International Journal of Mechanical Sciences 44 (2002) 79–101

Measurement of springback
W.D. Cardena , L.M. Genga , D.K. Matlockb , R.H. Wagonera;∗
a
Department of Materials Science and Engineering, The Ohio State University, 177 Watts Hall,
2041 College Road, Columbus, OH 43210-1179, USA
b
Colorado School of Mines, Department of Metallurgical Engineering, Golden, CO 80401, USA
Received 16 May 2000; received in revised form 2 July 2001

Abstract

Springback, the elastically-driven change of shape of a part after forming, has been measured under care-
fully-controlled laboratory conditions corresponding to those found in press-forming operations. Constitutive
equations emphasizing low-strain behavior were generated for three automotive body alloys: drawing-quality
silicon-killed steel; high-strength low-alloy steel; and 6022-T4 aluminum. Strip draw-bend tests were then
conducted using a range of die radii (3 ¡ R=t ¡ 17), friction coe6cients (0 ¡  ¡ 0:20), and controlled ten-
sile forces (0:5 ¡ Fb =Fy ¡ 1:5). Springback angles and curvatures were measured for bend and bend–unbend
areas of the specimen, the latter corresponding to the “sidewall curl” region, which dominates the geometric
change and the dependence on process variables. Friction coe6cient and R=t (die-radius-to-sheet-thickness)
greater than 5 have modest but measurable e;ects over the ranges tested. As expected, strip tension dominates
the springback sensitivity, with higher forces reducing springback. For 6022-T4, springback is dramatically
reduced as the tensile stress approaches the yield stress, corresponding to the appearance of a persistent an-
ticlastic curvature. The presence of this curvature, orthogonal to the principal curvature, violates the simple
two-dimensional models of springback reported in the literature. The measured springback angles and cur-
vatures are reported both in graphical summary and tabular form for use in assessing analytical models of
springback. ? 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Springback; Sheet metal forming; Anticlastic curvature; DQSK steel; 6022-T4 aluminum; HSLA steel; Draw–
bend tests; Plastic anisotropy

1. Introduction

Springback, the elastically-driven change in shape of a part upon unloading after forming, is a
growing concern as manufacturers increasingly rely on materials with higher strength-to-modulus
∗
Corresponding author. Tel.: +1-614-292-2079; fax: +1-614-292-6530.
E-mail address: wagoner.2@[Link] (R.H. Wagoner).

0020-7403/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 0 - 7 4 0 3 ( 0 1 ) 0 0 0 8 2 - 0
80 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Nomenclature
h arc height at the center of the strip
Fb actual back force
Fy yield force
Fb normalized back force
Ff actual front force
R tool radius
Ra anticlastic radius of curvature on the curl region
R
, 1 radius and corresponding angle for the region of sheet strip that has
undergone only bending
r
, 2 radius and corresponding angle for the region of sheet strip that has
undergone both bending and straightening
t sheet thickness
w arc width of the sheet strip cross section
overall springback angle of sheet strip
a anticlastic curvature on the curl region
 friction coe6cient

ratios than the traditional low-strength steel. High-strength low-alloy steel (HSLA) and 6022-T4
aluminum alloy are frequently considered as direct replacements for drawing-quality silicon-killed
steel (DQSK), the current mainstay for automotive body panels.
Corrections for springback are essential during die design in order to obtain speciFed Fnal shapes.
Correction curves based on empirical information or simple theories for small-springback in pure
bending cases have been available for many decades [1–3] based on the basic assumptions of engi-
neering beam bending. Many analytical solutions have been derived, and with reviews have appeared
for the early work [4,5], and more recently [6,7].
While the literature dealing with analysis of springback of metal sheets is extensive, carefully-
controlled experiments of springback under realistic forming conditions (i.e. involving bending and
unbending simultaneous with imposed tension and sliding over the tooling) are less common. Many
experiments have been carried out under pure-bending (i.e. with minimal tension) conditions of var-
ious kinds: cylindrical tooling [8–10], U-bending=channel bending [11–13], V-bending [12–14], and
Kanging [15]. Such experiments show springback increasing with R=t (tool radius=sheet thickness),
but such results have little application to situations where signiFcant sheet tension is present, because
sheet tension dominates other process variables in determining springback.
Stretch-bend tests [13,16,17] allow careful control of sheet tension during bending, but do not
typically exhibit bending and unbending, nor large sliding over the tooling common in press form-
ing operations. These tests do show, however, the pervasive e;ect of sheet tension on springback,
especially for sheet tensile stresses in the range of the material yield stress.
Actual forming operations, such as a two-dimensional idealization of draw=stretch bending of a
Kanged channel (also called a top-hat section) are used to assess practical springback. This geometry
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 81

is the most studied of springback cases [18–24] (including benchmark testing and simulations [25])
because of its practical importance, and because of the obvious, measurable presence of sidewall curl.
However, in such operations, the sheet tension is determined only indirectly, either from the blank
holder force, or from a drawbead simulation [26 –29], both of which depend on the coe6cient of
friction, which is usually known only approximately. Because of the dominant role of sheet tension
in springback, lack of control or direct measurement of this quantity is a serious drawback for
scientiFc use of such experiments in verifying simulation techniques. Such operations have common
features, including reduced springback and curl for increased blank holder forces (especially when
sheet tension approaches the yield stress). Curl has been shown to disappear as R=t approaches 2
for a variety of steels [18,22,24]. For smaller R=t, the curl may reverse direction.
The exceptional work is by Liu [30] for Kanged channels, and Kuwabara et al. [31] for draw-bending.
Liu devised a special channel Kanging experiment in which the restraining force was applied directly
by a controlled hydraulic cylinder. In this way, sidewall curl and springback could be assessed in
terms of known restraining forces for several materials. Kuwabara et al. used a draw–bend apparatus
that appears similar to the one in the current work, although dimensions and speciFcations were
not provided [31]. Reported springback angles decreased with increasing R=t and the usual depen-
dence on sheet tension was observed. The current draw=bend test [32–34], conceived for friction
testing, is similar in concept to Liu’s experiment, with four signiFcant di;erences of the current
device:

(1) much longer drawn distances are attainable, with correspondingly increased precision of shape
measurement,
(2) there is no bend at the bottom of the channel to complicate interpretation,
(3) there is only a single tool, thus avoiding changing clearances and geometries during the punch
stroke, and
(4) the material hardening curves were measured and Ft in some detail.

