Original Article

Proc IMechE Part L:

J Materials: Design and Applications
Formability limits by wrinkling 0(0) 1–12
! IMechE 2016
in sheet metal forming Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1464420716642794
pil.sagepub.com

JPG Magrinho, CMA Silva, MB Silva and Paulo AF Martins

Abstract
This paper presents a new combined experimental and theoretical methodology for determining the formability limits by
wrinkling in sheet metal forming. The methodology is based on the utilization of rectangular test specimens clamped
along its narrower sides and compressed lengthwise and is aimed at replicating the physics behind the occurrence of
wrinkling in deformation regions submitted to in-plane compression along one direction. The methodology draws from a
previous development in the field of flexible roll forming, and the overall objectives are to enhance and improve its
methods and procedures and to provide a new level of understanding on the onset of wrinkling in sheet metal forming.
Experimentation and finite element modelling of cylindrical deep-drawing without blank holder combined with the
utilization of the space of effective strain vs. stress triaxiality are employed to discuss the applicability and validity of
the new proposed methodology for determining the formability limits by wrinkling.

Keywords
Sheet metal forming, wrinkling, deep-drawing, experimentation, finite element method

Date received: 2 February 2016; accepted: 13 March 2016

The methodology for determining the formability

Introduction limits by wrinkling is not as consolidated as those
The formability limits in sheet metal forming set the utilized in necking and fracture. The reason for this
amount of deformation that can be attained without is attributed to the fact that wrinkling is influenced by
failure by necking, fracture or wrinkling. Figure 1 a significantly wider and more diverse range of par-
provides a schematic representation of these limits in ameters that include the mechanical properties of the
the principal strain space based on an original figure material, the geometry of the sheet metal part, the con-
made by Marciniak1 that was recently revised by tact conditions imparted by the tools and the applied
Martins et al.,2 as the result of new phenomenological level of stresses and strains.
and experimental evidence on the formability limits The idea behind the existence of a locus of in-plane
by fracture. strains delimiting the onset of wrinkling in sheet metal
The onset of necking is the most widely used form- forming is attributed to Havranek,6 who proposed the
ability limit in sheet metal forming, and it is generally concept of the wrinkling-limit curve (WLC) after mea-
plotted as a ‘V-shaped’ curve, designated as the form- suring the circumferential and radial strains on the
ing-limit curve (FLC). The FLC indicates the amount unsupported region of conical cups. The WLC is situ-
of deformation to be imparted to sheet metal parts for ated in the lower left-hand of the second quadrant of
which aesthetics difficulties and incipient fracture the principal strain space (Figure 1) but its exact loca-
derived from localized zones of thinning are likely to tion is commonly determined in a case-by-case basis
occur.3 depending on the regions of the sheet metal forming
The onset of fracture consists of a pair of curves processes to be analysed.
(designated as the fracture loci) that intersect at the
upper right-hand part of the second quadrant and
delimit the strain loading conditions above which
sheet metal parts fail by cracking opening modes I IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Lisboa,
(by tension) and II (by in-plane shear) of fracture Portugal
mechanics.4 The fracture loci and the circumstances
Corresponding author:
leading to fracture in sheet metal forming are compre- Paulo AF Martins, IDMEC, Instituto Superior Técnico, Universidade
hensively discussed in a recent state-of-the-art review de Lisboa, Av. Rovisco Pais, Lisboa 1049-001, Portugal.
published by Silva et al.5 Email: [email protected]

