You are on page 1of 22

Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007.

Measuring Stress Strain Curves to Large 1


Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

MEASURING STRESS STRAIN CURVES TO LARGE STRAINS ON SHEET


METAL

A. Bacha, M. Feuerstein, Ch. Desrayaud and H. Klöcker


Ecole Nationale Supérieure des Mines de Saint-Etienne
CNRS – UMR 5146
158 Cours Fauriel 42023 Saint-Etienne Cedex 2 France

Microstructures and Processing Department, Ecole des Mines de St Etienne, France


and CNRS Federation 2145, UMR 5146

Keywords : Large deformation, stress strain curve, aluminium automotive alloy,


rolled sheet

Abstract

The stress-strain response of aluminium sheet has been determined by a novel plane
strain compression test on laminated samples up to equivalent strains of 1.2. The test sample
is composed of several sheet layers glued together, machined to shape and then compressed in
a lubricated channel die along the “thin” directions. This simple test has been validated by a
comparison of FE simulations and experimental results. The stress-strain curves can be
considered accurate up to strains of 1.2.

1. INTRODUCTION

Large strain deformation testing on rolled sheets has become an important


requirement for the development of constitutive equations as used, for example, in finite
element simulations of shaping operations such as stamping, deep drawing or cutting. During
deep drawing or cutting operations local strains up to 2 are reached. Currently, no technique is
available for such large strain testing on rolled sheets. Due to necking, only strains of about
0.5 can be attained in classical tension tests on sheet specimens. Siebel [1] and Bridgman [2]
discussed the measurement of true stress strain curves in the neck of axisymmetric tensile
specimens and Needleman [3] and Le Roy [4] studied the necking of axisymetric specimens
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 2
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

by finite element analysis. Recently, Zhang et al. [5] showed how to determine material true
stress vs. true strain curves from tensile specimens with rectangular cross-section. But, in
tensile samples which neck, the strain is limited due to damage and there are practical
difficulties in measuring the local strain and stress. Standard uniaxial compression tests on
cold rolled sheets (typically 1mm thickness) are also impossible due to friction. Finally,
torsion tests on cold rolled 1mm-thick sheets would lead to extremely small sample sections.

The present work shows how to obtain σ(ε) curves for large strains (ε >1.2) on thin
metallic sheets, in this case 1mm thick aluminium sheets. It is proposed that large strain room
temperature deformations can be achieved by channel die multiple pass compression on
laminated samples.

Plane strain compression (PSC) has often been used to simulate metal behaviour
during hot rolling. The PSC “punching” technique is the most widely used for hot
deformation tests. It consists of compressing a preheated thick plate sample by means of a
punching tool which is longer than the sample. Half a century ago, Watts and Ford [6]
introduced this punching technique to determine the room temperature behaviour of strip
metal. However using this technique, the sheets can only be compressed along the normal
axis. The maximum strain is much less than unity and the deformed sample volume is very
small.

The channel-die test is an alternative plane strain compression test without any
lateral spreading, but which has so far been relatively little employed. Maurice and Driver
[7,8,9] have developed an original hot channel-die equipment which has the advantages of
imposing true plane strain compression without major friction. Channel die tests have been
used for room temperature testing on mono-block samples [10,11]. The present work shows
how the channel die equipment can be used to perform compression tests on laminated
samples of thin Al-alloy sheets.

2. TESTING TECHNIQUES

2.1. The channel die test rig

Figure 1 shows a schematic diagram of the channel-die apparatus first described by


Maurice and Driver [7]. The channel is composed of 3 separate steel plates of which the
centre plate has the same thickness (7mm) as the punch width. The equipment of Figure 1 was
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 3
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

installed in an a Schenck 100kN servohydraulic machine whose electronic and hydraulic


control systems enable constant plastic strain rate tests up to about 20sec-1 (according to
sample dimensions). Loads are measured by standard load cells and displacements by means
of a Linear Variable Differential Transformer (LVDT transducer) situated between the
crosshead and the base of the channel-die. More details concerning the channel die equipment
can be found in [7] and [9].

2.2. Plane strain compression samples

The PSC samples are 7mm wide (i.e. the channel width), 9 mm high (the
compression direction) and 10 long (direction of elongation). The sample dimensions were
optimised previously on solid mono-block samples in order to minimize friction effects and
deformation heterogeneities [9]. In the present work, the PSC samples consist of 7 layers of
1mm thick Al alloy sheets (Figure 2). Two reference frames are used to describe the sample
orientation. {RD, ND, TD} corresponds to the original rolling, normal and transverse
directions, whereas {x, y, z} correspond to the equivalent rolling, transverse and compression
directions of the PSC test (Fig. 2b).