The sensitivity of springback to a Bauschinger e;ect, particularly for bend=unbend operations

(as is the case for draw=bend tests and Kanged channel forming) has been noted in several anal-
yses [23,35 – 41]. However, until recently, the magnitude of the Bauschinger e;ect in sheet metal
tension–compression [42,43] or bend=reverse bend [44,45] had seldom been measured because of the
instability of metal sheets subjected to in-plane compression.
A variety of manufacturing techniques have been developed for dealing with springback. The
simplest involves designing tools to over-bend the sheet to compensate for springback. However,
this is not applicable to complex curved parts and, without additional control, this approach can
lead to large variations in practice as material properties and thickness vary. Other approaches
include the use of deformable tools [46,47], adjustments to small die radii and clearances [22,24],
through-thickness deformation [48], variable or stepped blank holder force control [21,36,49 –52],
multiple forming steps [19,36,53], or reconFgurable tooling [54]. Each of these methods can beneFt
from accurate knowledge of springback from measurement and analysis.
The current work is motivated by a lack of systematic springback measurements for deformation
under a range of carefully controlled laboratory conditions but applicable to industrial practice. In
particular, the role of friction coe6cient and R=t in springback has been equivocal, and data is
needed to evaluate the accuracy of simulation techniques.
82 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Table 1
Chemical compositions in weight percent [55,56]

Material C Mn P S Si Cu Ni Cr Mo Sn Al N V Cb
DQSK 55 0.035 0.22 0.009 0.010 0.007 0.03 0.01 0.02 0.01 0.002 0.031 0.004 0.000 0.001
HSLA55 0.070 0.477 0.010 0.004 — 0.02 0.02 0.03 — 0.001 0.022 0.005 — 0.039

6022-T456 Si—1.24 Fe—0.13 Cu—0.09 Mn—0.07 Mg—0.58 Ti—0.02

2. Experimental procedure

Strain hardening curves, with particular attention to the low-strain region, were generated by
standard tensile testing. Stretch=bend testing was conducted using apparatus and specimen procedures
developed by Matlock and coworkers [32–34] for the measurement of friction. Digitizing of the
specimen shapes and several forms of subsequent analysis were used to reveal the nature of the
springback after removal from the grips of the device.

2.1. Materials

Three materials were selected and provided under the auspices of a large cooperative project to
predict springback following automotive forming operations. The Frst alloy, DQSK steel [55], is a
current mainstay of the automotive industry for body panels. The other two alloys are promising or
planned alternatives suitable for reducing vehicle mass: HSLA steel [55], and 6022-T4 aluminum
[56]. The HSLA steel was hot-dipped galvanized while the other materials were uncoated. Chemical
compositions are listed in Table 1.
Standard tensile tests [57] were carried out using a 50 mm-gage extensometer to record strains,
at a nominal strain rate of 0.001=s. Three stress–strain curves were averaged to produce a single
composite set of data points for the rolling direction (parallel to the draw direction for all draw–bend
tests) of each material. Constitutive equations were then produced from this set of data points, with
special attention to the region near the elastic–plastic transition.
To describe the stress–strain curve accurately, three regions were deFned. The Frst is a linear Ft
to the initial elastic region, the second is a elastic–plastic transition region, and the third is large
strain plasticity. For the plastic region of the curve above 5% strain, typical hardening laws of the
Hollomon [58], Swift [59], and Voce [60] types were considered. The Holloman law Ft well for
the steels and the Voce law Ft well for 6022-T4. The transition zone was of particular interest
for its e;ects on springback response and simulation. Therefore, the plastic laws were modiFed to
include this region. A separate set of tensile tests involving repeated loading and unloading with
data acquired at smaller strain intervals was used to Ft the linear region to obtain Young’s modulus.
To describe the transition region, the deviation of this portion of the curve from the large-strain Ft
was determined, such that the deviation automatically approached zero at large strain. Trial and error
Fts for various kinds of equations were employed to improve the curve Ft in this region. Each of the
equations was Fnally Ft to the experimental curve over the entire strain range using a least-squares
technique to obtain the Fnal terms of the constitutive equations. The transition region and large-strain
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 83

Fig. 1. Measured and Ft strain hardening.

plastic regions of DQSK and 6022-T4 were combined without signiFcant degradation of the Ft and
thus only two equations. The irregularity of the HSLA yield point curve required the use of three
regions in the Fnal Ft. For each Ft, an overall correlation coe6cient greater than 0.99 was obtained.
The resulting constitutive equations and the comparison with the composite experimental data appear
in Fig. 1 and Table 2.

2.2. Draw=bend test

The draw=bend test [32–34] used in this study, Fig. 2, consists of two hydraulic actuators oriented
◦
90 to one another, with a Fxed or rolling cylinder at the intersection of their action lines to simulate
a tooling radius over which the sheet metal is drawn. The upper actuator (horizontal) is programmed
to provide a constant restraining force (or back force) while the lower actuator is set to displace at
a constant speed, thus drawing the strip over the cylinder. The material undergoes tensile loading,
bending, and unbending as it is drawn over the tooling. At the end of the test, the material is allowed
84 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Table 2
Material properties and hardening laws, rolling direction tensile tests

Material Strain Stress equation Constants

y : yield stress (c = correlation coe6cient)
rRD : plastic anisotropy, RD

DQSK  ¡ 0:0035  = E E = 212 GPa

(t = 1:50 mm)  ¿ 0:0035  = A[(B + )C − De−F ] A = 539 MPa
(y = 159 MPa) B = 0:00965
(rRD = 1:87) c = 0:99983 C = 0:274
D = 0:278
F = 1700

HSLA  ¡ 0:0012  = E E = 202 GPa

2 −F
(t = 1:50 mm) 0:0012 ¡  ¡ 0:0253  = A[B + C − (1 − D )e ] A = 592 MPa
(y = 414 MPa) B = 0:686
(rRD = 0:69) c = 0:99692 C = 1:044
D = 109000
F = 995
 ¿ 0:0253  = Kn K = 772 MPa
R = 0:99987 n = 0:163

6022-T4  ¡ 0:002  = E E = 69 GPa

−F
(t = 0:91 mm)  ¿ 0:002  = A[1 − B exp(−C) − De ] A = 389 MPa
(y = 172 MPa) B = 0:566
(rRD = 0:73) c = 0:99992 C = 8:44
D = 1:20
F = 1120

to spring back by removal of the specimen from the grips of the Fxtures. The parts of the strip
outside the formed region remain straight.
Strips were sheared to lengths of 508 mm along the sheet rolling direction, with widths of 50 mm.
After one end was gripped in the upper Fxture, each strip was hand-formed to an unloaded radius
◦ ◦
of 90 (± approximately 1 ), and the other end was clamped in the lower grip. The hand forming
of the initial radius conforming to the tooling is a possible source of variation from test-to-test,
although this formed region is small compared with the overall draw distance of 127 mm. For the
central radius value of 9:5 mm, the ratio between the length of the two regions (one corresponding to
a single bend and the other to a bend-and-unbend) is 9-to-1. For the high sheet tensions employed,
these small variations will be essentially removed by plastic deformation throughout a large part
of the sheet thickness as the strip originally formed to the radius passes through the rest of the
contact region and then exits. Simulations of elastic–plastic pure bending corresponding to the hand
forming operation show that the relief of all of the residual stresses results in an angular change
◦
less than 0:3 . Thus, it appears that the variations introduced by hand forming the initial radius are
inconsequential compared with other sources of test variation such as load control, material property
variation, and friction coe6cient changes.
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 85

Fig. 2. Draw=bend test geometry: (a) specimen shape after unloading; and (b) original and Fnal shapes during testing.

