You are on page 1of 10

Journal of Materials Processing Technology 234 (2016) 249–258

Contents lists available at ScienceDirect

Journal of Materials Processing Technology


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

Fracture toughness and failure limits in sheet metal forming


M.B. Silva a , K. Isik b , A.E. Tekkaya b , A.G. Atkins c , P.A.F. Martins a,∗
a
IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1049-001 Lisboa, Portugal
b
Institute of Forming Technology and Lightweight Construction, Technical University of Dortmund, Baroper Str. 303, D-44227 Dortmund, Germany
c
Department of Engineering, University of Reading Box 225, Reading RG6 6AY, UK

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

Article history: This paper proposes a link between plastic flow, void coalescence and growth, ductile damage, crack
Received 30 April 2015 opening modes and fracture toughness in sheet metal forming. This new integrated view is based on
Received in revised form 1 September 2015 an analytical framework that allows estimating the location of the fracture loci in the principal strain
Accepted 31 March 2016
space directly from material stress–strain response and from fracture toughness and thickness at fracture
Available online 1 April 2016
obtained from double-notched test specimens loaded in tension and plane torsion (in-plane shear). Exper-
iments in AA1050-H111 aluminium sheets with 1 mm thickness give support to the proposed analytical
Keywords:
framework.
Sheet metal forming
© 2016 Elsevier B.V. All rights reserved.
Fracture loci
Fracture toughness
Crack opening mode

1. Introduction in order to distinguish the circumstances under which different


processes fail by fracture (Isik et al., 2014).
Until recently, the formability limits by fracture have not been The commonly accepted fact that FLC’s are not material prop-
of interest to sheet metal forming because once a neck appears erties and that its determination is greatly influenced by strain
and spreads sideways under subsequent deformation, thinning will loading paths, by combination of in-plane loading and bend-
progress very fast under decreasing loads or pressures until the ing effects and by difficulties in measuring the onset of necking
sheet cracks. As a result of this, research has been focused on the (Centeno et al., 2014) further contribute to the growing interest in
formability limits at the onset of plastic instability (also known as the formability limits by fracture.
the forming limit curves, FLC’s). Nowadays, the experimental meth- The utilization of ductile damage mechanics for predicting the
ods and procedures for determining the FLC’s of metal sheets at onset of failure by fracture has a long pedigree and can be system-
room temperature are well established in the international stan- atized as a function of the associated theoretical background into
dard ISO 12004-2 (ISO, 2008) and involve carrying out Nakazima two different categories: (i) uncoupled procedures based on the
and Marciniak sheet formability tests. utilization of simple ductile damage criteria that are weighted inte-
The widespread utilization of finite element analysis in sheet grations of the effective plastic strain (Atkins, 1996) and (ii) coupled
metal forming relaunched the discussion on the utilization of duc- procedures based on micro-based damage mechanics (Tvergaard
tile damage mechanics for predicting the onset of failure by fracture and Needleman, 1984) built upon the macroscopic yield surface for
and on the experimental methods and procedures for determining porous materials (Gurson, 1977), or based on continuum damage
the fracture loci in the principal strain space and in the space of mechanics (Lemaitre, 1985).
effective strain vs. stress triaxiality. Some authors combine data Uncoupled procedures based on the utilization of ductile dam-
retrieved from sheet and bulk formability tests (Wierzbicki et al., age criteria due to Cockcroft–Latham, McClintock and Oyane,
2005) while others consider that the differences in plastic flow among others, do not account for the progressive deterioration
resulting from the plane stress conditions of sheet metal forming of the material during loading and unloading but are easier to
and the three dimensional stress conditions of bulk metal forming implement and to calibrate than fully coupled procedures. Recent
that are commonly used as a rationale to classify metal forming developments by Bai and Wierzbicki, (2010) and Li et al. (2010)
processes into two-different groups must be treated differently, allowed determining the fracture loci in both the principal strain
space and the space of effective strain vs. stress triaxiality by means
of new uncoupled ductile damage procedures that combine exper-
∗ Corresponding author. imentation in bulk and sheet metal formability tests and finite
E-mail address: pmartins@ist.utl.pt (P.A.F. Martins). element inverse calibration procedures.

http://dx.doi.org/10.1016/j.jmatprotec.2016.03.029
0924-0136/© 2016 Elsevier B.V. All rights reserved.
250 M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258

Fig. 1. Failure by fracture in tension. (a) Schematic representation of void growth and (b) fracture forming limit line (FFL) in the principal strain space.

