Journal of Materials Processing Technology 232 (2016) 19–33

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

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

Effect of through thickness strain distribution on shear fracture

hazard and its mitigation by using multilayer aluminum sheets
Maysam Gorji ∗ , Bekim Berisha, Niko Manopulo, Pavel Hora
ETH Zurich, Institute of Virtual Manufacturing, Tannenstrasse 3, 8092 Zurich, Switzerland

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

Article history: The formability of sheet metal is generally accepted to be limited by the so-called Forming Limit Curve
Received 7 August 2015 (FLC). This is valid for a majority of operations, given that localization immediately precedes fracture. For
Received in revised form 15 January 2016 the characterization of the latter, a microscopy based new method is applied on Nakazima specimens.
Accepted 16 January 2016
Furthermore a simple cup deep drawing test is employed to investigate the shear failure phenomenon
Available online 22 January 2016
and extend the fracture limit in the left hand side of the principal strain space. In addition, fracture
phenomenon in aluminum alloy AA6016 sheet sample is investigated by using different fracture criteria.
Keywords: f f
The measured principal fracture strains ε11 and ε22 are then implemented in the FE-code to predict
Forming Limit Curve (FLC)
Fracture strain
rupture. Moreover, initiation and propagation of cracks in monolithic and multilayer aluminum alloys,
Anisotropic hardening are studied. A newly designed triangle shape deep drawing part is used in the framework of this work to
Fracture model validate the constitutive and failure models as well as investigate the advantage of a multilayer material,
Shear loading compared to the monolithic one.
Multilayer aluminum alloy © 2016 Elsevier B.V. All rights reserved.

1. Introduction uniaxial tensile setup with different geometries in order to avoid

friction effects and by Hecker (1972) who kept the sheet geometry
Necking and consequent tearing are identified as important constant but adjusted the friction conditions to obtain the different
failure mechanisms in sheet metal forming. Therefore better under- strain paths. Nakazima test is recommended to determine the FLC
standing of this phenomena is essential. Furthermore, robust in order to cover a big variety of strain paths, according to DIN EN
processes, based on the necking criterion help reduce costs and ISO 12004-2 (2009). In this test, it is possible to obtain different
increase efficiently of manufacturing processes. A Forming Limit major to minor strain ratios, ranging from uniaxial to equi-biaxial
Curve (FLC) is widely accepted in sheet metal forming applications tension, by changing the width of the work-piece.
to evaluate deep drawing simulations. The pioneer research in this Despite some of the uncertainties and the overhead required,
field was conducted by Keeler (1961), who plotted the major strains experimental FLCs are still considered the most reliable way in pre-
against the minor strains for biaxial stretching. The research was dicting localization. The experimentally measured Forming Limit
continued by Goodwin (1968) for the tension-compression domain. Curves are subject to significant uncertainties. These are on the
Strains below and above the FLC are usually recognized as a safe and one hand due to the scattering material properties, which can
failed, respectively. Different methods for determining the strain sensitively affect the localization behavior. On the other hand,
limits have been proposed. The formability limit under the biax- due to the continuous nature of localization, it is challenging to
ial loading has been investigated by Olsen (1920) who used the define an objective evaluation of the results. In fact, Keeler (1961)
hydraulic bulge test to avoid friction effects. Later, a variable geom- and Goodwin (1968) originally proposed these limits as bands, to
etry experiment setup has been proposed by Marciniak (1965) include a certain amount of uncertainty.
which featured a flat cylindrical punch deforming the different The high cost of the FLCs, historically spurned the search for
specimens until rupture. Nakazima et al. (1971) proposed a vari- theoretical prediction of that. Many theoretical studies about the
ant of the latter using a spherical punch to ensure that the strains necking phenomenon are initiated in the fifties of the last century.
localize in the center of the specimen. Alternative methods have Hill (1952) proposed the failure criterion due to the localized neck-
been proposed by Brozzo et al. (1972) who advocated the use of a ing which is characterized based on the principal tensions (forces
per width) of the blank. This model is assumed based on the max-
imum major tension under the instability condition. One of the
∗ Corresponding author.
well-known approach to predict the localization is the model pro-
E-mail address: [email protected] (M. Gorji).
posed by Marciniak and Kuczynski (1967) (M–K). Based on this

http://dx.doi.org/10.1016/j.jmatprotec.2016.01.014
0924-0136/© 2016 Elsevier B.V. All rights reserved.
20 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

