is restricted by the testing equip- [2] H. Pijlman, Sheet material characterisation by multi-axial ex-
ment in terms of the maximum available tensile force, the periments, University of Twente, The Netherlands, 2001. ISBN
maximum available clamping force and the width of the 90-36516951.
[3] R. Knockaert, Numerical and experimental study of the strain lo-
wedges.
calization during sheet forming operations, Ph.D. Thesis, CEMEF,
• The specimen free length has a large influence on the re- Sophia-Antipolis, France, 2001.
sults. The specimen free length should be as small as pos- [4] P. Hora, A. Feurer, A. Wahlen, J. Reissner, Methods for handling of
sible in order to get a condition as close to plane strain as FEM input and output data with the goal of higher computational
possible in the middle region of the specimen. The strain reliability, in: Proceedings of the European Congress on Computa-
tional Methods in Applied Sciences and Engineering (ECCOMA),
evaluation is complicated by decreasing the specimen free
Barcelona, 2000.
length too much, however. [5] P. Keeler, Sheet Met. Ind. 42 (1965) 683.
• It can be concluded that the notch size has a large influ- [6] G.M. Goodwin, Application of strain analysis to sheet metal forming
ence on the stress and strain distribution in the specimen. problems in the press shop, SAE Paper no. 680093, 1968.
S. Holmberg et al. / Journal of Materials Processing Technology 145 (2004) 72–83 83

[7] Z. Marciniak, K. Kuczynski, Limit strain in the processes of [11] P. Hora, L. Tong, Prediction methods for ductile sheet metal failure
stretch-forming sheet metals, Int. J. Mech. Sci. 9 (1967) 609–620. using FE-simulation methods, in: Proceedings of the IDDRG’94,
[8] A. Graf, W.F. Hosford, Calculations of forming limit diagram for Lissabon, 1994, pp. 363–375.
changing strain paths, Metall. Trans. 24 (1993). [12] K. Galanius, Manual Aramis 4.7 Deformation measurement using
[9] P. Friedman, D. Houston, Effects of prestraining on the formabil- the grating method, GOM mbH Braunschweig, Germany, 2001.
ity of Al–Mg sheet alloys, Research Report SRR-1999–0091, Ford [13] Hibbit, Karlsson & Sorensson Inc., ABAQUS, Version 6.2, Paw-
Research Laboratories. tucket, RI, USA, 2001.
[10] T.B. Stoughton, A general forming limit criterion for sheet metal [14] R. Hill, The Mathematical Theory of Plasticity, Oxford University
forming, Int. J. Mech. Sci. 42 (2000) 1–27. Press, London, 1950.