This paper is built upon uncoupled ductile damage proce- 2. Theoretical background
dures but contrary to other publications in the field it is aimed at
establishing a link between void coalescence and growth, ductile 2.1. Ductile damage, void coalescence and growth
damage, crack opening modes and fracture toughness. The work
draws from two recently published works by Isik et al. (2014) and Crack propagation by void coalescence and growth in tension or
Martins et al. (2014) on the circumstances under which each crack shear stress fields may always be viewed as a process of continu-
opening mode will occur in terms of plastic flow and microstruc- ous re-initiation along the path of cracking. In tension, Atkins and
tural ductile damage and it is aimed at proposing an analytical Mai (1985) worked on McClintock’s (1968) continuum mechanics
framework to characterize fracture loci under plane stress con- of void growth to established a relation between the inter-hole l
ditions directly from material stress-strain response and from (inter-particle/inclusion) spacing, the diameter d of the hole (par-
fracture toughness and thickness at fracture obtained from exper- ticle/inclusion) and the stress triaxiality m / (defined as the ratio
iments with double-notched test specimens loaded in tension and of the average and the effective stress) at the onset of cracking
plane torsion, which fail by fracture in crack opening modes I and (Fig. 1a),
II of fracture mechanics. The fracture loci determined by means of
the new proposed analytical framework are checked against exper-
imental strain pairs at failure that were previously obtained by the
l εf
authors in plastic deformation scenarios that are distinct from the m
ln = dε (1)
fracture mechanics calibration test cases. d 
0

Fig. 2. Failure by fracture in shear. (a) Schematic representation of void growth and (b) in-plane shear fracture forming limit line (SFFL) in the principal strain space.
M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258 251

Fig. 3. Double notched test specimen under tension.


(a) Schematic representation of the volumes of the necked down region at the crack tip in which the cracks run during the applied tension.
(b) Schematic representation of the strain loading path in the principal strain space.

where the right hand side term is a simplified version of McClintock effective stress can be utilized to model the distortional changes in
(1968) ductile damage criterion. Eq. (1) allows concluding that the voids,
I
critical value of damage Dcrit calculated from of McClintock’s (1968)
criterion at the onset of fracture is related to crack opening by εf
II 
mode I of fracture mechanics because stress triaxiality is related Dcrit = dε (5)

to dilatation changes in voids rather than distortional changes,
0

II denotes the critical value of damage at the onset


εf The symbol Dcrit
I m of fracture by in-plane shear (crack opening mode II of fracture
Dcrit = dε (2)
 mechanics).
0 The derivation of Eq. (5) provides theoretical support to the
empirical ductile fracture criterion that was utilized by Isik et al.
In case of pure shear m / = 0, McClintock et al. (1966) (2014) and Martins et al. (2014) to characterize the fracture locus
proposed the following relation between the inter-hole l by in-plane shear in the principal strain space and in the space of
(inter-particle/inclusion) spacing, the diameter d of the hole (par- effective strain vs. stress triaxiality.
ticle/inclusion) and the shear strain  at the onset of cracking
(Fig. 2a), 2.2. Ductile damage and fracture loci
l 
ln = 1 + 2 (3) Martins et al. (2014) recently showed that by using the constitu-
d tive equations associated to Hill (1948) anisotropic yield criterion
and assuming plane stress loading conditions and rotational sym-
Contrary to Eq. (1) that is presented in integral form and, there-
metry anisotropy r˛ = r = r, where r is the normal anisotropy, it is
fore, can be readily taken as a measure of the accumulated void
possible to rewrite Eqs. (2) and (5) as a function of the major and
growth damage in tension, Eq. (3) is already the result of integra-
minor in-plane strains (ε1f , ε2f ) at the onset of fracture,
tion. This implies that Eq. (3) cannot be directly related to damage
in shear.
εf 
ε1f
 
