View metadata, citation and similar papers at core.ac.

uk brought to you by CORE

provided by Elsevier - Publisher Connector

International Journal of Solids and Structures 46 (2009) 3825–3835

Contents lists available at ScienceDirect

International Journal of Solids and Structures

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

Stress–strain behaviour of aluminium alloys at a wide range of strain rates

Y. Chen *, A.H. Clausen, O.S. Hopperstad, M. Langseth
Structural Impact Laboratory (SIMLab), Centre for Research-based Innovation, Department of Structural Engineering, Norwegian University of Science and Technology,
NO-7491 Trondheim, Norway

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

Article history: The stress–strain behaviour of extruded AA6xxx and AA7xxx aluminium alloys in T6 temper was studied
Received 24 July 2008 at a wide range of strain rates. Tensile tests at low to medium strain rates were performed in a standard
Received in revised form 1 July 2009 tensile test machine, while a split-Hopkinson tension bar was used to carry out tests at high rates of
Available online 25 July 2009
strain. Extruded aluminium alloys have anisotropic mechanical properties, and tests were therefore done
in three directions with respect to the extrusion direction. It is found that the AA6xxx alloys exhibit no
Keywords: significant rate sensitivity in the stress–strain behaviour, while moderate rate sensitivity was found for
Split-Hopkinson bar
the AA7xxx alloys. There seems to be no significant difference between the rate sensitivity in the three
Anisotropy
Rate sensitivity
tensile directions. The experimental data were used to identify the parameters of a thermo-viscoplastic
Constitutive modelling constitutive relation for the extruded alloys, which includes the effects of strain hardening, strain-rate
Fracture hardening, thermal softening and plastic anisotropy.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction temperature, they found a very low, yet slightly positive, increase
in flow stress with strain rate. Similar observations regarding rate
Aluminium alloys are attractive for use in automotive parts ow- sensitivity of aluminium alloys, now in tension, were reported by
ing to their low weight. Design and production of light and crash- Reyes et al. (2006), investigating the alloys AA7003-T79 and
worthy structural parts in aluminium entail development of alloys AA7108-T6. On the other hand, the flow stress and fracture strain
and manufacturing processes, structural design and crashworthi- of AA6005-T6 were shown by Børvik et al. (2005) to have rather
ness analysis. In particular, the increased use of numerical predic- strong positive strain-rate sensitivity. Considering AA7075 in ten-
tions in the design phase calls for material models with a sufficient sion as well as compression, El-Magd and Abouridouane (2006)
accuracy and a careful identification of the coefficients involved in found that the flow stress increases slightly with strain rate, whilst
the model. the fracture deformation at increasing rates was reported to in-
The initial phase of a car crash, involving crushing of the energy crease in compression and decrease in tension. Negative strain-rate
absorbing parts of the car body, may be over in fractions of a sec- sensitivity of flow stress, caused by dynamic strain aging, is found
ond. It follows that the rate-of-deformation of these structural for some alloys in the AA5xxx series, see e.g. Naka and Yoshida
components is high. Design of such components based on material (1999), Abbadi et al. (2002) and Clausen et al. (2004). Also measur-
properties obtained under quasi-static loading conditions may lead ing the fracture strains in all tension tests at different strain rates,
to solutions that are not optimal with respect to dynamic load Clausen et al. (2004) found that the fracture strain is rate indepen-
transfer and energy absorption. In addition, the quest for lighter dent in the dynamic strain aging regime, whilst it increases consid-
components leads to stronger utilization of the material and thus erably for higher rates. Smerd et al. (2005) considered the ductility
to increased probability of fracture. It follows that there is a need of AA5182 and AA5754 at different strain rates in tension, and re-
for experimental data concerning the dynamic behaviour (i.e. ported a small increase in the elongation at fracture.
strength and ductility) of aluminium alloys and further for mathe- Tests at a broad range of strain rates are indeed necessary to ac-
matical models of their behaviour that can be used in numerical quire relevant data for an assessment of crashworthiness behav-
simulations of crashworthiness. iour. Keeping in mind that servo-hydraulic test machines rarely
Tests on aluminium alloys at different strain-rate levels are re- can impose a strain rate higher than approximately 1 s1 , other
ported by a number of investigators. Oosterkamp et al. (1999) did techniques must be applied for obtaining the strain rates occurring
compression tests on AA6082 and AA7108 in tempers T6 and T79 in crash situations. A quite recent overview of test methods for ele-
at strain rates ranging from 0.1 to approximately 2000 s1 . At room vated strain rates, including drop weights, split-Hopkinson bars
and different impact tests, was provided by Field et al. (2004).
* Corresponding author. Tel.: +47 99 73 13 41; fax: +47 37 02 41 28. The split-Hopkinson bar, which may be used in compression, ten-
E-mail address: [email protected] (Y. Chen). sion and torsion testing, seems to be the most widespread method