The sample preparation was as follows. The sheets were machined to obtain seven
strips 1mm thick 9.01 mm high and 10.01 mm long. The seven strips placed on a standard
square were bonded with a cyanoacrylate glue. The thickness of the layered composite (alloy
sheets and glue) was determined with a fine micro-screw. The composite was then embedded
in a transparent acrylic resin. Two opposite faces (original RD-ND-plane) were polished in an
automatic polishing machine with pressure control on each sample. The sample is polished
under abrasive P280 to obtain the desired dimensions. Surface finishing was done with an
abrasive P1000. After decasting, the sample was embedded in a new acrylic resin and the 2
faces corresponding to the original RD-TD-plane were polished similarly. The two sample
faces corresponding to the original ND-TD-plane were polished manually since they are free
from any contact during channel die compression.

The thickness of the glue layer between two adjacent sheets is less than 10 µm
(Figure 2c). The area reduction due to the glue layer is thus smaller than 1 per cent.

2.3 Lubrication
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 4
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

Since four major surfaces of the sample are in contact with the tools, lubrication is a
critical aspect of the channel-die test. The sample is wrapped in PTFE (Teflon) ribbon (50µm
thickness). PTFE is known to possess very low friction coefficients. However, although PTFE
stretches and deforms with the sample as required it can undergo localised tearing at strains
typically of unity. Therefore two step deformations were used. During the first pass samples
were deformed up to strains of about 0.6.

The glue allows some relative movement of the different sample layers. But, after the
first compression pass, the glue interface still holds. After this pass, the sample ends were cut
and the sample was wrapped in a new PTFE ribbon.

3. RESULTS

Two aluminium alloys, AA5182 alloy and AA6016, widely used in sheet form
particularly in the automotive industry, have been deformed by channel die compression. For
each alloy two different grades, denoted a and b, were considered. Table 1 gives the chemical
composition of the alloys. Unless indicated otherwise the samples were deformed at a
constant strain rate of 10-1s-1. The AA6016 samples were deformed by single and two step
compression without any distinction between the rolling direction (RD) and the transverse
direction (TD).

The AA5182 samples were deformed by two step compression, distinguishing


between compression along the rolling direction (RD) and the transverse direction (TD). The
parameters of a modified Voce law [11,12] which closely describes the material behaviour of
these alloys, were determined experimentally as described below.

3.1. Overall load vs. overall displacement curves

Figure 3 shows typical load displacement curves corresponding to compression tests


on the AA6016 laminated sheet samples. Single pass and two step channel die compression
tests were done at constant strain rate. To obtain reliable stress strain curves, the displacement
and the overall load have been corrected as described below.

Displacement corrections

The displacement was corrected to take into account the deformation of the test rig
and the PTFE-layer [10] :
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 5
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

Qmeasured h teflon
d corrected = d measured − 1− (1)
k h0

where h0 and hteflon are the initial sample height and PTFE-layer thickness. dmeasured is the
displacement measured by the LVDT and dcorrected takes into account the deformation of the
PTFE layer. Qmeasured is the measured load. k is the stiffness of the test rig and was determined
by measuring the punch displacement without a sample present in the channel (k = 91
kN/mm).

Load corrections

The measured load has been corrected to take into account friction effects [10] as
follows :
Qmeasured
Qcorrected = (2)
W
1+ F
WP

where Qcorrected is the corrected value of the applied load Qmeasured. WF and WP are the work
rates associated respectively to friction and to plastic deformation. The ratio between these
work rates was estimated by [10] as

WF m l3 Z + 2h Z +l
= ln + 2 h 3 ln + 2 lhZ + l 2 e0 (3a)
WP 12 l0 h0 e0 4 Z − 2h Z −l

Z = l 2 + 4 h2 (3b)

where m is the Tresca friction coefficient whose value was estimated as described in section
4.3. h0, l0, and h, l are respectively the initial and current sample height and length. e0
corresponds to the sample and channel width (7mm).