However, a connection between
 l/d and ductile damage in shear I m (1 + r) ˇ+1 (1 + r) 
Dcrit = dε = dε1 = ε1f + ε2f (6)
can be achieved if the term 1 +  2 in the right hand side of Eq.  3 ˇ 3
(3) is approximated by /3 for the typical working range of shear 0 0
strains  < 3. Under these circumstances, the following integral
εf ε1f  
form of Eq. (3) can be built upon application of the Levy-Mises
 1 (1 + r) ˇ−1 1 (1 + r) 
constitutive equations in pure shear d = 3(/)dε,
II
Dcrit = dε = dε1 = ε1f − ε2f (7)
 2 (1 + 2r) ˇ 2 (1 + 2r)
0 0

l f εf The integrands in Eqs. (6) and (7) have the form (A + B/ˇ)
1 
ln ≈ d = dε (4) implying that the damage functions for a constant strain ratio
d 3 
0 0 ˇ = dε1 /dε2 , are independent of the loading path history. In other
words, the integrated values of Eqs. (6) and (7) are only dependent
The above integral form of Eq. (3) allows concluding that a duc- on the initial and final values of strain at the onset of fracture. This
tile damage criterion based on the ratio of the shear stress to the occurrence is comprehensively discussed by Atkins and Mai (1985)
252 M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258

loaded in tension (Fig. 3a), for example, the necked down process
region is located in-between the notches and is assumed to have
a height h < d, which should be similar to the thickness t of the
specimens (Hill, 1952).
Under these circumstances, the incremental work RdA dissi-
pated within an incremental volume hdA during crack nucleation
(or propagation) corresponds to the plastic work per unit of vol-
ume at the onset of fracture (please refer to the volumes that are
schematically shown in the detail of Fig. 3a),

εf
RdA ∼ R
= = dε (10)
hdA t
0

In a strain loading path that consists of two components (before


and after necking) the above equation may be written as,


εneck εf
R
= dε + dε (11)
t
Fig. 4. Forming limits by necking and fracture in the principal strain space of the 0 εneck
aluminium AA1050-H111 sheets with 1 mm thickness.
where εneck and εf are the effective strains at neck formation (FLC)
and justifies the reason why strain loading paths in Figs. 1b and 2b and fracture (refer to points ‘A’ and ‘B’ of Fig. 3b).
n
were assumed as linear. In case of a material with a stress-strain behaviour  = Kε , the
Moreover, it also follows from Eq. (6) that the critical value right hand side terms of Eqs. (10) and (11) corresponding to the
of damage DcritI associated with stress triaxiality and dilatational strain loading paths OB and OAB provide the same result,
changes in voids defines a straight line with slope ‘−1’ falling from n+1
R Kεf
left to right in close agreement with the FFL (fracture forming line, = (12)
t n+1
 condition
Fig. 1b) and the of critical reduction of thickness at frac-
ture ε3f = ln 1 − Rf , where Rf given by (t0 − tf )/t0 with t0 and tf The path independence result of Eq. (12) is understandable because
being the initial thickness and the thickness at fracture. Conversely, the integrand of the critical damage Dcrit I associated with crack
II asso-
Eq. (7) allows concluding that the critical value of damage Dcrit opening mode I (Eq. (6)) has the form (A + B/ˇ) and, therefore, it is
ciated with in-plane shear and distortional changes in voids defines also independent from the loading path history.
a straight line rising from left to right with a slope equal to ‘+1’ in The above analytical procedure and the resulting Eq. (12) can
agreement with the condition of critical distortion f along the SFFL also be applied to double notched circular test specimens loaded
(shear fracture forming line, Fig. 2b). in plane torsion because the critical damage DcritII associated with
In connection to what was said above about the fracture loci, it is crack opening mode II (Eq. (7)) has also the form (A + B/ˇ). The only
worth noting that if the lower limits of the integrals in Eqs. (6) and important difference that needs to be taken into consideration is the
(7) are equal to ε0 rather than zero, corresponding to situations utilization of different values of fracture toughness R for tension
where there is a threshold strain ε0 below which damage is not and shear loading conditions because the energy to nucleate and
accumulated, the FFL and the SFFL deviate from straight lines and propagate cracks should vary as a function of the crack opening
present ‘upward curvatures’ as it is schematically represented by mode.
the dashed solid curves in Figs. 1b and 2b, In connection to what was said above, it is worth noting that the
necked down region can also be formed as a result of changes in
εf     the relative ease of plastic flow in width and thickness directions
I m (1 + r) ˇ+1
Dcrit = dε = ε1f + ε2f − ε0 (8) rather than on classical mechanisms based on the development of
 3 ˇ
ε0
unstable plastic deformation (Isik et al., 2015).
Now, by taking into consideration that fracture toughness is a
εf     material property, which is not influenced by strain loading paths
II  1 (1 + r) ˇ−1 and by combination of in-plane loading and bending effects, it fol-
Dcrit = dε = ε1f − ε2f − ε0 (9)
 2 (1 + 2r) ˇ lows that Eq. (12) can be utilized to determine the fracture loci.
ε0 In this way, the effective strains at fracture εf for strain loading
paths giving rise to fracture by crack opening modes I and II can be
2.3. Fracture toughness and fracture loci estimated from,
R  n+1
1
The link between fracture toughness R and fracture loci is based mode (n + 1)
εf = (13)
on the assumption that there should be a connection between the tK
specific essential work of fracture, which characterizes the ability where Rmode is the fracture toughness R in crack opening modes I
of a sheet to resist crack initiation, and the work required locally or II and t = tf is the thickness at fracture. The effective strain at
to nucleate or propagate (under continuous nucleation) a crack in fracture εf may be written as a function of the anisotropy r, the
tension or shear stress loading conditions. strain ratio ˇ and the major strain at fracture ε1f ,
As discussed by Atkins and Mai (1985), the specific essential
work of fracture R may be converted to a local work done per unit