theory, localization concentrates in a narrow band because of an of this work, analytical framework to characterize fracture loci is
assumed inhomogeneity in the thickness. This approach became presented by Martins et al. (2014). Li et al. (2010) studied the shear-
the base for various further applications. For example Barlat (1987) induced fracture in deep drawing processes. They showed existing
used this methodology to investigate the role of anisotropy on models of traditional FLC have no solutions in this case.
formability limits. A similar aim was pursued by Wu et al. (2003), In addition, experiments and modeling of fracture loci under
who extended its use to 3D yield locus descriptions. More recently, various conditions are investigated by different researchers. Per-
Signorelli et al. (2009) coupled M–K model with polycrystalline formed compression-tension fracture experiments by Marcadet
simulations to elicit the effect of microstructure properties on and Mohr (2015) showed that applying a pre-strain increase the
localized necking. In 1994, Swift criterion has been modified by fracture strain. Based on their work, local thickening and transient
Hora and Tong for theoretical prediction of localized necking. Their hardening of the material during compression postpone the local-
approach has been named Modified Maximum Force Criteria (MMFC) ization. Roth and Mohr (2014) showed also increasing the ductility
(Hora and Tong, 1994). This approach has been developed in dif- of the material by increasing the loading speed. Many of these
ferent features during the last years. The influences of hardening empirical methods are representative of compression loadings or
behavior and yield loci on MMFC are discussed by Hora et al. in-plane shear stress which may be of less importance for the deep
(2013). Manopulo et al. (2015) extended MMFC to accommodate drawing of thin sheets. Thus, a new simple test is proposed in this
distortional hardening models. The role of phenomena such as work to extend the fracture line in the shear loading range.
Bauschinger effect, latent hardening and cross-loading contraction Proper definition of fracture limits is important but its exper-
on localization is also studied. imental evaluation and numerical modeling is still quite limited.
However, beside of the typical necking instability there are addi- In the present work, an optical measurement system is used to
tional types of material failure, which define process limits. Those capture the strain history of Nakazima experiments until rupture
are specially surface cracks occurring by e.g. extensive bending or occurrence. For the measurement of the actual fracture strain two
hemming process (Hora et al., 2012). Mattei et al. (2013) looked different methods are introduced: (I) a direct evaluation of the opti-
into strain localization of the aluminum alloy under the bending cal strain measurement and (II) the microscopic measurement of
operation. Plotted apparent rigidity (derivate of an obtained punch ruptured regions of the Nakazima specimens. Gorji et al. (2015)
force) of the material as a function of the displacement in situ showed that latter predicts the fracture strain more accurately.
bending test showed that surface cracks occur at the maximum In this study, prediction of rupture in deep drawing operation by
of this curve i.e., slightly before reaching to maximum punch force. using different fracture criteria is investigated. Numerical analysis
Castany et al. (2013) investigated influence of microstructure on shows that a proper definition of the fracture limit, plays a substan-
bendability of aluminum alloys. They observed bending surface tial role in predicting both the region of the crack in the specimen,
cracks initiate along shear bands at large intermetallic particles. and time of fracture appearance. In the framework of this study, the
Oya et al. (2010) focuses on the experimental investigation of fail- advantage of layered material compared to the monolithic material
ure in multi-layered materials. A similar study has been published is also discussed.
by the same authors which also covers the numerical aspect related
to the modeling of these materials (Yanagimoto et al., 2010). In
bending processes, strain values greatly exceed the FLC before fail- 2. Material characterization
ure occurs. For this reason Schleich et al. (2009) proposed bending
limit fracture instead of using the FLC. Similarly a Forming Limit 2.1. Material hardening
Stress Curve at Fracture (FLSCF) has been proposed by Hu et al.
(2010) based on the Cockroft–Latham fracture criterion in order to Demands on lightweight structural materials with better perfor-
bridge the mentioned gap. In addition, Park and Kim (2003) com- mance on strength and formability, specially in automobile bodies,
pared the FLC in conventional forming and incremental sheet metal are increasing. A new technology for improving the aluminum
forming. It is shown that FLC is unable of describing the failure in sheets deficit has been patented by Anderson et al. (2010) to comply
incremental forming applications i.e., strains exceed the conven- with the conflict of aims between the low weight and the max-
tional FLC without any failure. imum stability. This novel technology is named aluminum Fusion
An appropriate prediction of a crack initiation requires the technology. The underlying principal of this material is retrieved
proper definition of the fracture criterion. Shear crack of fracture from laminated metal matrix composites. Three layers of aluminum
(so-called also pure drawing type failure) is the other kind of fail- are cast simultaneously into one ingot (via conventional direct-chill
ure which has been occasionally seen in press shops because of casting process); consisting a core and two outer layers. Solidified
the tight punch/die radii in stamping process. An important thrust ingot is rolled then as a thin sheet. Correct control of process at the
on studies of the sheet metal forming in this type of failure is the molten metal stage produces a perfect metallurgical bond between
fact that FLC is unable to predict this type of fracture i.e., strain the alloy layers. Monolithic aluminum alloy AA6016 sheet sample
states are beyond of this criterion (Sriram et al., 2003). Thus, gen- with nominal thickness 1.02 mm and multilayer Fusion material
eralized FLC is proposed to overcome the shortage of conventional (AF200) are studied in this work. The investigated Fusion thickness
FLC. Numerous different experiments have been proposed by dif- in the framework of this study is 1 mm; contains about 88% alu-
ferent researchers in metal forming community to determine the minum alloy AA6016 as a core and 6% aluminum alloy AA5005 for
comprehensive fracture locus in the loading range from compres- each outer layer.
sion to equi-biaxial tension. Bao and Wierzbicki (2004) proposed Standard tensile test has been carried out to measure the flow
eleven different experiments including upsetting, shear and tensile curve of the studied aluminum alloy AA6016. Hydraulic bulge test
tests to characterize fracture ductility under different loading con- was also performed in order to characterize the material hardening
ditions. The authors subsequently published a comparative study in a higher range of equivalent plastic strain. The obtained values
of seven different fracture models able of fitting the measured frac- were then used to extend the uniaxial hardening curve using the
ture behavior (Wierzbicki et al., 2005). More recently, Lou and Huh work equivalence. Finally the measured flow curve has been extrap-
(2013) proposed an alternative set of experiments to be used in olated by using the Hockett and Sherby (1975) model (core AA6016
the tensile machine with specimens cut out of thin sheet. Isik et al. and Fusion AF200):
(2014) proposed a new approach to determine the fracture strain, n
exclusively based in sheet metal forming processes. Following ¯ = BHS − (BHS − AHS )e−mε̄ (1)
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 21

Table 1
Hardening parameters of studied aluminum alloys.

AHS BHS m n C1 C2 C3 C4 

AA6016 123.7 352.4 5.62 0.87 – – – – –

AA5005 39.7 123 88.72 1.23 8.9 234.8 0.003 0.34 0.6
AF200 119.2 345.4 5.44 0.862 – – – – –

Fig. 1. Hardening curve of studied materials out of the tensile and hydraulic bulge tests.

Combined approach of Gosh (1977) and Hockett–Sherby functions et al. (2003), m = 8 is recommended for face-centered cubic (FCC)
is more adequate to describe the hardening behavior of the clad materials e.g. aluminum alloys.
material (AA5005): Table 2 lists the mechanical properties of the investigated mate-
rial that are needed to fit the Yld2000-2d parameters (˛1 − ˛8 )
n
¯ = (C1 + C2 (C3 + ε̄)C4 ) + (1 − )(BHS − (BHS − AHS )e−mε̄ ) (2) which are also given in this table i.e., three measured directional
stresses ( 0 ,  45 ,  90 ) and corresponding r-values (r0 , r45 , r90 ) from
¯ and ε̄ represent the yield stress and the plastic strain, respectively. uniaxial stretching test as well as  b and rb from the equi-biaxial
Yield curve parameters of studied material are given in Table 1. hydraulic bulge and stack compression tests. Some researchers like
Flow curve of Fusion material is similar to the hardening of its core. Mutrux et al. (2008) and Güner et al. (2012) are taken rb equal to 1,
Fig. 1 shows elicited hardening curves of studied materials out of which is also applied in this research work.
the tensile and bulge tests as well as their derivatives. Maximum Note, yield stress values in each direction (respect to the rolling
force criterion is properly fitted for monolithic materials (AA6016 direction) are calculated at 4% equivalent strain in order to avoid
and AA5005) and also multilayer Fusion material (AF200). Swift oscillations at the beginning of the experiment. Fig. 2 demonstrates
(1952) presented the criterion for predicting the onset of the diffuse the yield loci of these three studied materials at 4% equivalent
necking with the assumption that plastic instability occurs at a load strain. As it can be seen, the yield function of Fusion material is sim-
maximum for proportional loading. As it can be observed in this ilar to its core (AA6016). Whereas the outer clad material (AA5005)
figure, material behavior of multilayer Fusion material (AF200) is has a much smaller yield locus compared to those two other mate-
similar to its core (AA6016). Maximum difference between them is rials.
around 2%.