Downloaded from pil.sagepub.com by guest on April 7, 2016

2 Proc IMechE Part L: J Materials: Design and Applications 0(0)

In addition to the above mentioned difficulties, the

Yoshida buckling test may be further limited by the
small levels of strain that are reached in the test,
which are considerably below the typical strain
values found in real sheet metal parts. As a result of
this, Cao et al.13 proposed a modification of the
Yoshida buckling test (designated as the ‘contact
buckling test’) to include a support of the blank in
order to replicate the curvature of the sheet and the
contact with the tool, which are known to delay the
occurrence of wrinkling.
The difficulty in combining the broad and diverse
range of parameters that influence the onset of
wrinkling into a universal testing methodology justi-
fies the development of alternative procedures that
are valid for specific sheet metal forming processes.
Figure 1. Schematic representation of the formability limits For example, Kim and Yang14 proposed an energy-
by necking, fracture and wrinkling in sheet metal forming.2 based criterion to determine the onset of wrinkling
for various sheet metal forming processes such as
cylindrical, spherical and elliptical cup deep-drawing.
Abe et al.,7 for example, proposed the utilization of Sivasankaran et al.15 proposed the utilization of arti-
a conical cup drawing test performed with a flat-bot- ficial neural networks to predict the onset of wrink-
tomed punch having a much smaller diameter than ling and the position of the WLC in the principal
that of the die opening to investigate the onset of strain space.
wrinkling at the wall and flange of circular blanks. This paper is focused on a new testing method-
Narayanasamy and Sowerby8 utilized the same con- ology to determine the onset of wrinkling in sheet
ical cup drawing test with two different flat-bottomed metal forming that circumvents the aforementioned
and hemispherical-ended punches and predicted the problems related to the occurrence of cracking prior
onset of wrinkling in partially draw cups by combin- to wrinkling. The development has its origins in the
ing the theories of plasticity and structural stability. work of Kasaei et al.16 to determine the process
They confirmed the existence of a WLC correspond- window of flexible roll forming and makes use of a
ing to critical values d"1 =d"2 of the strain loading path rectangular sheet clamped at two opposite edges and
for which wrinkling occurs. compressed in a double-action tool frame along its
The practical applicability of WLCs is sometimes longitudinal direction.
questioned because of problems related to uniqueness The aim and objectives of the presentation are three-
and existence. In fact, different strain-paths are fold. First, to introduce the experimental setup of the
known to change the position of the WLCs and modi- new testing methodology and to explain its advantage
fications in existing strain-paths to include a compres- in replicating the physics behind the occurrence of
sive in-plane strain component may eventually give wrinkling in sheet metal forming regions that generally
rise to wrinkling anywhere in the principal strain account for in-plane compression along one direction.
space.9 Second, to combine experimental and finite element
However, because typical strain-paths under which results to provide a new level of understanding of the
wrinkling is likely to occur in sheet metal forming do WLC. Finally, to compare the experimental results and
not vary significantly,10 it is relevant to carry out the observations of wrinkling obtained from cylindrical
experimental characterization of the WLCs under deep-drawing without blank holder against the WLC
such representative loading conditions, so that wrink- built upon the new proposed testing methodology.
ling can be prevented. This last objective is crucial to evaluate the potential
Yoshida et al.11 proposed the utilization of a buck- of the methodology for successfully predicting the
ling test consisting of a flat square sheet clamped at occurrence of wrinkling in sheet metal forming.
opposite corners and pulled in a tensile test frame
along the diagonal direction, as an attempt to propose
a characterization procedure that is independent from Experimentation
specific sheet metal forming processes. The test was
employed by various authors12 to investigate the influ-
Mechanical characterization of the material
ence of mechanical properties such as the yield The work was carried out in aluminium AA1050
strength, the strain hardening exponent and the sheets with various thicknesses that were annealed at
anisotropy coefficient on the occurrence of wrinkling 345  C during 3 h prior to testing. The mechanical
but its utilization may be limited by cracking of the characterization of the material was performed by
specimen prior to wrinkling. tensile testing on specimens that were cut out from

Downloaded from pil.sagepub.com by guest on April 7, 2016

Magrinho et al. 3

Table 1. Summary of the mechanical properties of aluminum AA1050-O sheets.

Modulus of Yield strength Ultimate tensile Elongation at Anisotropy

elasticity E (MPa) Y (MPa) strength UTS (MPa) break A (%) coefficient r

72472 22.1 76.2 47 r0 ¼0.97

r90 ¼1.05

Figure 2. The proposed experimental test to determine the onset of wrinkling in sheet metal forming: (a) schematic representation
of the experimental setup and (b) analogy with the stress-state at the edge of the flange in cylindrical deep-drawing.