3.2. Equivalent stress vs. equivalent deformation curves

The anisotropy of the (AA5182 and AA6016) sheets is small and will be neglected in
the following. Brunet et al [14] determined the Langford coefficient of the AA6016 sheets by
tension tests. For the two grades of AA6016 the relation between Hill’s stress function [15]
and the von Mises stress function is given by :

σ Hill = 0.98σ
(4)
ε Hill = 1.03ε
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 6
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

were σ Hill and ε Hill correspond to Hill’s stress function and strain rate respectively. σ and
ε correspond to von Mises stress and strain rate respectively. To determine the equivalent
stress σ and the equivalent strain ε , the overall load was corrected by equations (2,3) with a
value of m =0.02 (see section 4.3). However the shear stresses and strains due to friction were
neglected, i.e. the stress tensor and the strain rate tensor were assumed diagonal. Thus, the
equivalent von Mises stress and strain may be written :

3 Qcorrected d 2 h0
σ= 1 − corrected ε= ln (5)
2 e0l0 h0 3 h 0 − d corrected

Figure 4 shows equivalent stress vs. equivalent strain curves for the two grades of
AA6016.The overall stress vs. overall strain curves corresponding to single pass compression
are obtained straightforwardly by applying expression (5) to the corresponding load-
displacement curves after displacement (1) and load corrections (2,3). This procedure leads to
the equivalent stress equivalent strain curves marked “single pass” on figures 4a,b.

The same corrections (friction and test rig + PTFE deformation) were applied to the
load displacement curves of the two step compression tests. In this way the two branches
marked “pass 1/2 “ and “pass 2/2 (initial)” are obtained. In order to correctly match the two
branches of the equivalent stress equivalent strain curves, the elastic unloading of branch 1
(pass ½) and the elastic reloading of pass 2 (pass 2/2) should coincide. The curve
corresponding to pass ½ was used up to the maximum value of the equivalent stress (point A).
The unloading at the end of pass 1 was considered purely elastic. Fitting a straight line to the
results of the 2nd pass allows to determine the corresponding elastic part (OC on Fig. 4). The
curve corresponding to the second pass is shifted horizontally in order to bring C to B on
figure 4. AB is a vertical line segment. At the inter-pass a stress drop AB due to recovery may
be observed for certain Al-alloys.

3.3. Large strain material behaviour : extended Voce law

Determining the coefficients of the extended Voce law

For finite element simulations of shaping and deep drawing operations, the flow
stress has to be accurate at very large plastic strains. The aim of this paragraph is not to
discuss particular material laws but to illustrate one possibility of accurately fitting the
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 7
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

experimental data over a large strain range. To represent the material flow stress, σ Y , at
large deformations an extended Voce law has been chosen.

( )
σ Y = σ Yo + (σ ∞ − σ Yo ) 1 − exp ( −δ ε p ) + α ε p
σY (6)
εp = ε −
E

E is the Young’s modulus. ε and ε p are respectively the total and the plastic equivalent
strain. σ Yo is the initial flow stress. σ ∞ corresponds to the stress at very large plastic strains in
the Voce law. α ε p is an additional term accounting for small positive strain hardening at very
large plastic strains.

Single and two pass compression tests were carried out on AA6016 samples (figure
5a,b). On AA5182 samples two pass compressions along the rolling (RD) and the transverse
direction (TD) were performed (figure 5c,d). The initial flow stress σ Yo was determined
previously in simple tension. A straight line was drawn through the results of the second pass
(between points A and B on fig. 5c,d). This procedure gives a precise value of α. σ ∞ and δ
were determined by minimizing the total error between the extended Voce law and the two
pass experimental results. Table 2 summarizes the coefficients of the different Voce laws. The
most significant results are the different values of the α parameter for 5182 and 6016 alloys.
A small “difference” between the generalized Voce law and the experimental data for the two
6016 grades is observed. It can also be seen that the 6016 alloy undergoes softening at strains
about 0.5. This softening is analysed in section 4.3.b.

4. MECHANICAL ANALYSIS

4.1. Aim of the mechanical model

First, the friction between the sample and the channel die test rig has been analysed.
The value of the coefficient m describing friction between the sample and the test rig was
determined by comparing experimental and simulated load displacement curves.

Secondly, both the friction between the 7 sample layers and the strain homogeneity
of the sample have been analysed. The coefficient m layer describing friction between the
different sample layers was determined by comparing experimental and simulated deformed
sample shapes.

4.2. The finite element models


Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 8
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

Two finite element simulations of the channel die compression test were considered
here. The first model was used to analyse friction between the sample and the channel die
punch and thereby determine m . The second model was used to analyse slip of the 7 sheet
layers relative to each other. In both models, only the sample was meshed; the die and punch
were represented as non deformable surfaces. The Abaqus® explicit formulation with linear 8-
node elements and reduced integration was used for both models.