2.3. Non-proportional material hardening

2.2. Yield locus
The experimental data at specific deformation is used to eval-
A non-quadratic yield locus model (Yld2000-2d) is employed uate the Yld2000-2d yield function. Thus, considering constant
to describe the plastic behavior of studied alloys. Barlat et al. yield stresses along r-values at different rolling directions lead to
(2003) showed that Yld2000-2d is able of capturing pronounced constant Yld2000-2d parameters (˛1 to ˛8 ) and a constant ratio
anisotropy of aluminum sheet. The implementation of the model  xx / yy , which is set fixed at every level of the deformation. In this
for plane stress states has been thoroughly discussed in Yoon et al. section, a strain dependent Yld2000-2d model is employed to over-
(2004). This yield criterion is a function of the crystal structure expo- come the limitation of the standard yield criterion based on the
nent of the material which is assumed to be a real number between work of Peters et al. (2014). By measuring the yield stresses as a
2 and ∞. The higher the exponent value is, the bigger the curva- function of the equivalent strain, the evolution of the yield locus
ture of the rounded vertices of the yield surface. Based on Barlat can be derived.

Table 2
Mechanical properties and Yld2000-2d parameters of studied aluminum alloys at 4% equivalent strain.

y0  y 45  y 90 yb r0 r45 r90 rb

AA6016 189.26 182.61 184.02 186.69 0.686 0.5 0.666 1.0

AA5005 95.77 90.86 94.06 91.33 0.676 0.329 1.325 1.0
AF200 184.76 176.04 176.74 176.51 0.689 0.429 0.734 1.0

˛1 ˛2 ˛3 ˛4 ˛5 ˛6 ˛7 ˛8

AA6016 0.947 1.017 0.961 1.032 1.021 1.013 0.967 1.153

AA5005 0.8757 1.1421 1.1462 1.0152 1.0330 1.0469 0.9380 1.1879
AF200 0.9201 1.0806 1.0183 1.0508 1.0377 1.0842 0.9686 1.1564
22 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

2.4. Forming Limit Curve

Forming Limit Curve (FLC) is established extensively in industry

to detect failure. In DIN EN ISO 12004-2 (2009), the execution and
evaluation of experimental FLC by doing the Nakazima experiment
is standardized. The work-piece is drawn with a hemispherical
punch with velocity of 1 mm/s until fracture occurs. Different major
to minor strain ratios are traced, ranging from uniaxial to equi-
biaxial tension, by changing the width of the specimens. Digital
Image Correlation (DIC) is also employed to capture strains in situ.
Fig. 4 exhibits that all of the studied materials lead to similar
FLCs based on the standard cross section method. The limit strain of
Fusion material, no matter what method is used, does not improve
by adding soft clad AA5005 to core AA6016.

3. Fracture modeling
Fig. 2. Yld2000-2d loci of studied aluminum alloys at 4% equivalent strain.
3.1. Fracture characterization in Nakazima range

The yield stress in rolling direction ( 0 ) is taken at different lev- The critical fracture strain is determined by using two alterna-
els of equivalent plastic strain ε̄i p . The corresponding plastic work tive techniques. The first one is based on DIC. This allows detection
Wpl is then computed to obtain the other yield stresses ( 45 ,  90 of strain path history from the beginning of the test until rupture.
and  b ) at the same amount of the plastic work of rolling direction: The last detected values before rupture are considered as frac-
 ε̄i p
ture strains. Despite all of its deficiencies, this method is one of
the common ways to determine the fracture strain. As it is shown
Wpl = y (ε̄p )dε̄p (3)
by Gorji et al. (2015), weakness of this method is the fact that
0
especially at the end of the process, the strain rate increases expo-
Experimental observations show that r-values are almost constant nentially. The DIC averages the measured strains of the neighboring
for all of the studied materials thus their variation is not taken points to reduce noise, which undermines the accuracy of the frac-
into account. By combining four yield stresses and four anisotropic ture strain. An alternative method, the so-called thinning method
parameters as a function of plastic deformation, the yield locus evo- has been developed by considering the fracture thickness in this
lution can be derived. Fig. 3 illustrates the evolution of independent study. In this method, the microscopic picture of ruptured region
Yld2000-2d parameters (˛1 − ˛8 ). Deformation of yield surface is of deformed Nakazima specimen is used to determine the fracture
also shown in this figure. Aluminum alloy AA6016 has isotropic thickness (see Fig. 5). Assuming isochoric deformation and under
hardening whereas the yield function of AA5005 gets strongly dis- the plane strain condition, which governs between the localized
torted with increasing the plastic work. necking and rupture (Hora and Tong, 1994), the updated fracture

Fig. 3. Determined Yld2000-2d parameters ˛ as a function of equivalent plastic strain (ε̄p ) and deformation of yield surface.
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 23

3.2. Fracture limit beyond the Nakazima range

A pure drawing type failure (term shear crack broadly use to spec-
ify this fracture) has been occasionally seen in press shops. This
kind of fracture usually occurs on tight radii of tool edges. Besides
that, it is discussed in the previous section that the thinning method
covers the fracture line between uniaxial and equi-biaxial loading
directions. Therefore fracture induced by shearing (in the left hand
side of tensile loading) cannot be predicted by the aforementioned
technique. As it is mentioned in the introduction, great efforts have
been done in order to find appropriate approach to predict fracture
in this region. The goal is to have a simple method to evaluate the
fracture strain due to the shear loading that could be also represen-
tative of the fracture occurring on tight tool radii in real stamping
operations.
Different types of failure due to the shear stress are shortly
explained in the following. As it is explained by Emmens (2011),
shear stress is dominant when forming forces are parallel to the
surface of the material. Fracture by shear stresses in sheet metal
forming is subdivided into two cases: In-plane shear stress and out-
of-plane shear stress. Deformed state of schematic specimen under
these stresses is sketched in Fig. 6.
In the case of the in-plane shear stress, thickness of the specimen
Fig. 4. Determination of FLC according to standard ISO 12004-2, so-called cross
remains constant after the deformation (see Fig. 6-left). In-plane
section method.
shear test is one of the examples which fail under this condition.
In out-of-plane shear or deformation through the thickness condi-
major and minor strain can be defined. This method predicts higher tion, the length of the specimen increases, whereas the thickness
strains compared to the DIC method. decreases. A slant angle through the thickness is shown in Fig. 6-
The thinning method measures much higher fracture strain for right. Emmens (2011) explains also that bending effect is one of the
clad AA5005 due to the small fracture thickness (see Fig. 5). Soft mechanism which creates this type of fracture.
outer material in Fusion (clad) does not carry noticeable load An exemplary cup drawing experiment is illustrated in Fig. 7.
in Nakazima experiment. Microscopic picture of Fusion AF200 Microscopic view of ruptured region shows the slant angle through
through the thickness also proves this assertion. This is the main the thickness. It is shown in this study that in the case of the shear
reason why the fracture line of Fusion and core AA6016 is simi- crack in deep drawing, fracture occurs under the bending condition
lar. In this study, it is elaborated how this soft thin clad material (or out-of-plane shear state).
improves the formability of the material in more complex deep Therefore, introducing an alternative experiment as a represen-
drawing operations. tative of a real process in shear loading region is essential. For this

Fig. 5. Comparing the fracture line of core(AA6016) and clad(AA5005).