the annealed sheets at 0 and 90 degrees with respect analogy between the rectangular test specimen and
to the rolling direction. The values obtained for the the edge of the flange in cylindrical deep-drawing
modulus of elasticity E, the yield strength Y , the that is shown in Figure 2(b). The outer flange element
ultimate tensile strength UTS , the elongation at E plotted in dark grey is subjected to circumferential
break A and the anisotropy coefficient r are given in compressive loading and may be considered as
Table 1. The stress–strain response was approximated ‘clamped’ along its radial edges due to deformation
by the following Ludwik–Hollomon’s equation constraints along the circumferential direction.
In connection to this, it is worth noting that the
 ¼ 151:2 "0:3 ðMPaÞ ð1Þ utilization of alternative buckling tests combining ten-
sile and compression loading will not replicate the
stress-state at the edge but, somewhere, inside the
The results obtained for the anisotropy coefficients flange of the cylindrical cup. Therefore, they are not
r0 ffi r90 ffi 1 allow concluding that the annealed alu- suitable to reproduce the limiting strain conditions
minium sheets behave nearly isotropically. because the superposition of tensile stresses will min-
imize the tendency to wrinkle.
The testing apparatus developed by the authors is
Characterization of the onset of wrinkling schematically pictured in Figure 2(a) and consists pri-
The characterization of the WLC was performed by marily of two parts: the frame and the self-centring
means of rectangular test specimens clamped along its clamping system. The frame is built upon a double-
narrower sides and compressed lengthwise action cam sliding mechanism that converts the verti-
(Figure 2(a)). The utilization of compressive rather cal movement of the actuator (A) into the horizontal
than tensile loading as in case of the Yoshida buckling movement of the holders (H) containing the left and
test is aimed at replicating the physics behind the right clamping elements towards each other. The hori-
occurrence of wrinkling in deformation regions zontal movement of the holders (H) produces the
undertaking in-plane compression in one direction. compression loads that are applied on both clamped
This is schematically illustrated in the stress-state sides w0 of the test specimens. The self-centring

Downloaded from pil.sagepub.com by guest on April 7, 2016

4 Proc IMechE Part L: J Materials: Design and Applications 0(0)

clamping system is a simplified version of that used by Table 2. Rectangular test specimens that were utilized in the
Kasaei et al.16 and consists of two clamping shoes (S) characterization of the onset of wrinkling.
and two clamping elements (C). The clamping shoes Length Width
(S) serve to attach the clamping elements (C) to the Thickness l0 w0
holders (H). The clamping elements (C) provide the t0 (mm) (mm) (mm)
fixed end constraints of the rectangular test specimens
and ensure its self-alignment land in order to prevent t0 1, 2, 3, 4 20 50
35
deformation by pure bending. 50
As seen in Figure 2(a), the rectangular test speci- 20 100
mens subject to in-plane compression develop plastic
buckling in the centre along the width direction. The
evolution of the in-plane strain field during the test
required electrochemical etching a grid of interlaced
circles of 2 mm initial diameter d on the surface of the l0
sheets and measuring the lengths of the major a and
minor b axes of the ellipses (located on the middle of
w0
the width direction) that resulted from plastic deform-
ation of the original grid at different instants of time
(refer also to Figure 6(b)). The selected instants of 20
time correspond to progressively smaller distances (mm)
between the two opposite holders (H) containing the
left and right clamping elements.
The in-plane strain pairs ("1 , "2 ) resulting from the
above mentioned procedure are obtained at the grid The geometry of the tool and circular blanks are
points located at the outer surface of the buckled test summarized in Table 3.
specimens by means of conventional circle grid ana-
lysis, as follows
  Theoretical background
a b
"1 ¼ ln "2 ¼ ln ð2Þ Finite element modelling
d d
The procedure to characterize the WLC by means of
The deformed length l of the specimens and the the new proposed methodology combines experimen-
amplitude h of the plastic buckled waves tal data and finite element simulation. Finite element
(Figure 2(a)) also need to be measured for further analysis was carried out with the computer program
plots and calculations that will be addressed later in LS-Dyna, and the rectangular test specimens were dis-
the presentation. cretized by means of 4450 shell elements with five inte-
Table 2 presents a summary of the rectangular spe- gration points across the sheet thickness (Figure 3(a)).
cimens that were used in the characterization of the The rectangular test specimens were simulated as elas-
onset of wrinkling by means of the new proposed tic–plastic deformable objects and their mechanical
testing methodology. properties are supplied in equation (1) and Table 1.
The clamping elements were modelled as rigid
objects, and the application of compressive loading
Cylindrical deep-drawing benchmark cases
was accomplished by imposing a constant velocity at
Cylindrical deep-drawing without blank holder was the constrained regions of the specimens. The desired
utilized to check the occurrence of wrinkling against amount of total displacement was accomplished by
the WLC that resulted from the experiments with rect- means of an updated Lagrangian formulation that
angular test specimens that were described in the pre- made use of a succession of displacement increments
vious section (section Characterization of the onset of each of one corresponding to 10  6 mm in the longi-
wrinkling). tudinal direction. The overall central processing unit
The deep-drawing specimens consisted of circular (CPU) time for a typical analysis consisting of a finite
blanks that were cut out from the annealed sheets and element model similar to that shown in Figure 3(a)
electrochemically etched with a grid of interlaced cir- was 6 h on a computer equipped with an IntelÕ
cles of 2 mm initial diameter d on the surface of the CoreTM i7 CPU (2.93 GHz) processor and making
sheets. The corresponding in-plane strains were deter- use of four physical cores.
mined by circle grid analysis following a procedure
similar to that employed in the calculation of the
strains that are needed for the characterization of
Methodology to determine the onset of wrinkling
the WLC (refer to section ‘Characterization of the The combined experimental and numerical method-
onset of wrinkling’). ology to determine the onset of wrinkling is