0020-7683/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijsolstr.2009.07.013
3826 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835

for material characterization at the crash relevant strain rates be- fected by specimen geometry, the same type of specimen was used
cause a rather well-defined stress and strain state is feasible (e.g. in both testing machines. The comparatively small geometrical
Harding et al., 1960; Lindholm and Yeakley, 1968; Albertini and dimensions were dictated by the SHTB which has a diameter of
Montagnani, 1976; Nicholas, 1981; Staab and Gilat, 1991; Thakur 10 mm. The initial diameter of each specimen was measured at
et al., 1996; Verleysen and Degrieck, 2004). three different sections in the gauge area of the specimen prior
The aim of this paper is to establish data on the stress–strain to testing.
behaviour of several aluminium alloys with applications in auto- For the tests at low to moderate strain rates duplicate tests
motive crash components like crash boxes and bumper beams. were carried out for each alloy, direction and strain rate between
To this end, tensile tests are performed at a wide range of strain 103 and 1 s1 , while only one test was performed for strain rate
rates using two different testing techniques – a servo-hydraulic 3 s1 . This was the highest strain rate which could be obtained
testing machine at low and medium strain rates and a split-Hop- without serious noise in the acquired data. The threaded specimen
kinson tension bar (SHTB) at high strain rates. Since the alloys were was screwed into purpose-made fixtures and clamped by the
extruded and exhibit marked anisotropy in strength and plastic hydraulic grips of the machine. A special arrangement was used
flow, tests were performed in three directions with respect to the to avoid any bending moments in the specimen. The tests were
extrusion direction. In addition to the stress–strain response, the carried out in displacement control mode. The strain was mea-
fracture strain was measured on the ruptured specimens. The data sured by a one-sided extensometer with 3.8 mm gauge length.
obtained from the tests were used to identify the parameters of a Force, cross-head displacement and strain were logged with a fre-
thermo-viscoplastic constitutive relation (including strain harden- quency depending on the strain rate. By adjusting the logging fre-
ing, strain-rate hardening and thermal softening) and an aniso- quency, approximately same amount of test data was sampled at
tropic yield criterion for the extruded aluminium alloys. all strain rates.
In the SHTB, 6–8 tests at different strain rates between 100 and
1000 s1 were performed for each alloy and direction. As illus-
2. Experimental
trated in Fig. 3, the SHTB consists of two steel bars with a circular
cross section and diameter 10 mm: an incident bar AC with length
2.1. Material
8140 mm and a transmission bar DE with length 7100 mm. The
test specimen is connected by the threaded ends to the bars be-
The materials tested were the aluminium alloys AA6060,
tween C and D. A locking mechanism is located at B on the incident
AA6082, AA7003 and AA7108 in T6 temper. All four alloys were
bar, and there is a hydraulic jack at A used to pre-tension part AB.
provided by Hydro Aluminium, Sunndalsøra, Norway as extruded
Strain gauges are glued to the bar at the locations 1, 2 and 3. All
rectangular profiles with 10 mm thickness and 83 mm width. The
measurements during the test are based on signals from the strain
nominal chemical compositions of the materials are shown in Ta-
gauges at 2 and 3, while gauge 1 is used for monitoring the level of
ble 1. The materials were solution heat-treated and artificially aged
the pre-tension force N 0 . The bars have a small diameter, and the
to obtain temper T6 (the peak hardness condition) before ma-
incoming stress wave, having a length of twice the distance be-
chined to test specimens. Alloy AA6060 is recrystallized, while
tween A and B, is quite long. According to Kolsky (1963), this
the other alloys have fibrous, non-recrystallized grain structure.
low diameter to wave-length ratio serves to minimize the disper-
Extruded alloys invariably exhibit plastic anisotropy that differs
sion of the stress wave in the set-up shown in Fig. 3. Also, it was
markedly between recrystallized and fibrous grain structures (Li
checked experimentally with an extra strain gauge glued between
et al., 2005).
lock B and gauge 2 that any dispersion indeed is negligible. There-
fore, the strains at point C and D are assumed to be as measured
2.2. Tension tests with respectively gauges 2 and 3. The nominal stress s, strain e
and strain rate e_ in the sample are then obtained from one-dimen-
Uniaxial tensile tests at strain rates between 103 and 103 s1 sional wave theory (Kolsky, 1963). The sampling frequency was
were performed at room temperature to determine the mechanical 1 MHz for all tests.
behaviour of the extruded aluminium profiles. To investigate the The elastic stiffness determined from the SHTB measurements
anisotropy of the materials, tensile tests were performed in three is in general too low. According to Albertini and Montagnani
directions, 0°, 45° and 90°, with respect to the extrusion direction (1977), the source of this error is the deformation occurring out-
(ED) of the plate, see Fig. 1. The direction of the test specimens is side the gauge part of the specimen. They suggested to re-calculate
indicated with an angle a relative to the ED, i.e. a ¼ 0 means that the nominal strain as e ¼ e0  sðE  E0 Þ=ðEE0 Þ, where e0 and E0 are
the longitudinal direction of the specimen is parallel with the ED. measured values of nominal strain and elastic stiffness, respec-
The tests were carried out in two different testing devices tively. A standard value of E ¼ 70; 000 MPa was applied for Young’s
depending on the strain rate. Uniaxial tests at strain rate between modulus. Assuming plastic incompressibility, true stress r and
103 and 3 s1 were done in a servo-hydraulic test machine, while true strain e were determined from the nominal stress and strain
dynamic tensile tests involving strain rates in the range 100– with the well-known relations e ¼ lnð1 þ eÞ and r ¼ sð1 þ eÞ, while
1000 s1 were performed in a SHTB. For each alloy and direction, the plastic strain was calculated as ep ¼ e  r=E.
15–17 tests were performed at various strain rates, and the test Representative hardening curves in terms of true stress r and
programme comprised a total of 198 tests. Specification of the test true plastic strain ep in three directions at strain rate 103 s1 are
specimen’s geometry is provided in Fig. 2. To achieve results unaf- shown in Fig. 4. All curves are drawn to the point of necking. To
get rid of noise in the acquired data, the engineering stress–strain
curves were smoothed using a moving average with 40 points. It
Table 1
was checked that this smoothing process did not remove impor-
Nominal chemical compositions of the alloys in wt.%.
tant physical information from the test data. The hardening curves
Alloy Fe Si Mg Mn Cu Zn Zr Al in Fig. 4 were then determined based on the smoothed engineering
AA6060 0.19 0.42 0.47 0.015 – – – Bal. stress–strain curves. It is seen that the fibrous alloys AA6082-T6,
AA6082 0.20 1.04 0.67 0.54 – – – Bal. AA7003-T6 and AA7108-T6 exhibit marked plastic anisotropy.
AA7003 0.16 0.085 0.62 – 0.038 5.43 0.16 Bal.
The trend is that the strength is lower in the 45° direction than
AA7108 0.13 0.099 0.87 – 0.031 5.77 0.16 Bal.
in the 0° and 90° directions. The strength is highest in the ED for
 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835 3827