24 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

Fig. 6. Two types of fracture forming by shear.

Fig. 7. Cup drawing experiment of AA6016 with tight die radius 3 mm.

purpose, a simple cup deep drawing test with tight radius has been the thickness distribution of the simulation is compared with the
employed to extrapolate the fracture line to the left of the uni- deformed specimen, measured by optical system ATOS (developed
axial loading. The fracture strain is recovered by inverse analysis. by company gom).
Regarding to that and to obtain useful simulation data, FE parame- Fig. 8 illustrates the comparison of draw-in and thickness dis-
ters such as friction and tools’ stiffness were adjusted in a manner tribution between the experiment and simulation of cup drawing
that the simulation data correctly map the experiments. For this test at drawing depth 22.9 mm. Chosen process parameters for the
reason, multiple steps are considered: The draw-in of the exper- experiment are sheet size 140 mm × 140 mm and (200 kN) binder-
iment is compared with the draw-in of the simulation and also force.

Fig. 8. Comparison between the experiment and simulation at drawing depth 22.9 mm.
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 25

Constant parameters Cjc1 = 0.383, Cjc2 = 0.44 and Cjc3 = −2 have been
identified as best fitting with the experimental results out of the
Nakazima and cup drawing test (see Figs. 11 and 12). This model
indicates the dependency of fracture state on hydrostatic pressure.

4.4. Models based on maximum shear stress:

Ductile fracture is predicted with this criterion when the maxi-

f
mum shear stress in the material reaches the critical value (max ).
Maximum shear stress is obtained based on the principal stress  11 ,
 22 and  33 by the following relation:
 −   −   −  
11 22 22 33 11 33
max = max , , (5)
2 2 2
Based on this theory, accumulation of equivalent strain in pure
Fig. 9. Fracture strains determined from Nakazima and cup drawing experiments.
shear loading has the lowest fracture strain compared to uniaxial
compression and uniaxial tension loading. Highest ductility of the
To compare the experimental results with simulations in com-
material based on this approach is predicted in compression, uni-
mercial FE-code AutoForm, a proper simulation setup was chosen
axial tension and equi-biaxial loadings.
to match the experiment at drawing depths 22.9 and 27.4 mm. Lat-
Some proposed fracture models (e.g. Lou and Huh, 2013) predict
ter is the depth when fracture occurs. Maximum relative error of
the minimum equivalent fracture strain in pure shear loading. Teng
draw-in and thickness distribution are 1% and 4% with resulting
and Wierzbicki (2004) proposed to categorize loadings region and
initial parameter friction 0.03 and tools’ stiffness 500 in FE-code
using appropriate function to better fit experimental points. This
AutoForm. This small value of the friction is a consequence of the
approach is applied for the studied monolithic material AA6016, by
well lubricated specimen and tools in the experiment. The adjusted
using Eq. (6). Behavior of this approach is exhibited in Fig. 11 and 12.
simulation is then stopped at the experimental fracture drawing
This fitting is characterized under the assumption that the lowest
depth 27.4 mm to assist finding the correct fracture line. Obtained
accumulated equivalent fracture plastic strain occurs in the pure
fracture strain of aluminum alloy AA6016 with aforementioned cup
f f shear loading range and the highest value is in the compression
drawing test is ε11 = 0.58 and ε22 = −0.36. loading. This method has a better fitting with experimental frac-
Fig. 9 demonstrates the loading paths of Nakazima specimens ture points compared to the original maximum shear stress fracture
captured with the optical measurement system including the mea- f
criterion (max ). Representative models describing this kind of frac-
sured fracture strain with the thinning method. Mean value and
ture types are called in this study as a shear stress based fracture
deviation range of the latter method is also shown for each Nakaz-
criterion.
ima loading path. The fracture point based on cup drawing test is
shown in this figure by a red square point (position of the fracture
in this cup drawing test is demonstrated in Fig. 7). Based on the ε̄f = −10.2183 + 7.85662 − 0.8967 + 0.45; −1/3 <  ≤ 1/3
measurement results, it is shown that defining the fracture model
ε̄f = 18.123 − 20.8942 + 6.8879; 1/3 ≤  ≤ 2/3
linearly (in (ε11 , ε22 ) space) can satisfy the range between shear to (6)
equi-biaxial loading paths.

4.5. Transformation between strain and stress spaces

4. Comparative analysis of fracture loci

4.1. Fracture criteria As it is mentioned in the introduction, both strain and stress
based methods are used in the characterization of fracture loci.
An appropriate prediction of the crack initiation requires the This section aims to discuss a transformation method between the
appropriate definition of a fracture criterion. An important fact on different spaces, in order to facilitate the comparison between dif-
studies of shear failure with tight punch/die radii in stamping pro- ferent approaches in literature and to understand the structure of
cess is the fact that conventional FLC approach is unable to predict the derived fracture loci. Strain and stress states can be investi-
this kind of fracture. Beside of the explained linear fracture crite- gated in the following regions (for the case of isotropic constitutive
rion (linear in (ε11 , ε22 ) space), three other well-known fracture model):
criteria are studied in the remainder of this section.
1. Between uniaxial compression and pure shear loading
4.2. Constant equivalent fracture criterion: (−2 < ˇ < −1 and −1
3
<  < 0)
2. Between pure shear and uniaxial tension loading (−1 < ˇ < −1
2
Equivalent fracture strain is a simple model that postulates duc- and 0 <  < 31 )
tile fracture occurs when the accumulated equivalent strain reaches
3. Between uniaxial tension and plane strain loading ( −1
2
<ˇ<0
a critical value (ε̄f ).
and 31 <  < √1 )
3
4.3. Johnson–Cook: 4. Between plane strain and equi-biaxial loading (0 < ˇ < 1 and
√1 <  < 2 )
3 3
Johnson–Cook (JC) is a well-established fracture model which
is used extensively, especially in the bulk metal forming commu- where ˇ (= ε22 /ε11 ) and  (= H /) ¯ are strain ratio and stress
nity. In case of constant strain rate and temperature, this model triaxiality, respectively.
postulates that equivalent fracture strain (ε̄f ) is a function of stress The classical FLC is defined for tensile loading only and excludes
triaxiality (): the first and second mentioned regions. Corresponding lines in
strain space (ε11 , ε22 ), triaxiality diagram (, ε̄), yield locus ( 11 ,
ε̄f = Cjc1 + Cjc2 · exp(Cjc3 · ) (4)  22 ) are depicted in Fig. 10. These spaces are distinguished in four
26 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

relation between strain and stress ratios. The algorithm used for
this conversion, when the strain history is known, is explained in
Algorithm 1. For linear strain path assumption, calculation is done
in one step. Therefore, by having the relationship between (˛, ˇ),
(˛, f(˛)) and (˛, g(ˇ(˛))), a principal strains tuple (ε11 , ε22 ) can be
converted to equivalent strain/triaxiality pair (ε̄, ). More details
can be found in Gorji et al. (2015).
Algorithm 1. Computation of the triaxiality and Lode parameters
based on the strain history