The strip was drawn over the radius at a constant velocity of 40 mm=s. After forming and removal,
the shapes were traced onto paper; the traces being optically digitized later. Measurements were taken
from both the traces and the digitized images. Gage marks applied to the outer surface (50 mm apart)
were used for measurement of the axial strains.
Note. As part of quantifying day-to-day and test-to-test scatter, repeated measurements of U
were conducted over a period of time [61]. The results were consistent, within the measurement
scatter, for both steels. For 6022-T4 aluminum, however, the springback angles continued
to increase for periods up to several months after forming. A few tests of other aluminum
alloys: 2008-T4, 5182-O, and 6111-T4, showed similar behavior, presumably related to room-
temperature creep driven by the residual stress pattern. This was mainly observable for the
conditions producing large U . The angles reported throughout this paper refer to the long-time
measurements for 6022-T4.
In order to investigate the roles of tool radius, tension, and friction in springback, tests were
conducted with a selection of these process parameters. Multiple sets of tooling, both Fxed and
rolling, were constructed with radii of 3.2, 6.4, 12.7, and 25:4 mm. Draw restraint (via back force)
was set at a fraction of each material’s yield strength in the following increments: 0.5, 0.7, 0.9, 1.1,
1.3 and 1.5. Friction was modiFed by using lubricated rollers (minimum friction), lubricated Fxed
tools (medium friction), and unlubricated Fxed tools (high friction). A standard industrial drawing
lubricant [62] was used for all lubricated cases.
The friction coe6cient for each material and lubricant can be quantiFed by various analytical
schemes [32,34] based on the front (measured) and back (controlled) forces for each test. Use
86 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Fig. 3. Simulated force ratios as functions of friction coe6cient, with superimposed experimental results.

of these schemes gave inconsistent results and, in some cases, values of friction coe6cient out-
side of normally reported ranges. Instead, using a series of simulation techniques and elements, the
draw=bend test was simulated [63– 68]. By comparing simulated front forces with measured ones,
Fig. 3, it was possible to obtain friction coe6cients for each combination of material and lubrication
condition. These results were nearly independent of the simulation details, and they have the advan-
tage of incorporating actual conformance to the tool surface (rather than ideal conformance usually
assumed), variable pressure throughout the tool contact, and bending=unbending terms in the draw
force. The scatter shown in Fig. 3 for lubricated cases is calculated from the scatter of measured
front and back forces for a single condition. The e;ective friction coe6cients for DQSK and HSLA
under free-rolling conditions are not as low as expected ( = 0:06), presumably because of rolling
friction caused by bending of the roller under the higher draw loads required by these thicker mate-
rials. The unexpectedly low friction coe6cient for the dry case for HSLA ( = 0:09) may be related
to the lubricity of the coating for that material. The other two materials were uncoated.
Triplicate testing of all process combinations (comprising 648 experiments) was judged impracti-
cal, so a more limited series was devised to reveal the main features. The central combination, the
one considered most typical of industrial practice, consisted of lubricated, Fxed tooling, a normalized
tensile force (F b ) 0.9 and 1.1 times the yield strength, and a tooling radius of 9:5 mm. This radius
is a common one in automotive dies [69].
From this central combination, a series of tests was performed at various R=t or Fb while main-
taining the other variables constant. These are the principal combinations used to determine the
experimental sensitivity of springback to R=t and Fb . The extreme combinations of these variables
were also tested, although in some cases material or machine limitations intervened. HSLA failed
for F b greater than 1.1 of the yield strength, and the smallest tool radius did not allow rolling dur-
ing drawing, presumably because of roller bending, so reliable low friction conditions could not be
obtained. Because of the observed di6culties in obtaining low-friction conditions in some of these
cases, the results for low-friction conditions should be considered approximate for the smaller tool
radii. A complete set of test results for all tested combinations appears in Tables 3–5.
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 87

Table 3
Measured forces, strains, curvatures and springback angles from draw=bend tests of 6022-T4 aluminum alloya

R Fb Fb Ff U U 1 U 2 l Ra
(mm) (N) (N) (degrees) (degrees) (degrees) (mm)
Dry
6.35 0.9 6841 9056 9.8 −7:9 17.7 0.082 213
9.525 0.5 3807 5413 47.7 −7:0 54.7 0.038 1267
9.525 0.7 5329 7272 24.7 −6:5 31.2 0.059 556
9.525 0.9 6899 9701 5.1 −6:5 11.6 0.065 236
9.525 1.1 8416 11347 3.6 −3:7 7.3 0.103 269
12.7 0.9 6877 9145 2.2 −5:6 7.8 0.050 244

Lubricated
3.175 0.5 3060 2566 79.2 −10:2 89.4 0.085 1509
3.175 0.5 3732 2949 70.8 −16:6 87.4 0.097 1247
3.175 0.9 6881 9532 17.5 −5:1 22.6 0.137 284
3.175 1.1 8371 11360 5.4 −1:6 7.0 0.217 556
6.35 0.9 6917 8900 10.5 −6:8 17.3 0.085 224
6.35 1.1 8451 10622 6.3 −7:8 14.1 0.113 234
6.35 1.1 8981 11467 2.6 51.2 −48:6 0.223 996
9.525 0.5 3879 5186 52.2 −9:9 62.1 0.058 998
9.525 0.7 5369 6814 35.8 −9:5 45.3 0.064 729
9.525 0.9 6926 ± 22 8669 ± 67 6:9 ± 0:5 −8:5 ± 2:8 15:4 ± 2:8 0:079 ± 0:055 221 ± 10
9.525 0.94 7179 ± 9 8714 ± 209 5:2 ± 0:3 −8:0 ± 0:8 13:2 ± 1:1 0:042 ± 0:016 218 ± 8
9.525 1.1 8420 10328 4.6 −7:4 12.0 0.063 213
9.525 1.3 9861 12036 3.2 −6:2 9.4 0.065 290
9.525 1.4 10546 12739 0 −4:8 4.8 572
12.7 0.9 6952 ± 49 8513 ± 22 5:3 ± 0:6 −5:7 ± 1:4 10:9 ± 2:0 0:124 ± 0:052 249 ± 8
12.7 1.1 8500 ± 71 10168 ± 49 3:6 ± 0:6 −4:5 ± 4:8 8:1 ± 5:3 0:070 ± 0:008 251 ± 23
25.4 0.9 9439 10444 1 −10:6 11.6 274
25.4 1.1 −0:4 −3:0 2.6 0.061 246