Downloaded from pil.sagepub.com by guest on April 7, 2016

Magrinho et al. 5

Table 3. Geometry of the tool and circular blanks that were utilized in the cylindrical deep-drawing tests.

Thickness t0 (mm) Diameter D0 (mm)

1, 2, 3, 4 91

summarized in Figure 3(b) to (d). The methodology is The second procedure utilizes the finite element
valid for a specific sheet thickness and first requires evolution of the principal strains with the ratio h=l
ensuring compatibility between the experimental and of the amplitude h to the length l of the deformed
the numerically predicted evolutions of the compres- test specimens (hereafter designated as the ‘normal-
sion force with displacement (Figure 3(b)). Second, it ized amplitude of plastic instability waves’) to deter-
is necessary to calculate the critical strain pairs mine the critical strain pairs ("crit crit
1 , "2 ) at the instant
("crit crit
1 , "2 ) at the onset of instability for the entire of time when the critical force Fcr is attained and
set of test specimens and to plot them in the principal wrinkling starts (h 4 0mm, Figure 3(c)). The critical
strain space in order to define the WLC (Figure 3(c)). strain pairs for test specimens with different initial
Finally, it is necessary to transform the WLC from the lengths l0 are then employed to construct the WLC
principal strain space into the space of effective strain in the principal strain space (Figure 3(d)).
" vs. stress triaxiality m = (Figure 3(d)). In addition to finite element estimates, it is also
The first procedure is necessary because the onset important to obtain experimental values of the prin-
of wrinkling is not always attained simultaneously cipal strains by circle grid analysis at selected instants
along the entire width w of the specimens. This is of time, corresponding to progressively smaller dis-
particularly noticeable for specimens with small thick- tances between the two opposite holders (H) contain-
nesses and is attributed to the clamping conditions ing the left and right clamping elements. These
and to the straightness and perpendicularity toler- experimental values were taken after the onset of
ances in both opposite and adjacent sides of the spe- wrinkling and are mainly utilized to double-check
cimens. In fact, these two issues may prevent users the finite element simulations and to ensure the over-
from obtaining a perfect alignment of the specimens all quality of the predicted WLCs.
and may even cause spurious localized deformations As will be seen in the following section, the WLCs
by pure bending (Figure 3(b)). can subsequently be transformed from the principal
These localized deformations are observed at the strain space into the space of effective strain " vs.
beginning of the tests performed with thinner specimens stress triaxiality m =
and are responsible for triggering the onset of wrinkling
for smaller values of forces and displacements than Representation of the WLC in the space
those obtained if the specimens were perfectly flat. As
a consequence, it is necessary to introduce a quantity
of effective strain vs. stress triaxiality
designated as the ‘initial effective width’ we0 to adjust the The transformation of the WLCs from the principal
experimental settings and observations to the finite strain space into the space of effective strain " vs.
element simulative conditions, as follows (Figure 3(b)) stress triaxiality m = (Figure 3(d)) is based on the
application of Hill’s 48 yield plasticity criterion
we0 Fexp under plane stress conditions 3 ¼ 0
¼ cr ð3Þ
w0 Ffem
cr
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where Fexp fem
cr and Fcr are the experimental and finite 2r
element predicted critical forces at the onset of plastic  ¼ 12 þ 22  1 2 ð4Þ
ð1 þ rÞ
instability.16

Downloaded from pil.sagepub.com by guest on April 7, 2016

6 Proc IMechE Part L: J Materials: Design and Applications 0(0)

Figure 3. Summary of the methodology utilized to characterize the onset of wrinkling: (a) finite element model of the rectangular
test specimen before and after compression, (b) utilization of the initial effective width w0e to ensure compatibility between the
experimental and the finite element predicted evolution of the force-displacement curves, (c) utilization of the finite element estimates
to determine the critical strain pair at the onset of wrinkling and (d) schematic representation of the wrinkling-limit curve in the
principal strain space and in the space of effective strain vs. stress triaxiality.