The friction between the sample and the test rig was modelled with a mono-block
sample meshed with respectively 14, 16 and 19 elements in the sample width (x), length (y)
and height (z).

Relative slip of the sheet layers was analysed by modelling each layer with 2
elements in its width. The elements were cubes with a length of 0.5 mm. Thus, in this
analysis, the whole sample was simulated with respectively 14, 20 and 16 elements in the
sample width (x), length (y) and height (z). Figure 6 shows the deformed meshes of the mono-
block (a) and the laminated sample (b).

4.3. The load and displacement correction

a) The coefficient m describing friction between sample and test rig

The friction coefficient between the sample and test-rig m was assumed
independent of the sample deformation. Coulomb’s friction model was used. The following
relation between the Coulomb friction coefficients µ and the Tresca friction coefficient m was
taken.
m
µ pb =
2 1 − m2
(7)
m
µlat =
1 − m2

µpb represents the friction between the lower sample surface and the channel and between the
upper sample surface and the punch. µlat represents friction between the lateral sample
surfaces and the die.

The value of m was determined with the mono-block model. The load displacement
curves corresponding to the mono-block model were determined for several values of m and
compared to the experimental two-passes load displacement curves. The m -value finally
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 9
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

retained (i.e. m=0.02 ) corresponds to the minimum distance between the experimental and
simulated load displacement curves.

Figure 7 shows experimental (single and two passes) and finite element mono-block-
model results (with m=0.02 ) for the two grades of AA6016. At small strains, a good
correspondence between all the curves is observed. For displacements smaller than 2 mm
( ε ≤ 0.3 ) the friction correction is negligible. For displacements larger than 2 mm, a
difference between the single pass results and the finite element results is observed. But a
good agreement between two pass results and the finite element simulation is observed for
displacements up to 6 mm ( ε ≤ 1.2 ). Thus the 2 pass channel die compression test leads to
reliable load displacement curves up to deformations of 1.2.

b) Reproducing the extended Voce law

Figures 5 show that the experimental σ (ε ) curves are well represented by the
extended Voce law. However, as noted above, the 6016 alloy exhibits softening at ε ≈ 0.5
seen in the σ (ε ) curve. This behaviour can be reproduced from the Voce law by means of a
F.E. simulation assuming a small friction coefficient. Figure 8 shows the result of these
simulations for the mono-block model and a m = 0.02 and confirms this softening in 6016
and the absence of softening in the 5182 alloy.

4.4. Homogeneity of the deformation vs. slip between the different sample layers

The aim of present work is to determine the real stress strain curve of the sheet
material. Thus, an important question to answer is : Do the layers deform in an identical
fashion and if not does this have any influence on the load displacement curves ? This
problem is addressed in following section.

a) Validity of the load displacement curves

The influence of possible slip between different sample layers on their deformation
mode was analysed by the second finite element model. Friction between the sample and the
test rig was characterized by the above coefficient m = 0.02 . The friction between different
layers was described by the coefficient mlayer. Four very different values for the coefficient
-4 -3 -2 -1
mlayer were tested (10 , 10 , 10 , 10 ). The load displacement curves obtained with
different values of mlayer were compared to the experimental curve and to the load
displacement curve obtained with the mono-block simulation. Figure 9 shows the load
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 10
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

displacement curves corresponding to the two AA5182 grades. A perfect correspondence


between experimental and simulated load displacement curves is observed for all mlayer
values. Whatever the value of the inter-layer-friction coefficient the stress strain curves of the
layered sample are identical to those of the mono block.

b) Homogeneity of the sample deformation

Now, we are sure that the PSC test on laminated samples gives reliable stress strain
curves up to strains of 1.2. An other important question is : Is the PSC test on laminated
samples suited for microstructural analyses ?

Figure 10 shows a sample after an overall deformation of 1.2. Obviously some slip
occurred, but the sample layers are still bonded (glued together). The experimental behaviour
appears to be between the mono-block model (zero slip between the different sample layers)
and a layered model without friction.

To analyse the effect of slip between different sample layers, the shear strain
distribution in the sample was analysed. Figure 11 shows finite element simulations of the
shear strain distributions for both the layered sample and the mono block model. Very
different values of mlayer have been used to evaluate the effect of mlayer on the deformation
heterogeneity. The deformation is homogeneous in the sample centre (over 80 % of the sample
volume) for all the considered values of mlayer . Large shear strains are observed only at both
ends of the sample.