Free rolling
2.38125 0.3 2224 4524 81.7 −6:9 88.6 0.061 1725
2.38125 0.5 3763 ± 13 6374 ± 245 63:5 ± 4:5 −11:7 ± 5:4 75:1 ± 9:8 0:095 ± 0:044 1354
2.38125 0.9 6797 9279 17.9 −15:9 33.8 0.062 450
3.175 0.5 3794 ± 22 6067 ± 53 71:1 ± 2:9 −2:8 ± 3:4 73:9 ± 2:0 0:095 ± 0:005 1554 ± 218
6.35 0.5 4114 ± 0 4924 ± 31 63:4 ± 3:4 −13:3 ± 2:8 76:7 ± 0:6 0:092 ± 0:029 1651 ± 56
6.35 0.9 7264 8349 30.6 −6:2 36.8 0.007 795
6.35 1.4 10640 ± 62 12107 ± 231 4:1 ± 0:5 −1:6 ± 2:4 5:7 ± 1:9 0.157 ± 0:043 391 ± 69
6.35 ? 10840 12365 2.4 −1:2 3.6 0.043 533
9.525 0.5 3848 4154 57.8 −8:5 66.3 0.060 1458
9.525 0.7 5329 5627 45.6 −7:9 53.5 0.012 1186
9.525 0.9 6957 7237 33.3 −7:5 40.8 0.054 892
9.525 1.1 8758 8852 11.1 −7:2 18.3 0.055 394
9.525 1.4 10751 11182 5 −7:1 12.1 0.047 328
12.7 0.9 25.6 −5:4 31.0 0.011
25.4 0.5 3856 ± 9 3892 ± 9 35:9 ± 2:4 −14:8 ± 1:5 50:7 ± 3:9 0:131 ± 0:046 2487 ± 168
25.4 0.9 7103 7157 9.2 −8:3 17.5 0.015 490
25.4 1.4 10684 ± 4 10693 ± 9 −0:6 ± 0:4 −6:3 ± 1:5 5:7 ± 1:7 0:081 ± 0:078 330 ± 5
a
R: Tool radius, Fb : normalized back force, Fb : actual back force, Ff : actual front force, l : longitudinal strain in the
curl region and Ra : anticlastic curvature in the curl region.
88 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Table 4
Measured forces, strains, curvatures and springback angles from draw=bend tests of DQSKa

R Fb Fb Ff U U 1 U 2 l Ra
(mm) (N) (N) (degrees) (degrees) (degrees) (mm)

Dry
6.35 0.9 10715 16102 4.1 −4:3 8.4 0.099 881
9.525 0.5 5943 9345 7.3 −6:6 13.9 0.053 897
9.525 0.7 8371 12245 5.5 −4:3 9.8 0.065 770
9.525 0.9 10760 15323 4 −1:7 5.7 0.075 861
9.525 1.1 13322 18450 1.5 −2:5 4.0 0.091 1288
12.7 0.9 10746 16231 3.2 −4:4 7.6 0.062 876

Lubricated
3.175 0.9 10724 18259 −5:4 7.5 −12:9 0.182 3119
3.175 1.1 13202 ± 125 20181 ± 165 −5:1 ± 0:9 −16:0 ± 12:3 10:9 ± 11:4 0:219 ± 0:009 1857 ± 295
6.35 0.9 10786 15688 4.6 −6:3 10.9 0.101 892
6.35 1.1 13193 18130 2.6 −3:4 6.0 0.116 925
9.525 0.5 5885 9287 7.2 −3:2 10.4 0.039 1148
9.525 0.5 6272 9599 7.9 −1:8 9.7 0.040 1267
9.525 0.7 8402 12276 7.3 0.1 7.2
9.525 0.9 10809 ± 9 15123 ± 165 5:6 ± 0:5 −1:4 ± 2:2 6:9 ± 2:4 0:069 ± 0:005 988 ± 81
9.525 1.1 13233 17730 3 −4:6 7.6 0.085 955
9.525 1.3 15608 20358 2.2 −3:4 5.6 0.112 927
9.525 1.5 17930 21880 0.5 −3:1 3.6 0.154 2162
12.7 0.9 10813 ± 31 13375 ± 147 6:2 ± 0:9 −2:7 ± 1:5 8:8 ± 0:6 0:063 ± 0:010 823 ± 81
12.7 1.1 13268 15746 4.8 2.1 2.7 0.061 927
12.7 1.1 12819 15323 3.6 −2:1 5.7 0.071 925
25.4 0.9 10826 13295 2.8 −3:0 5.8 0.030 676
25.4 0.9 11124 13838 2 −3:8 5.8 0.033 729
25.4 1.1 13237 17365 0.3 −1:4 1.7 0.054 732
25.4 1.1 13219 17116 0.040 737

Free rolling
3.175 0.5 5934 ± 13 13175 ± 338 −3:8 ± 0:1 −17:2 ± 0:8 13:4 ± 0:9 0:118 ± 0:003 1417 ± 130
6.35 0.5 5974 9554 8.4 −7:3 15.7 0.057 1125
6.35 0.5 6258 9839 7.5 −5:3 12.8 0.064 1130
6.35 0.5 6730 10319 7 −6:8 13.8 0.061 1201
6.35 0.9 11093 14981 5.7 −8:2 13.9 0.098 1250
6.35 1.5 18130 ± 22 22569 ± 133 −0:1 ± 0:1 −7:6 ± 0:4 7:5 ± 0:6 0:183 ± 0:002 2367 ± 8
9.525 0.5 5987 7766 9.4 −6:7 16.1 0.036 1115
9.525 0.7 8358 10052 7.8 −4:3 12.1 0.053 1290
9.525 0.9 10795 12806 7.1 −4:5 11.6 0.061 998
9.525 1.1 13455 14905 3.2 −5:0 8.2 0.106 1346
9.525 1.5 17974 20870 0.7 −1:1 1.8 0.135 2014
12.7 0.9 10773 12188 5.7 −4:9 10.6 0.050 803
25.4 0.5 5996 ± 13 6485 ± 13 7:0 ± 0:5 −4:7 ± 1:0 11:7 ± 1:5 0:012 ± 0:005 886 ± 198
25.4 0.9 10947 11307 4 −4:2 8.2 0.024 688
25.4 1.5 18014 ± 9 18428 ± 27 −0:4 ± 0:0 −3:6 ± 0:2 3:2 ± 0:2 0:069 ± 0:006 767 ± 18
a
R: Tool radius, Fb : normalized back force, Fb : actual back force, Ff : actual front force, l : longitudinal strain in the
curl region and Ra : anticlastic curvature in the curl region.
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 89

Table 5
Measured forces, strains, curvatures and springback angles from draw=bend tests of HSLAa

R Fb Fb Ff U U 1 U 2 l Ra
(mm) (N) (N) (degrees) (degrees) (degrees) (mm)

Dry
6.35 0.9 28169 34205 0.6 −15:5 16.1 0.191 3254
9.525 0.5 15759 20470 16.6 −5:6 22.2 0.070 993
9.525 0.9 28214 33582 0.9 −7:4 8.3 0.126 1643
12.7 0.9 28280 ± 4 33556 ± 298 1:7 ± 0:6 −4:6 ± 1:3 6:3 ± 1:9 0:094 ± 0:007 897 ± 76

Lubricated
3.175 0.7 21826 29268 −8:3 −15:3 7.0 0.235 3035
6.35 0.9 28205 33018 −0:2 −7:9 7.7 0.167 1151
6.35 1 31332 35571 0 −4:8 4.8 0.208 2667
9.525 0.5 15572 19745 14 −7:2 21.2 0.063 874
9.525 0.7 23179 27400 7.9 −4:1 12.0 0.089 894
9.525 0.9 28231 ± 178 32337 ± 173 2:6 ± 0:7 −5:5 ± 1:6 8:0 ± 1:8 0:117 ± 0:003 1057 ± 79
9.525 1 31407 34930 0.9 −7:1 8.0 0.152 1377
9.525 1.02 32021 35379 0.8 −6:8 7.6 0.152 1425
9.525 1.04 32399 36318 0.6 −5:5 6.1 0.172 1923
9.525 1.04 32671 36144 0 0.0 0.0 0.211 2225
12.7 0.5 15644 19380 11.8 −5:8 17.6 0.044 693
12.7 0.9 28356 32088 0.9 −4:9 5.8 0.093 622
12.7 1 31447 34494 0.8 16.7 −15:9 0.127 884
12.7 1.02 32057 35050 0.8 −4:3 5.1 0.135 859
12.7 1.04 33284 36371 0.5 2.3 −1:8
25.4 0.9 28409 31065 −0:7 −3:7 3.0 0.045 632
25.4 1 31519 34170 −1:2 −3:1 1.9 0.061 759