In fact, by writing the constitutive equations asso- where the effective strain increment d" is obtained
ciated to Hill’s 48 criterion that relate the major d"1 from
and minor d"2 in-plane strain increments with the
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
applied stresses as follows 1þr 2r
d" ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d"21 þ d"22 þ d"1 d"2 ð6Þ
  ð1 þ 2rÞ ð1 þ rÞ
d" 1
d"1 ¼ ð1 þ r ð1  2 ÞÞ
 1 þ r
  It is possible to establish a relation between the
d" 1
d"2 ¼ ð2 þ r ð2  1 ÞÞ ð5Þ stress triaxiality m = and the slope  ¼ d"1 =d"2 cor-
 1 þ r
responding to the proportional strain loading path of

Downloaded from pil.sagepub.com by guest on April 7, 2016

Magrinho et al. 7

any arbitrary critical strain pair lying on top of the plastic instability waves obtained from rectangular
WLCs (refer, for example, to point ‘A’ in Figure 3(d)) test specimens subjected to the previously described
pffiffiffiffiffiffiffiffiffiffiffiffiffi correction procedure. The critical strains at the
m 1 þ 2r ð1 þ Þ onset of plastic instability ("crit 1 ¼ 0:0066, "2 ¼
crit
¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7Þ 0:0130) are obtained from the y-intercepts of the
 3 1 þ 2r  þ 2
ð1þrÞ major and minor in-plane strains (refer to the detail
in Figure 6(a)) and correspond to the instant of time
The above equation allows transforming the WLCs when the critical instability force Fcr is attained and
from the principal strain space into the space of effect- buckling is triggered (that is, when h 4 0 mm). The
ive strain vs. stress triaxiality that will be later utilized square markers correspond to the control experimental
in the analysis of wrinkling in cylindrical deep- values obtained from circle grid analysis (Figure 6(b)).
drawing. Now, by repeating the above mentioned procedure
for the entire set of test specimens that is listed in
Table 2, it is possible to determine the WLCs for
Results and discussion each sheet thickness t0 (Figure 7(a)). As seen in the
figure, the WLCs start presenting an initial slope
Evolution of the force with displacement d"1 =d"2 ¼ 1=2 similar to that of pure in-plane com-
Figure 4 shows the evolution of the force with dis- pression but later change direction towards smaller
placement for selected rectangular specimens taken slopes as the ratio l0 =t0 of the initial length l0 to the
from the entire set of testing conditions listed in initial thickness t0 of the rectangular test specimens
Table 2. As seen, the force increases steeply up to a decreases.
peak value corresponding to the critical instability The change in direction of the WLCs is identified
force Fcr after which starts to decrease at a much by means of point ‘B’ in case of the rectangular test
lower rate with increasing displacement. The value specimens with t0 ¼ 4 mm (Figure 7(a)) and corres-
of the critical instability force Fcr and the amount of ponds to a change in material flow towards an
displacement corresponding to the onset of plastic increasing amount of thickening before wrinkling.
instability are found to increase as the initial length This change in material flow is caused by the reduc-
l0 decreases (Figure 4(a)) and the initial thickness t0 tion of the initial length l0 available to trigger and
increases (Figure 4(b)). propagate the plastic instability waves and may pre-
The particular evolution of the force with displace- vent the occurrence of wrinkling in situations like that
ment obtained for the test specimen with l0 ¼ 20 mm of the rectangular test specimen with l0 ¼ 20 mm and
and t0 ¼ 4 mm (Figure 4(a)) corresponds to thicken- t0 ¼ 4 mm in Figure 4(a). As mentioned in previous
ing without the occurrence of plastic instability. This section ‘Evolution of the force with displacement’, the
mode of deformation is similar to those commonly mode of deformation of this particular test specimen
found in sheet-bulk metal forming,17 where the aim is typical of sheet-bulk metal forming processes and,
is to achieve thickening and/or produce local func- therefore, it will not provide a critical strain pair to be
tional features such as ribs and solid bosses outside included in the definition of the WLC corresponding
the original thickness of the sheets. to t0 ¼ 4 mm.
The necessity of defining an effective initial width The representation of the WLCs in the space of
we0 to ensure compatibility between the experimental effective strain " vs. stress triaxiality m =
and finite element predicted evolution of force with (Figure 7(b)) results from transformation of the results
displacement is illustrated in Figure 5. As seen, it is plotted in the principal strain space (Figure 7(a)) by
necessary to correct the initial width w0 by the effect- means of the analytical procedure that was previously
ive width we0 given by equation (3) for the test speci- described in section ‘Representation of the WLC in the
mens with lower thicknesses (Figure 5(a)), but this space of effective strain vs. stress triaxiality’. Point ‘B’
necessity disappears for the test specimens with illustrates the result of the transformation between the
larger thicknesses (Figure 5(b)). two spaces.
The discrepancies that are observed in Figure 5(a) The space of effective strain " vs. stress triaxiality
for the total amount of experimental and finite elem- m = will be utilized in the following section of the
ent predicted displacement at the peak forces is attrib- paper to compare the experimental results and obser-
uted to the localized deformation that is observed at vations of wrinkling obtained from cylindrical deep-
the beginning of the test (refer to section drawing without blank holder against the WLC built
‘Methodology to determine the onset of wrinkling’) upon the new proposed testing methodology.
because it facilitates the occurrence of wrinkling.
Assessment by cylindrical deep-drawing
Determination of the onset of wrinkling Figure 8 shows a typical finite element model of the
Figure 6(a) presents the finite element evolution of the cylindrical deep-drawing test cases that are listed in
in-plane strains with the normalized amplitude h=l of Table 3. As in case of the rectangular test specimens,