1+r 2r 1 1
of volume at the crack tip by dividing R by the height h of the necked εf =  1+ + 2 ε1f (14)
(1 + 2r) (1 + r) ˇ ˇ
down process region. In case of the double notched test specimen
M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258 253

Force WT
F t wT=
tl

ln
l
l2
d
RI
l1

W T1
w

F l1 l2 ln
Displacement l
(a) (b) (c)

Fig. 5. Method and procedure for determining fracture toughness RI in crack opening mode I.
(a) Schematic representation of a double-notched test specimen;
(b) Schematic evolution of the force with displacement for test specimens with different ligaments;
(c) Determination of fracture toughness from extrapolation of the specific total energy.

A first glimpse into Eq. (13) allows concluding that fracture loci and the results are summarized in Table 1. The stress–strain curve
derived from fracture toughness R may be seen as an effective strain was approximated by the following Ludwik–Hollomon’s equation,
based criterion where the ellipses of constant effective strain at
fracture ε = kf in the principal strain space (refer to Figs. 1b and 2b)  = 140ε0.04 (MPa) (15)
are modified in order to include a dependency on fracture tough-
ness RI or RII , sheet thickness t and stress-strain response of the The normal r anisotropy coefficient included in Table 1 was
material by means of constant K and strain hardening exponent n determined from the anisotropy coefficients r˛ of the tensile tests
(in case of a material following  = Kε ).
n performed in specimens cut out from the supplied sheets at 0◦ , 45◦
The comparison between the fracture loci derived from Eq. (13) and 90◦ degrees with respect to the rolling direction,
and from experimental procedures based on the determination r0 + 2r45 + r90
of gauge length strains at fracture will be provided in Section 4 r= (16)
4
and will be utilized to validate the proposed link between fracture
toughness and fracture loci.
3.2. Formability limits

3. Experimentation The formability limit by necking (FLC) was determined by means


of sheet formability tests (tensile, Nakazima, bulge and hemispher-
The investigation was carried out in aluminium AA1050-H111 ical dome tests) that covered strain paths from uniaxial to biaxial
sheets with 1 mm thickness and the experimental tests that are stretching conditions. The procedure utilized for determining the
needed to validate the proposed analytical framework to character- in-plane strains (ε1 , ε2 ) at the onset of necking involved electro-
ize fracture loci under plane stress conditions directly from fracture chemical etching of a grid of overlapping circles with 2 mm initial
toughness in crack opening modes I and II of fracture mechan- diameter on the surface of the test specimens before forming and
ics were retrieved from previous research works performed by measuring the major and minor axes of the ellipses that resulted
the authors. Consequently, this section will only present a brief from plastic deformation.
summary of the methodologies and results associated with the The characterization of the formability limits by fracture (FFL
mechanical characterization and determination of the formability and SFFL) included additional results from double-notched test
limits and fracture toughness of the material that are relevant for specimens loaded in tension and in plane torsion. The latter were
the aims and objective of this paper. Further details and experi- needed to characterize the SFFL because all the remaining sheet
mental testing conditions are given in Madeira et al. (2015) and formability tests failed by fracture under crack opening mode I.
Isik et al. (2014, 2015). The determination of the strains at fracture involved measuring the
thickness of the specimens before and after fracture at several loca-
tions along the crack in order to obtain the ‘gauge length’ strains.
3.1. Mechanical characterization The gauge length strains were subsequently fitted by two straight
lines; (i) one straight line (FFL) falling from left to right with a slope
The mechanical characterization of the aluminium AA1050- ‘−0.68’ in agreement with Eq. (7) of Section 2.2 and (ii) another
H111 sheets at room temperature was carried out by means of straight line (SFFL) rising from left to right with a slope equal to
tensile tests in specimens cut out from the supplied sheets at 0◦ , 45◦ ‘+1.39’ in agreement with Eq. (8) of Section 2.2.
and 90◦ degrees with respect to the rolling direction. The method- The formability limits by necking and fracture of the aluminium
ology followed the ASTM standard E8/E8 M (ASTM E8/E8 M, 2013) AA1050-H111 sheets with 1 mm thickness are shown in Fig. 4.