The 3 conditions (mono-block and laminated sample with mlayer=0.001, 0.1 ) lead to
quasi identical shear strain distributions in the layers close to the surface and the central layer.
For the smallest value of mlayer a difference between the sample central layer and the surface
layer is observed. For mlayer= 0.1 , no difference between the shear distribution at the sample
centre and surface is observed.

As the sheared sample volume is always small, the work rate due to friction ( WF ) is
always small compared to the plastic work rate ( Wp ) and the load displacement curves
obtained with the mono-block and the layered sample models coincide, as was shown in
Figure 9.

5. DISCUSSION
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 11
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

5.1. Importance of material testing to large deformations

Before discussing the reliability of the above results, it is important to recall the
problem of large strain testing on sheet metals. In sheet metal forming (deep drawing, cutting,
…) a reasonably high value of the strain hardening parameter is necessary to prevent strain
localization and necking. Thus to determine limit forming diagrams, a precise knowledge of
the stress strain curve at large strains is required. Here, the Voce law (6) with the
experimentally determined parameters (Table 2) was used to calculate the strain hardening
( )
parameter n = ∂ ( ln σ Y ) / ∂ ln ε p for the tested alloys. Figure 12 shows the variation of the
strain hardening parameter n vs. the equivalent von Mises deformation for both the AA6016
and the AA5182 grades.

Both the real variation of the strain hardening parameter with the overall deformation
and the strain hardening parameter obtained with a Voce law without the additional linear
term are shown in Figure 12. Obviously, neglecting the linear term (the large strain part of the
stress strain curve) leads to incorrect values of the strain hardening parameter n. The
Considère strain at the onset of necking in simple tension is much less than 0.18 for all the
grades considered. Figure 12 obviously confirms that one cannot extrapolate to large strains
the results of tensile test up to necking.

5.2. Reliability of the results

We believe that this work has demonstrated the potential of using the channel-die
compression test for detailed characterization of the mechanical response of sheet metal alloys
undergoing large strain compression at room temperature. The finite element analysis shows
that reliable flow curves can be obtained by the PSC test on laminated samples. Also as a
large part of the sample volume deforms homogeneously irrespective of the interlaminate
(glue) behaviour, the plane strain compression test also seems indicated for microstructural
analysis under larger deformation. The channel die compression tests works over large strains
essentially because the material is constrained, in compression along two faces, and can only
elongate in one direction. In this sense the glue has no real mechanical role during
compression but is indispensable to sample preparation.
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 12
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

The present method has been used for true strains up to about 1.2 by 2 pass
deformations. It is quite feasible to go to higher strains by multiple pass deformations. The
present test may be used to determine equivalent stress strain curves of metal alloy sheets
other than aluminium.

5.3. Comparison with other “large” strain testing techniques

Recently Vegter et al. [16] determined the initial yield stress of sheet metal by
several tests (tension, plane strain tension, compression of layered samples and shear tests).
The tension and compression tests are limited to strains smaller than 0.2 and the shear test is
limited to strains smaller than 0.4.

Barlat et al. [17] measured the initial yield surface by biaxial compression tests on
cubic specimens made from laminated binary Al-Mg alloy sheet samples. Baralat et al. used a
biaxial compression testing machine. Because of the physical constraints of the loading dies,
deformation step was limited to a strain of about 0.1. This technique is more expensive than
the simple one presented here.

To our knowledge all previous methods (tension, shear, compression) enable one to
measure the initial yield strength but are not very useful for analysing the strain hardening of
the material. The present method is therefore complementary to these techniques, but
particularly useful for obtaining large strain material behaviour.

6. CONCLUSIONS

In the present work, a reliable method for determining large strain (σ,ε)-curves on
sheet metal has been developed. Layered sheet metal samples have been deformed in channel
die (plane strain) compression up to strains of 1.2.

The parameters of a generalized Voce law were determined for two aluminium alloys
(AA5xxx and AA6xxx). The major importance of determining the strain hardening at large
strains has been confirmed.

The plane strain compression of layered samples was also analyzed by finite element
calculations. This mechanical analysis shows that, using Teflon film, the friction coefficient
between the sample and the tool is less than 0.02 and the friction coefficient between different
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 13
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

layers of the sample only play a very minor role. The sample deforms homogeneously over
more than 80 per cent of its volume.