Downloaded from pil.sagepub.com by guest on April 7, 2016

8 Proc IMechE Part L: J Materials: Design and Applications 0(0)

Figure 4. Force displacement evolution of selected rectangular test specimens of Table 2. (a) constant initial thickness t0 ¼ 4 mm
under the influence of the different initial lengths l0 , (b) constant initial length l0 ¼ 50mm under the influence of different initial
thicknesses t0 .

the circular blanks were discretized with 11972 shell As seen in Figure 9(a), the finite element evolution
elements with five integration points across the sheet of the loading path for the grid points located at the
thickness. The blanks were simulated as elastic–plastic edge of the flange evolves under pure compression up
deformable objects and its mechanical properties are to point ‘B1’ where the WLC corresponding to t0 ¼ 1
given in equation (1) and Table 1. The punch and die mm changes direction. The cause for this change in
were modelled as rigid objects and discretized by direction was already explained in section
means of spatial triangular and quadrilateral elements. ‘Determination of the onset of wrinkling’ (refer to
Figure 9 shows the finite element predicted loading Figure 7) and what is now relevant to understand is
paths in the space of effective strain vs. stress triaxi- if and how it will influence the loading path of the
ality for circular blanks with 1, 3 and 4 mm thickness. selected grid points located at the edge of the flange.
The loading paths were determined for points located In fact, after crossing point ‘B1’, the edge of the
at the edge of the flanges and include valleys and hills flange is no longer capable of undertaking thickening
in case of wrinkled cylindrical cups. The experimental without wrinkling and, therefore, its loading path
values are plotted as squares (solid squares for valleys bifurcates into the two extreme loading path condi-
and open squares for hills) for the circular blanks with tions of the grid points located at the bottom of the
3 and 4 mm thickness and were obtained after trans- valleys and at the top of the hills. The bottom of the
formation of the circle grid measurements by means valleys continues to be thick and this is the reason
of equations (6) and (7). why its loading paths continue to evolve under pure

Downloaded from pil.sagepub.com by guest on April 7, 2016

Magrinho et al. 9

Figure 5. Experimental and finite element predicted force-displacement evolution of selected rectangular test specimens of Table 2:
(a) initial length l0 ¼ 50mm and t0 ¼ 1mm and (b) initial length l0 ¼ 20mm and initial thickness t0 ¼ 4mm.