Table 1
Summary of the mechanical properties of aluminium AA1050-H111 sheets with 1 mm thickness.

Modulus of elasticity (GPa) Yield strength (MPa) Ultimate tensile strength (MPa) Elongation at break (%) Anisotropy coefficient

0◦ RD 72.7 115.4 119.0 7.1 0.71


45◦ RD 67.9 120.4 121.2 5.2 0.88
90◦ RD 71.8 123.0 120.8 5.6 0.87
Average 70.0 119.9 120.5 6.8 r =0.84
254 M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258

Fig. 6. Schematic link between plastic flow, void coalescence and growth, ductile damage, crack opening modes, and fracture toughness in sheet metal forming.

3.3. Fracture toughness To conclude, it is worth mentioning that fracture initiation


points are located in the middle of the notches because the spec-
The determination of fracture toughness in plane stress in crack imens were slightly indented with a razor blade in the opposite
opening modes I and II made use of double-notched test specimens sides of the notched tips before testing in order to localize crack
loaded in tension and plane torsion. The methodology followed opening.
the original developments of Cotterell and Reddel (1977) for crack
opening mode I and the extension for crack opening mode II that 4. Results and discussion
was recently proposed by Isik et al. (2015).
Characterization of fracture toughness RI (mode I) using double- 4.1. Comparison of theoretical models for predicting fracture loci
notched test specimens loaded in tension is summarized in Fig. 5
and involved three main procedures; (i) determination of the total Fig. 6 provides a schematic comparison of the theoretical esti-
energy WT directly from the force-displacement evolution as a mate of the fracture loci (FFL and SFFL) obtained from ductile
function of the starting ligament length l of each test specimen, damage and void coalescence/growth models associated to crack
(ii) determination of the specific total energy wT = WT /tl as a func- opening modes I and II (Sections 2.1 and 2.2) with the theoreti-
tion of the starting ligament length l of each test specimen and (iii) cal estimates provided by fracture toughness based models with
determination of the specific energy of fracture (fracture tough- values determined from double-notched test specimens loaded in
ness in crack opening mode I) RI by extrapolating the specific total tension and plane torsion (Section 2.3).
energy wT to the limiting conditions in which the starting ligament As seen in Fig. 6, the fracture loci obtained from fracture tough-
length l approaches zero. ness tests in crack opening modes I and II (Eqs. (13) and (14)) are
A similar procedure was employed for determining fracture represented in the principal strain space as two different ellipses
toughness in crack opening mode II (RII ) from double-notched test (ε = f (RI ) and ε = f (RII )) with major axis along pure shear ε1 = −ε2
specimens loaded in plane torsion in which the force F and the dis- and minor axis along equal biaxial stretching ε1 = ε2 . This allows
placement were replaced by the torque T and the degree of rotation concluding that the slope ‘+1’ of the in-plane shear fracture forming
(Isik et al., 2015). Table 2 summarizes the results obtained in both line (SFFL) associated to crack opening mode II and to the ductile
test cases. damage criterion based on the ratio of the shear stress to the effec-
tive stress (Eq. (7)) exactly matches the slope of the derivative to

Fig. 7. Finite element distribution of effective strain for double-notched test specimens loaded in tension and plane torsion.
(a) Double-notched test specimen with a ligament l = 15 mm loaded in tension at 1 mm vertical displacement;
(b) Double-notched test specimens with a ligament l = 6.5 mm loaded in plane torsion at 2.5◦ of rotation.
M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258 255

Table 2
Summary of the experimental tests to determine fracture toughness in aluminium AA1050-H111 sheets with 1 mm thickness.