Free rolling
6.35 0.5 15741 ± 27 20977 ± 67 13:5 ± 0:7 −8:5 ± 3:1 22:1 ± 3:5 0:094 ± 0:003 818 ± 25
6.35 0.9 28191 33627 0.7 −2:6 3.3 0.169
6.35 1.04 31492 37310 0.8 −7:1 7.9 0.260 1636
9.525 0.5 15764 17836 11.5 −9:6 21.1 0.058 859
9.525 0.7 22018 24331 9.8 −7:7 17.5 0.083 922
9.525 0.9 28289 32390 1.9 −7:8 9.7 0.118 1603
9.525 1 31421 34009 2 −2:9 4.9 0.137 2878
9.525 1.02 31937 ± 44 35895 ± 36 0:8 ± 0:2 −4:5 ± 2:8 5:3 ± 3:0 0:168 ± 0:004 3193 ± 84
12.7 0.9 28414 29628 3.4 −1:6 5.0 0.076 838
25.4 0.5 15839 16471 13.7 −8:7 22.4 0.022 917
25.4 0.9 28400 28841 2.1 −4:4 6.5 0.045 693
a
R: Tool radius, Fb : normalized back force, Fb : actual back force, Ff : actual front force, l : longitudinal strain in the
curl region and Ra : anticlastic curvature in the curl region.

2.3. Measurement of springback

The Fnal specimen shape can be deFned by two radii of curvature, R
and r
(see Fig. 2), and
two corresponding angles, 1 and 2 . R
and 1 are the unloaded quantities representing the region
90 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

remaining in contact with the tool at the end of the test (before unloading), which has undergone
only bending, not straightening. r
and 2 represent the region which has been drawn over the tool
and which has undergone bending and straightening. The two measures are related by Fxed arc
lengths of R=2 (tool contact length) and 127 mm (the driving actuator displacement), respectively.
The springback angles are the changes of these measures upon unloading:
U 1 = 1 − loaded
1 = 1 − =2 = =2 (R=R
− 1); (1)

loaded
− 0 = 127 mm=r
;
◦
U 2 = 2 − 2 = 2 (2)

U =U 1 + U 2: (3)
The change of 1 (U 1 is generally negative) represents relaxation from a single bend to conform
to the tool radius while 2 corresponds to a region of straightening from this radius. U 2 , generally
positive, is a measure of “sidewall curl” [70,71] in industrial channel forming.
The shapes of the samples were carefully traced onto paper immediately after unloading. (For
6022-T4, the shapes were retraced at long times, and these long-time shapes are reported here.)
was measured directly from the traces by constructing straight lines through the specimen legs.
The overall springback angle (or U ) determined from CAD image and tracing were Frst compared,
◦
and were found to agree to a typical tolerance of 0:2 . Because of the large draw distance involved
in the draw=bend test, the strains occurring during springback can be resolved with high precision.
◦
The measurement scatter of 0:2 corresponds to a strain resolution at the outer Fber of the sidewall
curl area of 0:00001 or 0:001%:
Several techniques were tested to obtain the component angles 1 and 2 . The most consistent
was based on the use of digitized images and CAD software. A circle was numerically Ft to an arc
length of (=2)R in the single-bend region. This circle was superimposed digitally, and the tangent
points with both the upper leg and curl region were determined (separated by a Fxed arc length
of (=2)R). The tangent at the tool radius=curl boundary deFnes both 1 and 2 . Comparison of
angles determined by this procedure for multiple experiments under identical conditions revealed
◦ ◦
that typical measurement uncertainties for 1 and 2 (or U 1 and U 2 ) are 1 –3 , the larger errors
corresponding to smaller tool radii. Therefore, the measurement of U is more accurate than its
components by an order of magnitude.
For certain test results, principally aluminum specimens with intermediate sidewall curling, a sec-
ondary curvature in the sidewall curl region orthogonal to the principal curvature could be observed.
This secondary curvature, which occurs as a result of di;erential contraction in the width direction
from the top to bottom sides of the strip, is known as anticlastic curvature 1 [7,72]. The existence of
the persistent secondary curvature during unloading increases the e;ective moment of inertia of the
cross section about the bending axis, thus reducing the springback angle. The anticlastic curvature
was measured on each specimen at a position 51 mm from the location of the boundary between
the straight upper leg and Frst curve (76 mm for the test cases of tool radii of 25:4 mm), thus near

1
It should be noted that the principal and secondary curvature in the sidewall curl region are typically both concave on
the tool side of the strip, which violates the formal deFnition of “anticlastic”, although this term has acquired the meaning
of a secondary, orthogonal curvature resulting from a primary bending.
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 91

Fig. 4. Typical Fnal specimen shapes for variation of back force (R = 9:5 m) and tool radius (F b = 0:9).

the center of the sidewall curl region. The arc height at the center of the strip (h) and width of the
section (w) were measured and the radius of curvature (Ra ) was computed as follows:
1 h w2
Ra = = + ; (4)
2 8h
where is the curvature. The measured anticlastic curvatures are presented in Tables 3–5.

3. Results

Typical shapes of the unloaded strips are presented in Fig. 4. As expected, sheet tension has a
clear and dramatic e;ect on reducing total springback. (The back force is presented in normalized
form by dividing the measured back force by the force to yield the material in tension, as presented
for each material in Table 2.) The role of increasing R=t (for a normalized back force of 0.9) is less
consistent, with a signiFcant decrease of U for 6022-T4 and little measurable e;ect for HSLA. For
DQSK, only the extreme values of R=t have a signiFcant e;ect, including a reverse sidewall curl
at the smallest R=t tested (2:1), a trend reported by others for drawing steel and channel forming
[11,22,24].
92 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Fig. 5. Variation of total component (U 1 , U 2 ) and total springback angles (U ) with back force.

Figs. 5 and 6 show the relationship of total springback angle (U ) to its components: U 2
(corresponding to sidewall curl springback) and U 1 (corresponding to springback over the die
radius). For the three tested materials, U 1 is small and approximately constant while U 2 changes
markedly with the level of back force. In terms of R=t, U 1 is slightly sensitive for 6022-T4 (Fig. 6),
but no trend could be observed outside of the scatter. The variation of overall shape is thus dominated
by the sidewall curl, as would be expected for the large draw distance (sidewall curl length) relative
to the tool contact length. Because of the much better precision available for measuring U compared
to U 1 or U 2 , the remainder of the results will be presented in terms of U , although all three
angles are available in Tables 3–5 for detailed comparisons with theory or simulation.
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 93

Fig. 6. Variation of springback angle with the radius of tool-to-sheet thickness (R=t).