I Evaluation of the functions f(˛), ˇ(˛) and g(ˇ(˛)):

for ˛ = 0 :1 do
f(˛);
∂ ∂
ˇ(˛) = ∂22
/ ∂ ;
11
g(ˇ(˛));
end for

II Evaluation of  and ¯ for the different strain paths. ε11 and ε22 of strain
path captured from optical measurement system:
for t = 0 : tend do
t
ε11 = t ε11 − t−t ε11
t
ε22 = t ε22 − t−t ε22
t ε
ˇ= t ε
22
11
ε̄ = g(ˇ(˛)) · ε11
t
ε̄ = ε̄
t
t
 = t f (˛) · 1+3 ˛
= −1
t¯ 3
2
· (t
f (˛)) · (2 − t ˛) · (2 · t ˛ − 1) · (1 + t ˛)
end for

4.6. Conversion of fracture models

Conversion of aforementioned fracture criteria in different

spaces is shown in this section. It is worth mentioning that because
of lacking strain histories, these conversions between principal
strains space (ε11 , ε22 ) and stress space (, ε̄) are calculated under
the linear deformation assumption. A summary of the predic-
Fig. 10. Corresponded loading line at different spaces.
tion of explained four fracture models, equivalent fracture strain,
Johnson–Cook, shear stress and linear fracture criterion (linear in
different regions. Correlation of fracture modes between principal (ε11 , ε22 ) space) at two different spaces ((ε11 , ε22 ) and (, ε̄)) is dis-
strains space (ε11 , ε22 ) and stress space (εeq , ) is studied in the played in Figs. 11 and 12. Orange points represent the conventional
following. FLC values. Red points show the measured fracture points out of the
For this purpose, auxiliary functions f (˛) = (11 /) ¯ thinning method and cup drawing test.
and g(ˇ(˛)) = (ε̄/ε11 ) are needed. ˇ = (ε22 /ε11 ) and Equivalent fracture strain is defined as a material constant.
˛ = ( 22 / 11 ) are principal strain and stress ratios, respectively. Equivalent strain ε̄f = 0.56 maps the measured fracture points,
Flow curve and constitutive model play a significant role in the evaluated by thinning method and cup drawing test for aluminum

Fig. 11. Comparison of prediction of different fracture criteria in (ε11 , ε22 ) space for Fig. 12. Comparison of prediction of different fracture criteria in (, ε̄) space for
aluminum alloy AA6016 sheet sample. aluminum alloy AA6016 sheet sample.
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 27

Table 3
Performed triangular deep drawing experiment for both monolithic (AA6016) and
√
Fusion (AF200) material at different drawing depth. represents the experiment
without rupture and × shows that fracture occurs

Die radius 3 mm Die radius 5 mm

Height [mm] 35 40 45 50 55 35 40 45
√ √ √ √ √
AA6016 ×
√ √ √ √ √ √ √ √
Fusion (AF200)

Note that results with a die radius of 5 mm are not critical for
nor the monolithic AA6016 neither Fusion (AF200) materials (see
Table 3).
Exemplary failed monolithic AA6016 specimen is illustrated in
Fig. 15; crack initiates and propagates in the wall of the specimen,
right above the die edge. There is no failure in the case of the mul-
tilayer Fusion material (AF200). Experimental results show Fusion
Fig. 13. Dimensions of the punch; units are in [mm].
material can tolerate the highest possible drawing depth (55 mm)
without any failure.
alloy AA6016 best (see Fig. 12). The constant fracture equivalent
strain and shear stress based fracture criteria exhibit significant dif-
5.2. FEM based analysis of the fracture phenomenon
ferences with the other methods especially in the region of shear
and compression loadings. Both Johnson–Cook and linear fracture
The FE analysis of the triangular shape deep drawing is done
model consider more accumulation of equivalent strain in the zone
with the dynamic explicit FE-code LS-DYNA. Tools are modeled as
where pressure increases. Latter approach predicts the fracture
rigid bodies. In addition, in order to predict the localization and
more accurately in the range between the uniaxial and equi-biaxial
crack propagation more accurately, the blank is modeled with a rel-
tension.
atively fine element size of 0.5 mm. Total element number is over
300,000 and punch stroke is defined as 75 mm/s. As it is also dis-
5. Validation of the fracture models in a real deep drawing
cussed in Section 3.2, the first step to obtain useful simulation is
operation
to adjust the simulation parameters, such as friction, in a manner
5.1. New triangular deep drawing shape that the simulation results are comparable with the experiments.
To achieve this, the deviation between the draw-in of real part and
The new designed triangle deep drawing shape is used in the simulation is minimized by adjusting the friction coefficient.
framework of this study to validate the constitutive model and After defining the unknown parameters, principal strains dis-
compare the behavior of the Fusion material with the monolithic tribution are compared with the experiment to validate the
material. Two die radii 3 mm and 5 mm are designed and produced constitutive model. Fig. 16 illustrates the major strain difference
for this operation. Drawing of the punch is sketched in Fig. 13. It and minor strain difference between the experiment and simulation
can be seen that the punch is characterized by three rounded cor- for aluminum alloy AA6016 at drawing depth 35 mm in the case
ners with different radii. These special convex–concave shape of of the die radius 3 mm. There is a good agreement between the
the punch produces different strain states especially in the wall experimental measurement and numerical results. Used constitu-
of the deformed blank. The basic goal of the development of this tive model in this simulation is modified Yld2000-2d yield function
tool is to reach high strains in shear deformation range for further dependent on equivalent strain with crystal structure exponent
investigation of the material behavior. m = 8 (see details in Section 2.3). As it can be seen in Fig. 16, this
Multiple variants of the triangular shape deep drawing process model is not able of predicting the strains in the compressed region.
were accomplished which can be found in Table 3. Deformed spec- Employing the thick shell element can mitigate the inaccuracy of
imens were sent then to company gom for the evaluation of strains. the FE results in this region. Note that the standard Yld2000-2d
DIC system ARGUS (developed by company gom) is used to evalu- model delivers similar output. This is due to the isotropic behavior
ate the strain distribution of the studied materials (see Fig. 14). As of the studied material. Fig. 3 illustrates that in case of monolithic
it is depicted in this figure, bulk of the strains is placed between aluminum alloy AA6016, yield surface does not significantly deform
uniaxial and pure shear loadings. during the deformation.

Fig. 14. Evaluated strain values of AA6016 sheet sample at drawing depth 35 mm, using optical measurement system ARGUS (no fracture).
28 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

Fig. 15. [Left] Exemplary failed monolithic AA6016 (fracture occurs between the drawing depth 40 and 45 mm) and [Right] Multilayer Fusion material at drawing depth
45 mm.