Test Geometry (mm) Fracture toughness R (kJ/m2 ) Thickness at fracture tf (mm)

t=1 56.9
w = 50
d=3
l = 5–25

t=1 67.1
r0 = 40
ri = 21
d=1
l = 4–11.5

the ellipse ε = f (RII ) at the fracture point ε1f = −ε2f corresponding 4.2. Assessment of theoretical and experimental fracture loci
to pure shear loading conditions.
In case of the fracture forming line (FFL) associated to crack Fig. 7 shows the finite element predicted distribution of effective
opening mode I and to the ductile damage criterion based on the strain for two double-notched test specimens taken from Table 2.
stress triaxiality ratio (Eq. (6)), there is no exact match between the As seen, the plastic deformation region has a slightly elliptical shape
slope ‘-1’ of the FFL and the slope of the derivative to the ellipse and the necked down zone where material experiences high values
ε = f (RI ) at the fracture point ε1f ∼ = −ε3f that is typical of near of effective strain is limited to a small volume located in-between
plane strain loading conditions of the double notched test spec- the notches with an average height h < d. This result corroborates
imens loaded in tension. However, not only the ellipse ε = f (RI ) the assumption h ∼ = t that was previously made in the analytical
presents little curvature in the segment from equal biaxial stretch- development of Section 2.2.
ing (where the derivative has a slope of ‘−1’) to plane strain, as the Fig. 8a shows the experimental strain loading paths in the prin-
observed increase in the slope of the derivative is compatible with cipal strain space for the double-notched test specimens loaded in
the upward curvature of the FFL when a threshold strain ε0 below tension and plane torsion. The open markers correspond to val-
which damage is not accumulated, is assumed to exist (refer to Eq. ues that were obtained from in-plane strain measurements with
(8) and to the dashed solid curve ε0 > 0 in Fig. 6). the Aramis commercial system whereas the solid markers corre-
To conclude the comparison of the theoretical methods for pre- spond to ‘gauge length’ strains at fracture that were obtained from
dicting fracture loci in sheet metal forming it is worth noting that through-thickness measurements along the cracks.
the linear loci with slope ‘−1’ (FFL) and ‘+1’ (SFFL) as well as the The strain loading paths given by the two lines (please notice
corresponding elliptical loci obtained from the fracture toughness that one of these lines is coincident with the vertical axis) cor-
based model were built upon the assumption that the through- respond to the finite element estimates provided by the test
thickness strain at fracture is constant for each crack opening mode specimens that are shown in Fig. 7. The agreement between exper-
and that fracture occurs without previous necking. The influence imental and numerically predicted strain loading paths is good up
of these two assumptions in the overall agreement between theory to the transition of experimental data from necking to fracture,
and experimentation will be discussed in the next section. which is plotted in a simplified manner as a change into vertical
direction (refer to the vertical dashed lines in Fig. 8a). In fact, finite
elements were not able to model the change of the strain loading
256 M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258

Fig. 8. Forming limits by fracture of the aluminium AA1050-H111 sheets with 1 mm thickness.
(a) Comparison between the experimental fracture loci determined from the gauge length strains at fracture and that calculated from fracture toughness by means of Eqs.
(13) and (14).
Comparison against experimental values of fracture determined from tensile tests and from single point incremental forming of truncated conical, lobe conical and pyramidal
shapes with varying drawing angles.

path into vertical direction in case of the double-notched test spec- By taking the stress-strain curve of the material (Eq. (1)) and the
imen loaded in plane torsion and the reason for this is attributed to experimental values of fracture toughness and thickness at frac-
the modelling conditions of uncoupled damage that were utilized ture from the double-notched test specimens loaded in tension
in order to cope with the analytical framework for estimating the and plane torsion (Table 2), and by then applying Eqs. (13) and
fracture loci in the principal strain space. (14) it is possible to obtain the two different ellipses ε = f (RI ) and
M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258 257