The test-to-test scatter of measured angles was investigated by repeating tests of the central condi-
tions (F b =0:9; R=9:5 mm, lubricated) at least seven times over several days. The standard deviation
◦ ◦
of the experimentally determined springback angle ( or U ) was 0:5 for DQSK, 1:1 for 6022-T4,
◦
and 0:7 for HSLA. The magnitude of test-to-test scatter correlates approximately with the magnitude
of the springback. For comparison, the scatter in measurement of U (from a single experiment) is
◦ ◦
approximately 0:2 , while the scatter of the two component angles, U 1 or U 2 , is 1–3 .
Fig. 7 shows the expected trend of decreasing springback with back force (or sheet tension), as
reported in virtually every reported analytical and experimental study. While DQSK and HSLA show
a continuously decreasing springback, 6022-T4 shows a plateau for normalized back force greater
than 0.9. This plateau of springback at higher sheet tensions has been reported by others [14,43].
Fig. 7 also shows that friction in normal industrial ranges (i.e. well-lubricated to dry conditions)
has no measurable e;ect on springback. Even for the free-rolling condition, corresponding to the
minimum friction attainable, only the 6022-T4 showed a signiFcant di;erence from the lubricated=dry
tests, mainly at intermediate back forces. Experimental results in the literature correlate springback
to friction conditions. However, in most such experiments or forming operations, the back force is
not controlled independently, as it is for the present experiments. The discrepancy in results suggests
that the role of friction in practical forming lies in the modiFcation of sheet tension.
Tooling radius has a modest e;ect on springback, Figs. 8 and 9. For back forces less than the
yield force, increasing R=t beyond 5 reduces springback for both 6022-T4 and DQSK. There are few
data for R=t less than 5, but the trend reverses, with springback approaching zero as R=t approaches
2. This trend is consistent with reported observations [18,22,24], and the critical ratio of 5 – 6 is
generally taken as the boundary between bending of thin and thick shells [6]. For large back forces,
and for all HSLA results, R=t has little or no e;ect on springback. These results suggest that for
low hardening induced during bending (i.e. for HSLA or at high sheet tension), the e;ect of R=t is
diminished.
The measured anticlastic curvature ( a ) for 6022-T4 increases dramatically for normalized back
forces in the range of 0.8–1.2, Fig. 10a. For R=t ¡ 7, the critical range is narrower, perhaps F b =
0:9–1.0. In either case, as the sheet tension approaches the yield stress, persistent and signiFcant anti-
clastic curvature is measured. For this range of back force, where the anticlastic curvature is large,
94 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Fig. 7. Role of friction on measured total springback angle (U ).

the springback angle drops precipitously, Fig. 5a. This correlation can be understood by considering
the moment of inertia of the deformed strip of 50:8 mm width. This moment controls the principal
bending and unbending. Three cases illustrate the range of moment of inertia (I ) for a Kat sheet
(ideal two-dimensional case) of width 50:8 mm, a sheet with curvature of 0:001 mm−1 (for F b ¡ 0:9
or F b ¿ 1:2), and a sheet with curvature of 0:004 mm−1 (for 0:8 ¡ F b ¡ 1:2). These values may be
computed using standard formulas [73].

I( a = 0:0 mm−1 ; Ra = ∞) = 3:2 mm4 ; (5)

I( a = 0:001 mm−1 ; Ra = 1000 mm) = 3:6 mm4 ; (6)

I( a = 0:004 mm−1 ; Ra = 250 mm) = 10:0 mm4 : (7)

 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 95

Fig. 8. Variation of total springback angle (U ) with R=t ratio (free rolling condition).

Thus, anticlastic curvature has at most a modest e;ect on the section moment of inertia outside of
the range of critical back forces, contributing 13% or less to the moment of inertia. In the critical
region of R=t, the anticlastic curvature increases the Kat-sheet moment of inertia by a factor of 3,
thus potentially reducing springback by a similar factor. This analysis agrees with the trends in the
measured springback values shown in Fig. 5a.
The anticlastic curvature for the thicker (t = 1:5 mm) DQSK and HSLA steels is much lower,
Figs. 10b and c. It is nearly constant, within experimental scatter, over the range of back forces
and R=t values tested. For comparison, the moment of inertia for these thicker sections is increased
less than 5% by the typical anticlastic curvature of 0:001 mm−1 (Ra = 1000 mm). For the highest
observed anticlastic curvatures for steels ( = 0:015 mm−1 ; Ra = 667 mm), the increase of section
96 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

Fig. 9. Variation of springback angle with R=t ratio (lubricated condition).

moment of inertia is 11%. Thus, the steels are expected to show a smoothly decreasing springback
with back force, as is the case in the measured values presented in Figs. 5b and c.
Fig. 11 illustrates the rapid change of the ratio of springback (primary) curvature-to-anticlastic
(secondary) curvature with back force for 6022-T4. For F b ¡ 0:8, the primary curvature is much
greater than the anticlastic curvature (up to a ratio approaching 20). For F b ¿ 0:8, the ratio reverses,
and changes at a much lower rate.

4. Conclusions

Draw-bend tests have been conducted and analyzed to investigate the role of typical process vari-
ables on springback and to provide accurate data for evaluation of springback models and simulations
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 97

0.005
R/t = 3.5 6022-T4
R/t = 6.9 Lubricated
R/t = 10.4
0.004
R/t = 13.9
Anticlastic Curvature (1/mm)

R/t = 27.8

0.003

0.002

R/t >7

0.001

0
0.3 0.5 0.7 0.9 1.1 1.3 1.5

0.0025
Anticlastic Curvature (1/mm)

R/t = 2.1
R/t = 4.2 DQSK
0.002 R/t = 6.3 3 < R/t < 8 Lubricated
R/t = 8.5
0.0015 R/t = 16.9

0.001

0.0005

0
0.3 0.5 0.7 0.9 1.1 1.3 1.5
Anticlastic Curvature (1/mm)

0.0025
HSLA
0.002 Lubricated
3 < R/t < 8
R/t = 2.1
0.0015
R/t = 4.2
0.001 R/t = 6.3
R/t = 8.5
0.0005 R/t = 16.9

0
0.3 0.5 0.7 0.9 1.1 1.3 1.5
Normalized Back Force

Fig. 10. Variation of measured anticlastic curvature with normalized back force.

of three sheet alloys: DQSK and HSLA steels, and 6022-T4 aluminum alloy. The following conclu-
sions were reached:
1. The shape change associated with simple bending springback is much smaller than that associated
with sidewall curl because of the extent of the a;ected area. The remaining conclusions are derived
from overall shape change, which is dominated by sidewall curl.
98 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

R/t = 3.5
R/t = 6.9
R/t = 10.4
R/t = 13.9
R/t = 27.8

Fig. 11. Variation of primary-to-anticlastic curvature with normalized back force for 6022-T4 aluminum.