5.3. Comparison of different fracture models phenomenon more accurately, it is necessary to wipe out each inte-
gration point as soon as it reaches the specified limit. To fulfill
It is mentioned that conventional FLC is inefficient in predicting this condition, the stiffness of each integration point through the
the failure for the processes which feature strains in the shear load- thickness has been set to zero when it is reached the fracture limit.
ing range. An appropriate prediction of the crack initiation requires The resulting experimental strain states of monolithic alu-
the adequate definition of a fracture criterion. In the remainder minum alloys AA6016 in the (ε11 , ε22 ) space with die radius 3 mm
of this study, it is shown how the previously discussed fracture and drawing depth 40 mm is shown in Fig. 17. As it can be seen here,
approaches (Sections 3.2 and 4.1) can be applied to real deep most of the strains are placed between the shear and the uniaxial
drawing applications. It is shown that fracture limit plays a substan- stretching loadings in the left hand side of the FLC. The most inter-
tial role in predicting the region of the crack and time of fracture esting point in the fracture analysis of this geometry is that there
appearance. Triangular deep drawing process with tight die radius exist two competing regions where fracture may occur. Although
(3 mm) is studied by using the modified Yld2000-2d constitutive a bigger absolute strain value is observed in ZoneII, fracture occurs
model. at ZoneI. The way different models react to this phenomenon is
Shell element deletion is employed to eliminate each element discussed in the following.
which encroaches the fracture limit. In most of the FE software e.g.
LS-DYNA, an element is deleted when all or a pre-defined percent- 5.3.1. Constant equivalent strain fracture criterion
age of the integration points through the thickness satisfy the frac- Fig. 18 shows the strain distribution of lower layer (punch side
ture criterion. To overcome this defect and to predict the fracture of blank) at drawing depth 30.6 mm when the first group of strains

Fig. 16. Strain difference of aluminum alloy AA6016 between the experiment (35 mm depth, 3 mm die-radius) and simulation.
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 29

Fig. 17. Experimental strain states of AA6016 sheet sample with die radius 3 mm and drawing depth 40 mm, using optical measurement system. Fracture occurs at ZoneI.
Note, this sample has no cracks and white points on the specimen could not be detected by DIC.

propagation is correct by applying this fracture approach. This

criterion might be helpful for some less complex deep drawing
processes (e.g. cup drawing test), but this triangular deep draw-
ing example shows that using the equivalent fracture strain is not
an appropriate approach to predict the crack initiation.

5.3.2. Johnson–Cook Criterion

This fracture model predicts the fracture zone correctly. Fig. 19
demonstrates the strain distribution of the upper layer (die side
of blank) and lower layer (punch side of blank) for the triangular
deep drawing process at drawing depth 43.2 mm. As it is mentioned
before, fracture in the experiment occurs at a height of 40 mm
which is comparable with FE results. Based on this model both
Zone I and Zone II in Fig. 19 are in the critical region and close to
the failure limit. Fracture occurs finally in Zone I which agrees with
experimental observations.

Fig. 18. Strain distribution of lower layer for the triangular deep drawing process at 5.3.3. Shear stress based failure criterion
drawing depth 30.6 mm; Zone II passes the fracture line earlier in the case of constant FE-results elaborate that using the maximum shear stress or ana-
equivalent strain fracture criterion ε̄f = 0.56 (in the reality fracture occurs at Zone I). lytical fitting of the fracture points in a manner that accumulation
of plastic strain has the lowest value in pure shear domain predicts
cross the equivalent fracture strain ε̄f = 0.56 (as it is shown in the fracture wrong, both the position of crack initiation and depth
Fig. 12, this value had a best fit with experimental fracture points). of the fracture.
Neither the fracture depth (fracture occurs at drawing depth Elements in Zone II (see Fig. 18) reach to limit before the ones
between 40 and 45 mm in the experiment), nor the place of crack in Zone I. Fig. 20 illustrates the strain states of lower layer (punch

Fig. 19. Strain distribution of lower and upper layer for the triangular deep drawing process at drawing depth 43.2 mm; Zone I passes the fracture line earlier in the case of
Johnson–Cook fracture criterion (in the reality fracture occurs also at Zone I).
30 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

Fig. 20. Strain distribution of lower layer for the triangular deep drawing process at drawing depth 30.6 mm; Zone II passes the fracture line earlier in the case of shear stress
based fracture criterion (in the reality fracture occurs at Zone I).

side of blank) of triangular deep drawing operation at depth around performed experiments, fracture does not happen at drawing depth
30 mm. Strains in this layer cross fracture limit earlier than the ones 40 mm. In the simulation, it is seen that large part of the elements
in the upper layer (die side of blank). Predicted rupture position is are beneath the fracture line. Few elements of the upper layer (2 or
not compatible with experimental observations. 3) pass this limit line which could be due to the numerical outliers.
And finally in 45 mm, numerous elements are passed the fracture
5.3.4. Linear fracture criterion line.
The global principal strains of the integration points located in Simulation results for smaller die-radius triangular deep draw-
bottom and top layers in drawing depths 40 and 45 mm are plot- ing configuration by considering shell element deletion criterion
ted in Fig. 21. The fracture line in this figure (red line) is derived is depicted in Fig. 22. Numerical prediction of the fracture is illus-
by combination of thinning method and cup drawing test. In the trated at a drawing depth of 40.0 mm.

Fig. 21. Strain distribution of different layers for triangular deep drawing experiment of AA6016 at height [Up] 40 mm, [Down] 45 mm with die radius 3 mm by considering
modified Yld2000-2d.
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 31

Fig. 24. Illustration of material description of the Fusion material in LS-DYNA using
[Left] *ELEMENT SHELL, [Right] *PART COMPOSITE.

whereas no fracture occurs in reality. Summary of discussed frac-

Fig. 22. Simulated deep drawing experiment of aluminum alloy AA6016. ture models are summarized in Table 5

Table 4 5.4. Fusion material

Critical depth and crack positioning of different fracture models.