ε = f (RII ) corresponding to crack opening modes I and II that are failure by fracture in sheet metal forming. In other words, they
plotted in Fig. 8a and b. The theoretically predicted fracture locus is assumed the FFL and the SFFL to meet in a point.
constructed from the solid curved segments taken from these two However, a transition loci between the FFL and the SFFL is not
ellipses, as it is schematically illustrated in Fig. 8a and b. only likely to exist inside the grey quadrilateral of Fig. 8 as its occur-
As seen in Fig. 8a, the upper ellipse resulting from frac- rence makes sense due to the existence of infinite ellipses located
ture toughness experiments with double-notched test specimens in-between the upper and lower ellipses corresponding to crack
loaded in tension agrees well with the FFL that had been previously opening modes I and II.
determined by the authors (Isik et al. (2014), refer to Section 3.2). The above conclusion may justify future interest in extending
This may be attributed to the fact that the slope (‘−0.68’) of the the overall procedure based on the utilization of double notched
experimental FFL is closer to that of a secant between the fracture test specimens loaded in tension and plane torsion to include the
strain pairs (ε1f , ε2f ) at plane strain and equal biaxial stretching double notched staggered test specimens loaded in tension that
(where the derivative has a slope equal to ‘−1’). The dashed segment were originally proposed by Cotterell et al. (1982) to produce frac-
of the upper ellipse does not account for the fracture loci because ture in mixed modes I and II along the inclined direction, which
the strain loading paths in this region of the tension-compression forms at an angle to the tips of the notches.
quadrant do not fail by cracking in opening mode I.
However, Fig. 8a also discloses a poorer agreement between the 5. Conclusions
lower ellipse and the SFFL in the region of the tension-compression
quadrant located at the vicinity of pure shear, ε1 = −ε2 (refer to This paper enlarges a recently proposed vision for the fracture
the solid segment of the lower ellipse). This is attributed to the loci in sheet metal forming that makes use of fundamental concepts
following three main reasons. Firstly, the experimental SFFL has a of plastic flow, ductile damage and void coalescence and growth to
slope larger than that of the theoretical SFFL. Secondly, the double- include fracture toughness in two different crack opening modes.
notched test specimens loaded in plane torsion fail by fracture An analytical framework is proposed to estimate the location
with previous necking. Thirdly, the scattered measurements of the of the fracture loci in the principal strain space directly from the
in-plane strains of fracture in the necked down region of the lig- experimental measurements of fracture toughness and thickness
ament by the Aramis commercial system gives rise to a certain at fracture in double notched test specimens loaded in tension and
degree of uncertainty in the determination of the fracture strain plane torsion.
pairs (ε1f , ε2f ). The comparison between the fracture loci of AA1050-H111
The first of these reasons is justified by the fact that the slope aluminium sheets with 1 mm thickness obtained by means of
(‘+1.39’) of the experimental SFFL is larger than the slope (‘+1’) of the new proposed analytical framework and that retrieved from
the theoretical SFFL and of the derivative to the ellipse at the frac- experiments with different sheet formability tests shows a good
ture point ε1f = −ε2f corresponding to pure shear. This difference agreement in case of failure by tension and a fair agreement in case
moves the SFFL away from the ellipse. of failure by in-plane shear. The investigation also allows infer-
The second reason allows explaining a better agreement ring the existence of a transition fracture loci associated to crack
between the SFFL and the lower ellipse if fracture toughness RII opening by mixed modes.
(mode II) were larger than that given in Table 2 as a consequence of
hypothetical double-notched test specimens loaded in plane tor- Acknowledgements
sion that would fail by cracking without previous necking.
The third reason is also exclusive of double-notched test spec- K. Isik and A. E. Tekkaya gratefully acknowledge funding by
imens loaded in plane torsion because the fracture strain pairs the German Research Foundation (DFG) within the scope of the
(ε1f , ε2f ) are constructed from the experimental values of the Transregional Collaborative Research Centre on sheet-bulk metal
in-plane minor strains at fracture ε2f obtained from the Aramis forming (SFB/TR 73) in the subproject C4 ‘Analysis of load his-
commercial system and the values of in-plane major strains tory dependent evolution of damage and microstructure for the
at fracture ε1f = −ε3f that are obtained from through-thickness numerical design of sheet-bulk metal forming processes’. M.B. Silva
measurements at several locations along the cracks by using a and P.A.F. Martins would like to acknowledge Fundação para a
microscope. The reason why this problem does not affect the Ciência e a Tecnologia of Portugal and IDMEC under LAETA −
double-notched test specimens loaded in tension is because their UID/EMS/50022/2013 and PDTC/EMS-TEC/0626/2014.
strain loading paths experience near plane strain conditions ε2f ∼ =
0. References
Fig. 8b provides additional comparison of the theoretically pre-
dicted fracture locus constructed from the solid curved segments ASTM E8/E8 M, 2013. Standard Test Methods for Tension Testing of Metallic
taken from the two abovementioned ellipses against experimental Materials. ASTM International, West Conshohocken, USA.
Atkins, A.G., Mai, Y.W., 1985. Elastic & Plastic Fracture, Chichester, Ellis Horwood.
fracture strain pairs obtained by the authors in earlier investiga- Atkins, A.G., 1996. Fracture in forming. J. Mater. Process. Technol. 56, 609–618.
tions on truncated conical and pyramidal shapes (Isik et al., 2014) Bai, Y., Wierzbicki, T., 2010. Application of extended Mohr–Coulomb criterion to
and truncated lobe conical shapes (Soeiro et al., 2015) with varying ductile fracture. Int. J. Fract. 161, 1–20.
Centeno, G., Bagudanch, I., Martínez-Donaire, A.J., García-Romeu, M.L., Vallellano,
drawing angles produced by single point incremental forming. The C., 2014. Critical analysis of necking and fracture limit strains and forming
agreement is good and demonstrates the predictive capability of forces in single-point incremental forming. Mater. Des. 63, 20–29.
the new proposed vision and associated analytical framework for Cotterell, B., Reddel, J.K., 1977. The essential work of plane stress fracture. Int. J.
Fract. 13, 267–277.
loading conditions that are distinct from the calibration test cases. Cotterell, B., Lee, E., Mai, Y.W., 1982. Mixed mode plane stress ductile fracture. Int.
Fracture strain pairs retrieved from tensile tests are also included J. Fract. 20, 243–250.
for reference purposes. Gurson, A., 1977. Continuum theory of ductile rupture by void nucleation and
growth. I: Yield criteria and flow rules for porous ductile media. J. Eng. Mater.
To conclude it is worth noting that the differences between the
Technolo.—Trans. ASME 99, 2–15.
ellipses and the fracture loci at the transition between the FFL and Hill, R., 1948. A theory of yielding and plastic flow of anisotropic metals. Proc. R.
the SFFL (refer to the grey schematic quadrilateral in Fig. 8a) are Soc. Lond. Ser. A 193, 281–297.
easy to explain because Isik et al. (2014) assumed no crack open- Hill, R., 1952. On discontinuous plastic states, with special reference to localized
necking in thin sheets. J. Mech. Phys. Solids 1, 19–30.
ing by mixed modes in their analytical framework to characterize Isik, K., Silva, M.B., Tekkaya, A.E., Martins, P.A.F., 2014. Formability limits by
fracture in sheet metal forming. J. Mater. Process. Technol. 214, 1557–1565.
258 M.B. Silva et al. / Journal of Materials Processing Technology 234 (2016) 249–258