compression. In contrast, the top of the hills starts to For example, in case of a sheet with t0 ¼ 1mm, the
thin and its corresponding loading paths develop maximum initial length lmax 0 ffi 35mm (Figure 9(a))
towards the right side of the space of effective strain and, therefore, the expected total number of wrinkles
vs. stress triaxiality. Both strain loading paths are the is 8. This value is identical to that observed in the
result of bending caused by wrinkling during which cylindrical deep-drawing experiments.
the neutral plane moves towards the inside of the Finally, in what concerns the remaining grid points
bending regions. located in between the bottom of the valleys and the
The crossing point ‘B1’ can also be understood top of the hills, it is worth noting that they are located
from another perspective by referring to the analogy on the loading envelope given by the dotted curve
between the new proposed test and cylindrical deep-- than is included in Figure 9(a), for at a particular
drawing in Figure 2(b). In fact, because ‘B1’ corres- instant of time. In other words, all possible values
ponds to the maximum initial length lmax 0 of a of stress triaxiality m = are located within the two
rectangular test specimen that undergoes wrinkling above-mentioned extreme loading paths.
after previous thickening exclusively under pure The bifurcation of the loading path at the point
compression loading, it can be utilized to estimate the where the WLC experiences a change in direction is
total number of wrinkles n that will be triggered in the also observed for the cylindrical deep-drawing tests per-
flange of a deep-drawing cup at the onset of wrinkling formed with a sheet thickness t0 ¼ 3 mm (Figure 9(b)).
D0 The point is now designated as ‘B3’ and the finite elem-
n¼ ð8Þ ent predictions are in good agreement with the experi-
lmax
0 mental values included as square markers.

Downloaded from pil.sagepub.com by guest on April 7, 2016

10 Proc IMechE Part L: J Materials: Design and Applications 0(0)

Figure 6. In-plane strains for a test specimen with l0 ¼ 50mm and t0 ¼ 2mm: (a) Experimental control values and finite element
evolution of the in-plane strains with the normalized amplitude h=l of plastic instability waves and (b) Photograph showing a buckled
test specimen and the in-plane principal strain directions.

Figure 7. Wrinkling-limit curves (WLCs) for different sheet thicknesses: (a) Representation in the principal strain space and (b)
representation in the space of effective strain vs. stress-triaxiality.

The result shown in Figure 9(c) is somewhat differ- flanges. This is confirmed by observation of the
ent from those included in Figure 9(a) and (b) because photograph of the cylindrical cup included in
the working range of the effective strain is now very Figure 9(c). The overall agreement with the experi-
close to the threshold point ‘B4’ where the WLC mental values is also good.
changes direction. As a result of this, the signs of From what was mentioned above, it may be con-
bifurcation in the loading path are unnoticeable and cluded that besides the relevance of determining the
no significant wrinkling is observed at the edge of the shape of the WLC by means of the new proposed

Downloaded from pil.sagepub.com by guest on April 7, 2016

Magrinho et al. 11

Figure 8. Finite element model utilized in the numerical simulation of the cylindrical deep-drawing without blank holder.

Figure 9. Assessment of the WLCs in the space of effective strain vs. stress triaxiality by means of cylindrical deep-drawing without
blank holder using circular blanks with (a) t0 ¼ 1 mm, (b) t0 ¼ 3 mm and (c) t0 ¼ 4 mm.