Fracture height Initiation of rupture The strength capability of Fusion material is obtained with the
[mm] alloy used in the core; while the alloy used for the two outer layers
provides the additional bendability and corrosion resistance.
In order to adequately model the layered structure of the Fusion
Experiment 40–45 mm Zone I
material, the *PART COMPOSITE functionality of LS-DYNA has been
employed instead of regular shell element (see Fig. 24). Latter case
ε̄f = 0.56 31 mm Zone II × requires the material properties of homogenized Fusion as a single
√
Johnson–Cook (JC) 43 mm Zone I material whereas in the composite shell element, the mechanical
√
Linear fracture 40 mm Zone I
Shear stress based 31 mm Zone II ×
properties and thickness distribution of each layer can be described
separately.
Eleven integration points (11 IPs) through the thickness of the
As it can be seen in this figure, location and propagation of the composite shell elements are employed to define the Fusion mate-
crack, predicted by FE simulation has a good agreement with real- rial. 1 IP for each clad outer layer with the thickness of 0.06 mm
ity (see Fig. 15). In addition, the fracture drawing depth is also and 9 IPs for the core with thickness of 0.0978 mm for each IP with
comparable with the experiment. The fracture occurs at a depth total core thickness of 0.88 mm. Defined integration points of core
between 40 and 45 mm. In the simulation crack initiates at a height and clad have their own material properties i.e. hardening curve
approximately 40 mm which is a little bit earlier than in reality. (see Table 1), modified Yld2000-2d yield function (see Fig. 3) and
Critical depths and place of crack initiation based on the inves- fracture limit (see Fig. 5).
tigated models are listed in Table 4. Comparison of experimental output and numerical prediction
of the Fusion material in deep drawing operation at drawing depth
5.3.5. Fracture phenomenon in larger die radius 55 mm is exhibited in Fig. 25. This is the maximum drawing depth
Die radius 5 mm experiments are summarized in Table 3. Trian- which has been performed. As it can be seen in this figure, Fusion
gular deep drawing shape with this die radius does not fail (neither material can undergo combined stretching and bending without
AA6016 nor Fusion). Fig. 23 plots the strains distribution in (ε11 , occurrence of failure, which has been predicted with numerical
ε22 ) space for different layers. All of the elements are under the analysis, properly.
fracture limit. Fig. 26 elaborates the key advantage of Fusion material com-
Strain distribution captured with optical system and also pared to the monolithic material clearly. The aluminum alloy
numerical calculated strains prove that constant equivalent strain AA5005 used at outer layers provides the additional formability
and shear stress based fracture criteria are inadequate to analyze in shear loading range. High fracture strain of clad material does
the fracture in this tools setup. Both models predict the rupture not allow initiation and therefore propagation of the surface cracks

Fig. 23. Strains distribution in (ε11 , ε22 ) space for triangular deep drawing experiment of AA6016 at height 45 mm with die radius 5 mm by considering modified Yld2000-2d
constitutive model for lower and upper layer.
32 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33

Table 5
Prediction of fracture in the case of die radius 5 mm.

Experiment Die radius 5 mm ε̄f = 0.56 Johnson–Cook (JC) Shear stress based Linear fracture line

√ √
No failure × ×

Fig. 25. Plastic strain of simulated triangular deep drawing experiment of Fusion material at drawing depth 55 mm for different layers.

Fig. 26. Strain distribution in (ε11 , ε22 ) space for triangular deep drawing experiment of Fusion material at height 55 mm with die radius 3 mm by considering modified
Yld2000 constitutive model (the fracture limit for the clad material is significantly higher than the core and thus uncritical. In order to keep the figure scales constant, these
limits were not plotted).

in the material. The Fusion material can tolerate such a large defor- that considering the equivalent strain and shear stress based as a
mation without rupture because the strain state at its core is below fracture criterion is not proper to define the crack initiation in such a
its fracture limit. Whereas in monolithic AA6016 material, surface complex deep drawing processes. The linear fracture criterion has a
cracks created in upper layer propagate through the thickness (see better agreement with experimental results in the considered tri-
Fig. 21). axiality ranges, especially in the equi-biaxial loading. Both linear
fracture and Johnson–Cook models deliver similar results for the
triangular shape deep drawing process. It is also shown that cre-
6. Conclusions
ated surface crack in shear loading region has been ceased by using
the soft clad material in multilayer Fusion material.
Firstly, thinning method (measuring the fracture thickness in
Nakazima test) is used to obtain the fracture strains of the material
in the range of uniaxial until equi-biaxial loading. In deep drawing, Acknowledgments
as soon as the material crosses from the flange towards the wall the
state will switch from shear towards plane strain and failure will The authors are very grateful to the CTI (The Swiss Innovation
happen in out-of-plane shear (see Fig. 7). Therefore, the cup draw- Promotion Agency) for the financial support of this work within the
ing experiment is proposed which is a closer representative of the project 13082.1 PFIW-IW. Moreover, thanks Dr. Jürgen Timm and
kind of shear fracture experienced in deep drawing as conventional Dr. Etienne Combaz (team of Novelis) to provide the materials and
in-plane shear tests. As far as the identification of the failure strain thanks Dr. Ernst Müller (gom GmbH) for assisting with the evalu-
is concerned, the approach adopted is phenomenological. In fact, ation of the optical measurements. Finally, the authors very much
an inverse analysis is carried out to properly match the simulated appreciate the support by Dr. Mike Selig (AutoForm Development
strain distribution to the corresponding measured thickness distri- GmbH) and Mr. Klaus Wiegand (Daimler AG) the other industrial
bution at the fracture depth. This enables better fitting the linear partners of this CTI project.
fracture model on the left hand side of the FLC which otherwise
would be unable of predicting the fracture there. Fracture concept References
is also investigated using four different models: Constant equiv-
M.D. Anderson, K.T. Kubo, T.F. Bischoff, W.J. Fenton, E.W. Reeves, B. Spendlove,
alent fracture strain, Johnson–Cook, shear stress based and linear R.B. Wagstaff, Method for casting composite ingot, October 26, 2010. US Patent
fracture criterion (linear in (ε11 , ε22 ) space). It has been displayed 7,819,170.
 M. Gorji et al. / Journal of Materials Processing Technology 232 (2016) 19–33 33