Isik, K., Silva, M.B., Atkins, A.G., Tekkaya, A.E., Martins, P.A.F., 2015. A new test for McClintock, F.A., Kaplan, S.M., Berg, C.A., 1966. Ductile fracture by hole growth in
determining fracture toughness in plane stress in mode II. J. Strain Anal. 50, shear bands. Int. J. Fract.Mech. 2, 614–627.
221–231. McClintock, F.A., 1968. A criterion for ductile fracture by the growth of holes. J.
Lemaitre, J., 1985. A continuous damage mechanics model for ductile fracture. J. Appl. Mech. Trans. ASME 35, 363–371.
Eng. Mater. Technol. 107, 83–89. Soeiro, J.M.C., Silva, C.M.A., Silva, M.B., Martins, P.A.F., 2015. Revisiting the
Li, Y., Luo, M., Gerlach, J., Wierzbicki, T., 2010. Prediction of shear-induced fracture formability limits by fracture in sheet metal forming. J. Mater. Process.
in sheet metal forming. J. Mater. Process. Technol. 210, 1858–1869. Technol. 217, 184–192.
Madeira, T., Silva, C.M.A., Silva, M.B., Martins, P.A.F., 2015. Failure in single point Tvergaard, V., Needleman, A., 1984. Analysis of the cup-cone fracture in a round
incremental forming. Int. J. Adv. Manuf. Technol. 80, 1471–1479. tensile bar. Acta Metall. 32, 157–169.
Martins, P.A.F., Bay, N., Tekkaya, A.E., Atkins, A.G., 2014. Characterization of Wierzbicki, T., Bao, Y., Lee, Y.W., Bai, Y., 2005. Calibration and evaluation of seven
fracture loci in metal forming. Int. J. Mech. Sci. 83, 112–123. fracture models. Int. J. Mech. Sci. 47, 719–743.

You might also like