This work can be developed further. The present work has been limited to 1mm thick
aluminium sheet tested at room temperature. Very thin sheets (of thickness less than 1 mm)
are used widely in industry. The stress strain curves of these sheets are even more difficult to
obtain then the stress strain curves of 1mm thick sheet metal. Determining the minimum layer
thickness that avoids instabilities and gives reliable results in plane strain compression would
be interesting for very thin sheet metal.

The present work concerned room temperature compression. No microstructural


evolution of the material has been considered. Hot plane strain compression, with
temperatures up to 500°C of layered samples could be developed on the same test facility.
This is of course of primary interest for the hot working of sheet metal alloys.
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 14
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

Acknowledgements

The authors wish to thank Gilles Blanc for his help in sample preparation. The results
concerning the AA6016 grades have been obtained in the context of the PhD of Alexis Bacha
with Alcan (Centre de Recherche de Voreppe). The results concerning the AA5182 grades
have been obtained during the PhD of Matthieu Feuerstein. This research was partially funded
by the Région Rhône Alpes (“Rupture des particules intermétalliques dans des alliages légers:
amélioration de la formabilité et de l’aspect de surface des tôles laminées”) The authors
would like to thank the Alcan (Centre de Recherche de Voreppe) for for their collaboration
and support. Funding by the Région Rhône Alpes is gratefully acknowledged.

References

References
[1] E. Siebel and A. Pomp, Mitteilungen Kaiser-Wilhelm Inst. für Eisenforschung,
1927, vol. 9, p. 157-165.
[2] W. Bridgman, 1944, The stress distribution at the neck of a tension specimen, Trans.
Am. Metals, vol. 32, pp. 553-574.
[3] A. Needleman, 1972, A numerical study of necking in circular cylindrical bars, J.
Mech. Phys. Solids, vol. 20, pp 111 – 127.
[4] G. Le Roy, J.D. Embury, G.Edwards, M.F.Ashby, ,Model of ductile fracture based
on the nucleation and growth of voids, Acta Metallurgica, 1981, vol. 29, pp. 1509-
1522.
[5] Z.L. Zhang, M.Hauge, J. Ødegård, C.Thaulow, 1999, Determining material true
stress-strain curves from tensile specimens with rectangular cross-section, Int. of
Solids and Structures, vol. 36, pp. 3497-3516.
[6] A.B. Watts, H. Ford, 1955, Proc. Inst. Mech. Eng., vol. 169, pp. 1141 – 1156.
[7] Cl. Maurice and J.H. Driver, 1993, High temperature plane strain compression of
cube-oriented aluminium crystals, Acta Metall. Mater., vol. 41, pp. 1653-1664
[8] K.F. Karhausen, J. Savoie, C.M. Allen, D. Piot and R. Luce, 2002, Material testing,
constitutive modelling and implementation of material models into hot rolling
models for alloy AA3103, Proc. ICAA8, Mater. Sci. Forum, vols. 396-402, pp. 371-
378.
[9] Cl. Maurice, D. Piot, H. Klöcker, J.H. Driver, 2005, Hot plane strain compression
testing of aluminium alloys by channel-die compression, Metallurgical and
Materials Transactions vol. 36A, pp. 1039 –1047.
[10] J.H. Driver, A. Skalli, 1983, “L' essai de compression plane de monocristaux
encastrés: méthode d' étude du comportement d' un cristal soumis à une déformation
plastique imposée“,Rev. Phys. Appl, vol. 17, pp. 447-451.
[11] B. Orlans-Joliet, J.H. Driver, F. Montheillet, 1990, “Plane strain compression of
silicon-iron single crystals”. Acta Metall. Mater., vol. 38, pp. 581-594.
[12] E. Voce, 1948, The relationship between stress and strain for homogeneous
deformation, J. Inst. Metals, vol. 74, pp. 537 - 562
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 15
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

[13] E. Voce, 1948, True Stress-Strain Curves and Their Application to Cold-working
Processes, Metal Treatment, vol. 15, pp. 53 – 60.
[14] M.Brunet, F.Morestin, Experimental and analytical necking studies of anisotropic
sheet metals, Journal of Materials Processing Technology, 2001, vol. 112, 214-226
[15] R. Hill, 1948, A theory of the yielding and plastic flow of anisotropic metals, Proc.
Roy. Soc., vol. 193, pp. 281-297.
[16] H. Vegter, C. H. L. J. ten Hor, Y An, E. H. Atzema, H. H. Pijlman, T. H. van den
Boogaard, H. Huetink, 2003, Characterisation and modelling of the plastic material
behaviour and its application in sheet metal forming simulation,VII International
Conference On Computational Plasticity, Complas VII, ed E. Onate, D.R. Owen,
Barcelona, pp. 1 – 20.
[17] F. Barlat, Y. Maeda, K. Chung, M. Yangawa, J.C. Brem, Y. Hayashida, D. J. Lege,
K Matsui, S.J. Murtha, S. Hattori, R.C. Becker and S. Makosey, 1997, Yield
function development for aluminum alloy sheets, J. Mech. Phys. Solids, vol. 45, no.
11/12, pp. 1727-1763.