Bao, Y., Wierzbicki, T., 2004. On fracture locus in the equivalent strain and stress Manopulo, Niko, Hora, Pavel, Peters, Philip, Gorji, Maysam, Barlat, Frédéric, 2015.
triaxiality space. Int. J. Mech. Sci. 46, 81–98. An extended modified maximum force criterion for the prediction of localized
Barlat, F., 1987. Crystallographic texture, anisotropic yield surfaces and forming necking under non-proportional loading. Int. J. Plast.
limits of sheet metals. Mater. Sci. Eng. 91, 55–72. Marcadet, S.J., Mohr, D., 2015. Effect of compression-tension loading reversal on the
Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Choi, S.H., Pourboghrat, F., Chu, strain to fracture of dual phase steel sheets. Int. J. Plast.
E., Lege, D.J., 2003. Plane stress yield function for aluminum alloy sheets – Part Marciniak, Z., 1965. Stability of plastic shells under tension with kinematic boundary
I: Theory. Int. J. Plast. 19, 1297–1319. condition. Archiwum Mechaniki Stosorwanej 17, 577–592.
Brozzo, P., Deluca, B., Rendina, R., 1972. A new method for the prediction of forma- Marciniak, Z., Kuczynski, K., 1967. Limit strains in the process of stretch-forming
bility in metal sheet, sheet metal forming and formability. In: In: Proceedings of sheet metal. Int. J. Mech. Sci. 9, 609–620.
the 7th Biennial Conference of the IDDRG, pp. 161–166. Martins, P.A.F., Bay, N., Tekkaya, A.E., Atkins, A.G., 2014. Characterization of fracture
Castany, P., Diologent, F., Rossoll, A., Despois, J.F., Bezencon, C., Mortensen, A., 2013. loci in metal forming. Int. J. Mech. Sci. 83, 112–123.
Influence of quench rate and microstructure on bendability of aa6016 aluminum Mattei, L., Daniel, D., Guiglionda, G., Klöcker, H., Driver, J., 2013. Strain localization
alloys. Mater. Sci. Eng.: A 559, 558–565. and damage mechanisms during bending of AA6016 sheet. Mater. Sci. Eng.: A
DIN EN ISO 12004-2. Metallic materials – sheet and strip – determination of forming- 559, 812–821.
limit curves – Part 2: Determination of forming-limit curves in the laboratory, Mutrux, A., Hochholdinger, B., Hora, P., 2008. A procedure for the evaluation and vali-
2009. dation of the hydraulic biaxial experiment. In: In: Proceedings of the NUMISHEET
Emmens, W.C., 2011. Formability: A Review of Parameters and Processes that Con- Conference, Interlaken, Switzerland, pp. 67–71.
trol, Limit or Enhance the Formability of Sheet Metal. Springer Science & Business Nakazima, K., Kikuuma, T., Hasuka, K., 1971. Study on the formability of steel sheets.
Media. In: Yawata Technical Report (284)., pp. 678–680.
Goodwin, G.M., 1968. Application of strain analysis to sheet metal forming problems Olsen, T., 1920. Machines for ductility testing. Proc. Am. Soc. Test. Mater., 398–403.
in the press shop. Soc. Autom. Eng. (680093), 380–387. Oya, T., Tiesler, N., Kawanishi, S., Yanagimoto, J., Koseki, T., 2010. Experimental and
Gorji, M., Berisha, B., Hora, P., Barlat, F., 2015. Modeling of localization and fracture numerical analysis of multilayered steel sheets upon bending. J. Mater. Process.
phenomena in strain and stress space for sheet metal forming. Int. J. Mater. Technol. 210 (14), 1926–1933.
Form., 1–12. Park, Jong-Jin, Kim, Yung-Ho, 2003. Fundamental studies on the incremental sheet
Gosh, A.K., 1977. Tensile instability and necking in materials with strain hardening metal forming technique. J. Mater. Process. Technol. 140 (1), 447–453.
and strain-rate hardening. Acta Metall. 25, 1413–1424. Peters, P., Manopulo, N., Lange, C., Hora, P., 2014. A strain rate dependent anisotropic
Güner, A., Soyarslan, C., Brosius, A., Tekkaya, A.E., 2012. Characterization of hardening model and its validation through deep drawing experiments. Int. J.
anisotropy of sheet metals employing inhomogeneous strain fields for yld2000- Mater. Form. 7 (4), 447–457.
2d yield function. Int. J. Solids Struct. 49 (25), 3517–3527. Roth, C.C., Mohr, D., 2014. Effect of strain rate on ductile fracture initiation in
Hecker, S.S., 1972. A simple forming limit curve technique and results on aluminum advanced high strength steel sheets: experiments and modeling. Int. J. Plast.
alloys. In: In: IDDRG1972, Amsterdam, Netherland, pp. 51–58. 56, 19–44.
Hill, R., 1952. On discontinous plastic states with special reference to localized neck- Schleich, R., Sindel, M., Liewald, M., 2009. Investigation on the effect of curvature
ing in thin sheets. J. Mech. Phys. Solids 1, 19–30. on forming limit prediction for aluminium sheet alloys. Int. J. Mater. Form. 2,
Hockett, J.E., Sherby, O.D., 1975. Large strain deformation of polycrystalline metals 69–74.
at low homologous temperatures. J. Mech. Phys. Solids 23-2, 87–98. Signorelli, J., Bertinetti, M., Turner, P., 2009. Predictions of forming limit diagrams
Hora, P., Berisha, B., Gorji, M., Manopulo, N., 2012. A generalized approach for the using a rate-dependent polycrystal self-consistent plasticity model. Mater. Sci.
prediction of necking and rupture phenomena in the sheet metal forming. In: Eng. 25, 1–25.
In: IDDRG2012, Mumbai, India, pp. 79–93. Sriram, S., Wong, C., Huang, M., Yan, B., 2003. Stretch-bendability of advanced high
Hora, P., Tong, L., 1994. Prediction methods for ductile sheet metal failure using strength steels. In: In: Proceedings of SAE 2003 World Congress, Detroit, MI,
FE-simulation. In: In: Proceedings of the IDDRG, Lisbon, Portugal, pp. 363–375. USA.
Hora, P., Tong, L., Berisha, B., 2013. Modified maximum force criterion, a model for Swift, H.W., 1952. Plastic instability under plane stress. J. Mech. Phys. Solids 1, 1–18.
the theoretical prediction of forming limit curves. Int. J. Mater. Form. 6, 267–279. Teng, X., Wierzbicki, T., 2004. Effect of fracture criteria on high velocity perforation
Hu, X., Lin, Z., Li, S., Zhao, Y., 2010. Fracture limit prediction for roller hemming of of thin beams. Int. J. Comput. Methods 1 (01), 171–200.
aluminum alloy sheet. Mater. Des. 31 (3), 1410–1416. Wierzbicki, T., Bao, Y., Lee, Y.W., Bai, Y., 2005. Calibration and evaluation of seven
Isik, K., Silva, M.B., Tekkaya, A.E., Martins, P.A.F., 2014. Formability limits by fracture fracture models. Int. J. Mech. Sci. 47, 719–743.
in sheet metal forming. J. Mater. Process. Technol. 214 (8), 1557–1565. Wu, P.D., Jain, M., Savoie, J., MacEwen, S.R., Tugcu, P., Neale, K.W., 2003. Evaluation
Keeler, S.P., 1961. Plastic instability and fracture in sheet stretched over rigid of anisotropic yield functions for aluminum sheets. Int. J. Plast. 19, 121–138.
punches. (Ph.D. thesis). Massachusetts Institute of Technology, Boston. Yanagimoto, J., Oya, T., Kawanishi, S., Tiesler, N., Koseki, T., 2010. Enhancement of
Li, Y., Luo, M., Gerlach, J., Wierzbicki, T., 2010. Prediction of shear-induced fracture bending formability of brittle sheet metal in multilayer metallic sheets. CIRP
in sheet metal forming. J. Mater. Process. Technol. 210 (14), 1858–1869. Ann.-Manuf. Technol. 59 (1), 287–290.
Lou, Y., Huh, H., 2013. Prediction of ductile fracture for advanced high strength steel Yoon, J.W., Barlat, F., Dick, R.E., Chung, K., Kang, T.J., 2004. Plane stress yield function
with a new criterion: experiments and simulation. J. Mater. Process. Technol. for aluminum alloy sheets – Part II: FE formulation and its implementation. Int.
213, 1284–1302. J. Plast. 20, 495–522.