2. As consistently reported in the literature, applied tension drastically reduces springback, although
an unexpected drop and plateau region of smaller dependence appears for back forces in the range
of 0.8–1.2 times the material yield stress.
3. For 6022-T4, the rapid drop of springback angle at back forces near yield corresponds to the
sudden increase of persistent anticlastic curvature. This appears to be the Frst measurement of
persistent anticlastic curvature in connection with springback, and its correlation with process
variables.
4. Friction in the range normally encountered in sheet forming practice has little e;ect on springback,
although very low friction conditions increase springback for 6022-T4. This conclusion di;ers from
some results appearing in the literature [13], presumably because for many experiments the sheet
tension cannot be controlled independently of friction, as with the current one.
5. Springback decreases for larger tool radii (R=t greater than about 5). This result is in agreement
with some of the literature [20] and is contrary with other reports [26]. For the steels, an inKec-
tion occurs for R=t near 5, and the trend reverses, with trends toward zero springback for R=t
approaching 2. This trend agrees with reports in the literature [18,22,24]. There appears to be an
opportunity of reducing springback by using very small tool radii.

Acknowledgements

We would like to thank the support of National Institute of Science and Technology for an
Advanced Technology Project under the auspices of PNGV, CAMMAC (Center for Advanced Ma-
terials and Manufacturing of Automotive Components), and Dr. Kaiping Li for helpful discussions.
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 99

References

[1] Schroeder W. Mechanics of sheet-metal bending. Transactions of the American Society for Mechanical Engineers
1943;65:817–27.
[2] Sachs G. Principles and methods of sheet metal fabricating, vol. 100. New York: Reinhold Publishing Corp., 1951.
[3] Levy BS. Empirically derived equations for predicting springback in bending. Journal of Applied Metalworking
1984;3:135–41.
[4] Nadai A. Theory of Kow and fracture of solids, vol. 1. New York: Ronald, McGraw-Hill, 1950.
[5] Phillips A. Introduction to plasticity. New York: Ronald, 1956.
[6] Huang M, Gerdeen JC. Springback of doubly curved developable sheet metal surface. Analysis of Autobody Stamping
Technology, Society of Automotive Engineers 1994;103:125 –38.
[7] Yu TX, Zhang LC. Plastic bending, theory and applications. Singapore: World ScientiFc, 1996.
[8] Yu TX, Johnson W. Cylindrical bending of metal strips. Material Technology 1983;10:439–47.
[9] Yuen WY. Springback in the stretch-bending of sheet metal with non-uniform deformation. Journal of Materials
Processing Technology 1990;22:1–20.
[10] Sanchez L. RD, Robertson JC. Gerdeen Springback of sheet metal bent to small radius=thickness ratios. SAE Paper
960595, Warrendale, PA, Society of Automotive Engineers, 1996.
[11] Sudo C, Kojima M, Matsuoka T. Some investigations on elastic recovery of press formed parts of high: : : : Eighth
Biennial Congress of IDDRG, 1974. p. 192–202.
[12] Chakhari ML, Jalinier JN. Springback of complex parts. Thirteenth Biennial Congress of IDDRG, 1984. p. 148–59.
[13] Hino R, Goto Y, Shirashi M. Springback of sheet metal laminates subjected to stretch bending and the subsequent
unbending. Advanced Technology of Plasticity 1999;II:1077–82.
[14] Zhang ZT, Hu SJ. Mathematical modeling in plane strain bending. SAE Technical Publication No. 970439. Society
of Automotive Engineers, 1997.
[15] Wang NM. Predicting the e;ect of die gap on Kange springback. Proceedings of the 13th Congress of the IDDRG.
Melbourne, Australia, 1984. p. 133– 47.
[16] Kuwabara T, et al. 2-D springback analysis for stretch-bending processes based on total strain theory. Automotive
Stamping Technology, SP-1067, SAE paper 950691. Warrendale, PA: Society of Automotive Engineers, 1995. 1–10.
[17] Ueda M, Ueno M, Kobayashi M. A study of springback in the stretch bending of channels. Journal of Mechanics
Working Technology 1981;5:163–79.
[18] Davies RG. Sidewall curl in high-strength steels. Journal of Applied Metalworking 1984;3:120–26.
[19] Ayres RA. SHAPESET: a process to reduce sidewall curl springback in high strength steel rails. Journal of Applied
Metalworking 1984;3:2:127–34.
[20] Bayraktar E, Altintas S. Square cup deep drawing and 2d-draw bending analysis of HadFeld steel. Journal of Materials
Processing Technology 1996;60:183–90.
[21] Schmoeckel D, Beth M. Springback reduction in draw-bending process of sheet metals. Annals of the CIRP
1993;42:1:339 – 42.
[22] Hayashi Y. Analysis of surface defects and side wall curl in press forming. IDDRG 13th Biennial Congress, 1984.
p. 565 –80.
[23] Pourboghrat F, Chu E. Springback in plane strain stretch=draw sheet forming. International Journal of Mechanical
Sciences 1995;36:3:327–41.
[24] Umehara Y. Technologies for the more precise press-forming of automobile parts. Journal of Materials Processing
Technology 1990;22:239–56.
[25] Makinouchi AE, Nakamachi E, Onate, Wagoner RH, editors. NUMISHEET ’93. The Institute of Physical and
Chemical Research, 1993.
[26] Nine HD. In: Koistnen DP, Wang N-M, editors. Drawbead forces in sheet metal forming. Mechanics of sheet metal
forming. New York: Plenum, 1978. p. 179–211.
[27] Nine HD. The applicability of Coulomb’s law to drawbeads in sheet metal forming. Journal of Applied Metalworking
1982;2:200–10.
[28] Nine HD. In: Wagoner RH, editor. Testing lubricants for sheet metal forming, novel techniques in metal deformation
testing. The Metallurgical Society of AIME, Warrendale, PA, 1983. p. 31– 46.
100 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101