Grade
Mg Si Cu Fe Mn
element
5182 (a) 4,25 0,09 0,06 0,29 0,35
5182 (b) 5,2 0,06 0,06 0,15 0.09
6016 (a) 0,33 1,03 0,12 0,3 0,11
6016 (b) 0,43 0,98 0,15 0,3 0,11

Table 1. : Chemical composition of the different grades

Alloy/grade E (GPa) σo (MPa) σ∞ (MPa) δ α (MPa)


AA5182a 70 130 264 14.6 101
AA5182b 70 105 253 14.95 95
AA6016a 130 310 16 15
AA6016b 70 248 12 14

Table 2. Voce law parameters for the of AA5182 and AA6016 grades.

Figure 1. Schematic of the channel-die equipment [7,9]


Figure 2. Layered sheet sample (a) 3D view, (b) view along the compression direction z, (c) SEM image of
interface between two sheets. {x, y, z} correspond respectively to the rolling, transverse and compression
directions of the PSC test.
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 16
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

Figure 3. Load displacement curves for the two AA6016 grades. Pass ½ means the first of two compression
passes etc.
Figure 4. Equivalent stress equivalent strain curves for the two AA6016 grades. To obtain these curves,
displacement (1) and load correction (2,3) with an m value of 0.02 were applied to the load displacement
curves of Figure 3.
Figure 5. Equivalent stress vs. equivalent strain curves for the two AA6016 and the two AA5182 grades:
Exprimental results and generalized Voce fit.
Figure 6. Deformed finite element meshes of (a) the mono-block model and (b) the layered model for a unit
equivalent strain ( ε = 1 ). The friction values of the friction coefficients are m=0.02 and mlayer=10 -3 .

Figure 7. Load displacement curves for the two AA6016 grades. Experimental single and two pass results. Finite
element results obtained with the generalized Voce law (6) applied to a mono-block sample with a friction
coefficient m = 0.02.
Figure 8. Comparison between experimental and simulated stress strain curves for the two AA6016 and the two
AA5182 grades. The geometrical softening observed experimentally for the AA6016 grades is reproduced by the
mono-block finite element model with the Voce law (6) and m = 0.02. For the AA5182 grades no softening is
observed.
Figure 9. Influence of mlayer value on the load displacement curves for the two AA5182 grades. A good
agreement between experimental data (A), mono-block sample (B) and layered sample finite element (C)
simulations is observed for mlayer = 0.0001 and mlayer =0.1. The value of m was taken as 0.02.

Figure 10. Deformed sample shape ( ε = 1.2 ) .


Figure 11. Finite element simulation of the shear strain distribution in a layered AA5182 sample deformed up to
an equivalent unit strain ε = 1 . The Voce law (6) and a friction coefficient m = 0.02 were used with two values
of mlayer , i.e. mlayer = 0.001 (a,b) and mlayer = 0.1 (c,d). (e) corresponds to the mono-block model. (a,c,e)
correspond to surface layers and (b,d) correspond to central layers.
Figure 12. Variation of the strain hardening parameter n with equivalent deformation for the 6016 (a) and the
5182 grades (b). The strain hardening parameter n is represented for the Voce law (6) and the parameters of
Table 2. The behaviour of a hypothetical strain hardening parameter for a Voce law without the αε p –term is
also represented.

Figure 1. Schematic of the channel-die equipment [7,9]


Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 17
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

-a- -b- -c-

Figure 2. Layered sheet sample (a) 3D view, (b) view along the compression direction z, (c) SEM image of
interface between two sheets. {x, y, z} correspond respectively to the rolling, transverse and compression
directions of the PSC test.

-a- -b-

Figure 3. Load displacement curves for the two AA6016 grades. Pass ½ means the first of two compression
passes etc.