Downloaded from pil.sagepub.com by guest on April 7, 2016

12 Proc IMechE Part L: J Materials: Design and Applications 0(0)

testing methodology, it is of paramount importance to 2. Martins PAF, Bay N, Tekkaya AE, et al.
identify the critical strain pair at which the slope of Characterization of fracture loci in metal forming. Int
the WLC changes towards an increasing thickening J Mech Sci 2014; 83: 112–123.
rate before wrinkling. In fact, because the edge of 3. Atkins AG. Fracture in forming. J Mater Process Tech
1996; 56: 609–618.
the flange in a deep-drawing operation is not capable
4. Isik K, Silva MB, Tekkaya AE, et al. Formability limits
of changing its loading path towards an increasing
by fracture in sheet metal forming. J Mater Process
rate of thickening due to the stress boundary condi- Tech 2014; 214: 1557–1565.
tions, it will wrinkle and the loading path will bifur- 5. Silva MB, Isik K, Tekkaya AE, et al. Fracture loci in
cate accordingly. sheet metal forming: a review. Acta Metall Sin (Eng
Lett) 2015; 28: 1415–1425.
6. Havranek J. Wrinkling limit of tapered pressing. J Aust
Conclusions Inst Met 1975; 20: 114–119.
The new testing methodology based on the double- 7. Abe H, Nakagawa K and Sato S. Proposal of conical
sided compression of a rectangular sheet along its cup buckling test. In: Proceedings of the congress of the
longitudinal direction is capable of replicating the international deep drawing research group, Kyoto,
Japan, May 1981, pp.1–19.
physics behind the occurrence of wrinkling in the
8. Narayanasamy R and Sowerby R. Wrinkling of sheet
sheet metal forming deformation regions submitted
metals when drawing through a conical die. J Mater
to in-plane compression along one direction. The Process Tech 1994; 41: 275–290.
onset of wrinkling is determined by combining 9. Thomson PF and Gunasekera JS. On the wrinkling of
circle grid analysis and finite element modelling of sheet metal during deep drawing. In: Proceedings of
the rectangular test specimens and is plotted as the NAMRC X – the 10th North American manufacturing
WLC in the principal strain space and in the space research conference, Hamilton, ON, Canada, 24–25
of effective strain vs. strain triaxiality. The WLC is May 1982, pp.173–179. Society of Manufacturing
made of two parts, an initial straight line and a final Engineers (SME).
curved shape that matches the increasing rate of 10. Szacinski AM and Thompson PF. Investigation of the
thickening before wrinkling. The transition point existence of a wrinkling-limit curve in plastically-
deforming metal sheet. J Mater Process Tech 1991; 25:
between the two parts of the WLC corresponds to
125–137.
the last admissible critical strain pair that is compat-
11. Yoshida K, Hayashi H, Miauchi K, et al. Yoshida
ible with the stress boundary conditions that are buckling test. In: Proceedings of the international sym-
commonly found at the edge of the flanges of posium on new aspects of sheet metal forming, Tokyo,
deep-drawing cups. Thus, any loading path going Japan, 14–15 May 1981, pp.125–148. The Iron and Steel
beyond this transition point in a deep-drawing oper- Institute of Japan.
ation performed without blank holder will give rise 12. Szacinski AM and Thompson PF. The effect of mech-
to wrinkling and to bifurcation of the loading path anical properties on the wrinkling behavior of sheet
into the two extreme conditions that are found at the material in the Yoshida test and in a cone forming
bottom of the valleys and at the top of the hills. The test. In: Proceedings of the 13th Biennial congress of
initial length of a rectangular test specimen corres- the international deep drawing research group,
Melbourne, VIC, Australia, 20–25 February 1984,
ponding to the transition point can also be utilized
pp.532–542. Australian Institute of Metals.
to estimate the number of wrinkles that will be trig- 13. Cao J, Cheng SH, Wang HP, et al. Buckling of sheet
gered in the flange of a deep-drawing cup at the metals in contact with tool surfaces. Ann CIRP 2007;
onset of wrinkling. 56: 253–256.
14. Kim JB and Yang DY. Prediction of wrinkling initi-
Declaration of conflicting interests ation in sheet metal forming processes. Eng Comput
2003; 20: 6–39.
The author(s) declared no potential conflicts of interest with 15. Sivasankaran S, Narayanasamy R, Jeyapaul R, et al.
respect to the research, authorship, and/or publication of Modelling of wrinkling in deep drawing of different
this article. grades of annealed commercially pure aluminium
sheets when drawn through a conical die using artificial
Funding neural network. Mater Des 2009; 30: 3193–3205.
16. Kasaei MM, Moslemi Naeini H, Liaghat GH, et al.
The author(s) disclosed receipt of the following financial
Revisiting the wrinkling limits in flexible roll forming.
support for the research, authorship, and/or publication
J Strain Anal 2015; 50: 528–541.
of this article: The authors would like to acknowledge the
17. Sieczkarek P, Isik K, Ben Khalifa N, et al. Mechanics of
support provided by Fundação para a Ciência e a
sheet-bulk indentation. J Mater Process Tech 2014; 214:
Tecnologia of Portugal and IDMEC under LAETA –
2387–2394.
UID/EMS/50022/2013 and PDTC/EMS-TEC/0626/2014.

References
1. Marciniak Z. Assessment of material formability. Adv
Technol Plast 1984; 1: 685–694.

Downloaded from pil.sagepub.com by guest on April 7, 2016