[29] Wang NM. A mathematical model of drawbead forces in sheet metal forming. Journal of Applied Metalworking
1982;2:193.
[30] Liu YC. The e;ect of restraining force on shape deviations in Kanged channels. American Society for Mechanical
Engineers—Journal of Engineering and Material Technology 1988;110:389–94.
[31] Kuwabara T, Takahashi S, Ito K. Springback analysis of sheet metal subjected to bending–unbending under tension
Part II (experimental veriFcation). Advanced Technology of Plasticity. In: Altan T, editor. Proceedings of the 5th
ICTP, vol. II. Columbus, Ohio, 1996. p. 747–50.
[32] Wenzlo; GJ, Hylton TA, Matlock DK. A new procedure for the bending under tension friction test. Journal of
Material Engineering and Performance 1992;1:5:609–13.
[33] Haru; JP, Hylton TA, Matlock DK. Frictional response of electrogalvanized sheet steels. The physical metallurgy
of zinc coated steel. The Minerals, Metals and Materials Society 1993.
[34] Vallance DW, Matlock DK. Application of the bending-under-tension friction test to coated sheet steels. Journal of
Material Engineering and Performance 1992;1:5:685–93.
[35] Baba A, Tozawa Y. E;ect of tensile force in stretch-forming process on the springback. Bulletin of JSME 1964;7:
834–43.
[36] Tozawa Y. Forming technology for raising the accuracy of sheet-formed products. Journal of Materials Processing
Technology 1990;22:343–51.
[37] Zhang ZT, Lee D. E;ect of process variables and material properties on the springback of 2d-draw bending parts.
Automotive Stamping Technology 1995;11–8.
[38] Kuwabara T, Takahashi S, Ito K. Springback analysis of sheet metal subjected to bending–unbending under tension
Part I (theory and results of numerical analysis). Advanced Technology of Plasticity. In: Altan T, editor. Proceedings
of the 5th ICTP, vol. II. Columbus, Ohio, 1996. p. 743– 6.
[39] Tang SC. Application of an anisotropic hardening rule to springback prediction. Advanced Technology of Plasticity.
In: Altan T, editor. Proceedings of the 5th ICTP, vol. II. Columbus, Ohio, 1996. p. 719 –22.
[40] Focellese A, Fratini L, Gabrielli F, Micari F. The evaluation of springback in 3d stamping and coining processes.
Journal of Materials Processing Technology 1998;80 –81:108–12.
[41] Kuwabara T, Seki N, Takahashi S. A rigorous numerical analysis of residual curvature of sheet metals subjected
to bending–unbending under tension. In: Geiger M, editor. Advanced Technology of Plasticity 1999, vol. II. Berlin:
Springer, 1999. p. 1071–5.
[42] Balakrishnan V. Measurement of in-plane Bauschinger e;ect in metal sheets. MS Thesis, The Ohio State University,
1999.
[43] Kuwabara T, Morita Y, Miyashita Y, Takahashi S. Elastic–plastic behavior of sheet metal subjected to in-plane
reverse loading. Journal of Japanese Society of Technology and Plasticity 1995;36:769 –74 (in Japanese).
[44] Jiang S. Springback investigations. MS Thesis. The Ohio State University, 1997.
[45] Shen Y. The reverse-bend test: indirect measurement of the bauschinger e;ect in metal sheets. MS Thesis. The Ohio
State University, 1999.
[46] Zhang LC, Lu G, Leong S. V-shaped sheet forming by deformable punches. Journal of Materials Processing
Technology 1997;63:134–9.
[47] Zhang LC, Lin Z. An analytical solution to springback of sheet metals stamped by a rigid punch and an elastic die.
Journal of Materials Processing Technology 1997;63:49–54.
[48] Chou I, Hung C. Finite element analysis and optimization on springback reduction. International Journal of Machine
Tools and Manufacture 1999;39:517–36.
[49] Hishida Y, Wagoner RH. Experimental analysis of blank holding force control in sheet forming, SAE Paper Number
930285, Sheet Metal and Stamping Symposium, SAE SP-944, Warrendale, PA, 1993. p. 93–100.
[50] Sunseri M, KaraFllis AP, Cao J, Boyce MC. Methods to obtain net shape in aluminum sheet metal forming using
active binder control. Mechanics in Materials Processing and Manufacturing (ASME) 1994;194:167–84.
[51] Sunseri M, Cao J, KaraFllis AP, Boyce MC. Accommodation of springback error in channel forming using active
binder force control: numerical simulations and experiments. Transactions of ASME 1996;118:426–35.
[52] Han SS, Park KC. An investigation of the factors inKuencing springback by empirical and simulative techniques.
Proceedings of Numisheet’99, Besancon, France, 1999. p. 53–7.
[53] Nagai Y. High precision U-bending technique for moderately thick plate. Japanese Society of Technology and
Plasticity 1987;28:143–9 (in Japanese).
 W.D. Carden et al. / International Journal of Mechanical Sciences 44 (2002) 79–101 101

[54] Kutt LM, Nardiello JA, Ogilvie PL, Pifko AB, Papazian JM. Non-linear Fnite element analysis of springback.
Communications in Numerical Methods in Engineering 1999;15:33–42.
[55] United States Steel Corporation (US Steel) Material data sheet provided with the material, 1996.
[56] Aluminum Company of America (ALCOA) Material data sheet provided with the material, 1996.
[57] Annual Book of ASTM Standards, American Society of Testing and Materials, West Conshohocken, PA, vol. 3.01,
Standard E-8M-96, 76-96.
[58] Hollomon JH. Tensile deformation. AIME Transactions 1945;162:268.
[59] Swift HW. Plastic instability under plane stress. Journal of Mechanics and Physics of Solids 1952;1:1.
[60] Voce E. The relationship between stress and strain for homogeneous deformation. Journal of the Institute of Metals
1948;74:537– 62,760–71.
[61] Carden WD, Wagoner RH, Matlock DK. Unpublished research.
[62] Parco Prelube MP 404, manufactured and distributed by Parker Amchem, Henkel Surface Technologies Corporation,
32100 Stephenson Highway, Madison Heights Michigan, 48071, phone: (810) 583-9300.
[63] Wagoner RH, Carden WD, Carden WP, Matlock DK. Springback after drawing and bending of metal sheets. In:
Chandra T, Leclair SR, Meech JA, Verma B, Smith M, Balachandran B, editors. Proceedings of the IPMM ’97—
Intelligent Processing and Manufacturing of Materials, University of Wollongong, 1997, vol. 1. p. 1–10.
[64] Li KP, Wagoner RH. In: Huetink J, Baaijens FPT, editors. Simulation of springback. Simulation of materials
processing. Rotterdam: A.A. Balkema, 1998. p. 21–32.
[65] Li KP, Wagoner RH. Simulation of deep drawing with various elements. In: Gelin JC, Picart P, editors. Proceedings
of NUMISHEET’99. University of Franche-Compte, Besancon, France, 1999. p. 151– 6.
[66] Li KP, Geng LM, Wagoner RH. Simulation of springback with the draw=bend test. In: Meech JA, Veiga MM, Smith
MH, LeClair SR, editors. IPMM’99, 1, IEEE. Vancouver, BC, Canada, ISBN 0-7803-5489-3.
[67] Li KP, Geng LM, Wagoner RH. Simulation of springback: choice of element. In: Geiger M, editor. Advanced
technology of plasticity 1999, vol. III. Berlin: Springer, 1999. p. 2091–8.
[68] Geng LM, Wagoner RH. Springback analysis with a modiFed combined hardening law, sheet metal forming: Sing
Tang 65th Anniversary Volume, SP-1536, SAE Publication 2000-01-0768. Warrendale, PA: Society of Automotive
Engineers, 2000. p. 21–32.
[69] Wenner ML. Private communication, General Motors Corporation, March, 1996.
[70] Mickalich MK, Wenner ML. Calculation of springback and its variation in channel forming operations SAE
Paper 880526, in P-206, Advances and Trends in Automotive Sheet Metal Stamping. Warrendale, PA: Society
of Automotive Engineers, 1988. p. 99 –100.
[71] Thompson NE, Ellen CH. A simple theory for “side-wall curl”. Journal of Applied Metalworking 1985;4:1:39–42.
[72] Horrocks D, Johnson W. On anticlastic curvature with special reference to plastic bending: a literature survey and
some experimental investigations. International Journal of Mechanical Sciences 1967;8:835–61.
[73] Young WC. Roark’s formulas for stress and strain. 6th ed. Table I, Case 20. New York: McGraw-Hill, 1989. p. 68.