-a- -b-

Figure 4. Equivalent stress equivalent strain curves for the two AA6016 grades. To obtain these curves,
displacement (1) and load correction (2,3) with an m value of 0.02 were applied to the load displacement
curves of Figure 3.
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 18
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

350 AA6016a
400 AA6016b
300 350

equivalent stress (MPa)

equivalent stress (MPa)


250 300
250
200
200
150
single pass 150 single pass
100 two passes two passes
Voce 100 Voce
50 50
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
equivalent strain equivalent strain

-a- -b-

400 400
AA5182a AA5182b
350 350
equivalent stress (MPa)

equivalent stress (MPa)


A B A
300 300 B
250 250
compression along RD compression along RD
200 compression along TD
200 compression along TD
150 Voce 150 Voce

100 100
50 50
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
equivalent strain equivalent strain

-c- -d-

Figure 5. Equivalent stress vs. equivalent strain curves for the two AA6016 and the two AA5182 grades:
Exprimental results and generalized Voce fit.

z z
y y

x x

-b-
-a-

Figure 6. Deformed finite element meshes of (a) the mono-block model and (b) the layered model for a unit
equivalent strain ( ε = 1 ). The friction values of the friction coefficients are m=0.02 and mlayer=10 -3 .
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 19
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

AA6016a AA6016b experim.


40 experim. 50

40
30

load (kN)

load (kN)
f.e.
30 f.e.
20
20
single pass
10 two passes 10
f.e. simulation

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
displacement (mm) displacement (mm)

-a- -b-

Figure 7. Load displacement curves for the two AA6016 grades. Experimental single and two pass results.
Finite element results obtained with the generalized Voce law (6) applied to a mono-block sample with a
friction coefficient m = 0.02.

350 AA6016a
400 AA6016b
300 350
equivalent stress (MPa)

equivalent stress (MPa)

250 300
FE
FE 250
200
200
150
single pass 150 single pass
100 two passes two passes
finite element simulation 100 finite element simulation
50 50
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
equivalent strain equivalent strain

-a- -b-
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 20
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

400 400
AA5182a AA5182b
350 350

equivalent stress (MPa)

equivalent stress (MPa)


A B FE A
300 300 B FE

250 250
compression along RD compression along RD
200 compression along TD
200 compression along TD
150 finite element simulation 150 finite element simulation

100 100
50 50
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
equivalent strain equivalent strain

-c- -d-

Figure 8. Comparison between experimental and simulated stress strain curves for the two AA6016 and the
two AA5182 grades. The geometrical softening observed experimentally for the AA6016 grades is reproduced
by the mono-block finite element model with the Voce law (6) and m = 0.02. For the AA5182 grades no
softening is observed.

-a- -b-

Figure 9. Influence of mlayer value on the load displacement curves for the two

AA5182 grades. A good agreement between experimental data (A), mono-block


sample (B) and layered sample finite element (C) simulations is observed for mlayer =

0.0001 and mlayer =0.1. The value of m was taken as 0.02.


Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 21
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

z x

10 mm

Figure 10. Deformed sample shape ( ε = 1.2 ) .

-a- -b-

-c- -d-

-e-

Figure 11. Finite element simulation of the shear strain distribution in a layered AA5182 sample deformed up to
an equivalent unit strain ε = 1 . The Voce law (6) and a friction coefficient m = 0.02 were used with two values
of mlayer , i.e. mlayer = 0.001 (a,b) and mlayer = 0.1 (c,d). (e) corresponds to the mono-block model. (a,c,e)
correspond to surface layers and (b,d) correspond to central layers.

0.3 0.35
0.30 )
Pa )
M
0.25 01 Pa
0.2 = 1 95 M
AA6016a α
0.20 a( =
AA6016b 82 ( α
n n 51 2b
0.15 AA 518
AA
0.1
0.10
MPa )
(α = 15 AA5182a +b
1 4 MPa ) 0.05 α=0
(α =
0.0 α=0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
equivalent strain equivalent strain

-a- -b-

Figure 12. Variation of the strain hardening parameter n with equivalent deformation for the 6016 (a) and the
5182 grades (b). The strain hardening parameter n is represented for the Voce law (6) and the parameters of
Table 2. The behaviour of a hypothetical strain hardening parameter for a Voce law without the αε p –term is
also represented.
Bacha A., Feuerstein M., Desrayaud Ch. and Klöcker H., 2007. Measuring Stress Strain Curves to Large 22
Strain on Sheet Metal, Journal of Evaluation and Testing, 35(2), 157-166

You might also like