Extrusion Direction (ED)

0°
90° 45°

Fig. 1. Orientation of test specimens with respect to the extrusion direction of the flat profile.

M5 strain rates illustrate the scatter. As one can observe from these fig-
Ø3 Ø3 ures, alloys AA6060 and AA6082 are rather insensitive to the strain
M5 rate, while alloys AA7003 and AA7108 exhibit small, positive
9 R =14 strain-rate sensitivity with respect to flow stress. There is no signif-
15 5 15 icant difference in the rate sensitivity between the three tensile
directions. From Fig. 6 it is further seen that there is some variation
Fig. 2. Geometry of test specimen for tensile tests. Measures in millimetre. in the strain to necking with strain rate for some of the alloys. This
indicates that the hardening to some degree depends on the strain
rate.
all the fibrous alloys. It is further seen that the plastic anisotropy is
stronger for the AA7xxx alloys than for the AA6xxx alloys. For the
recrystallized AA6060-T6, the tests at 103 s1 shown in Fig. 4 indi- 2.3. Plastic anisotropy
cate some strength anisotropy, where the 90° direction is some-
what stronger than the other two directions. Significant scatter To characterize the plastic anisotropy of the alloys, the flow-
between the duplicate tests was observed for this particular case, stress ratio ra is defined as the ratio between the flow stress raf
and the results are thus somewhat uncertain. However, based on in a tensile test in direction a and the flow stress r0f in a reference
the tests with higher strain rate, it seems that the strength anisot- test in the ED for the same amount of plastic work

ropy of AA6060-T6 is small, which is consistent with previous re- raf 
sults for AA6060 in temper T1 (Reyes et al., 2006). For all tests at ra ¼  ð2Þ
low to medium strain rate, the average difference between the
r0f W p
0.2% proof strengths obtained from duplicate tests was 3.2 MPa
where W p is the specific plastic work. The flow-stress ratio for the
for AA6060, 4.4 MPa for AA6082, 7.6 MPa for AA7003 and reference test in the 0° direction is equal to unity by definition.
3.7 MPa for AA7108.
For a given value of plastic strain epa , the specific plastic work W p
In the SHTB tests, the measured stress is only valid for strains in a tensile test in the direction a is defined as
beyond some few percent. The reason for this is that the tensile
Z epa
specimen is not in equilibrium in the first phase of the dynamic
test. To obtain comparable hardening curves without the small Wp ¼ raf depa ð3Þ
0
oscillations typically observed in results from SHTB tests, the Voce
hardening rule was used to represent each individual true stress- For each alloy, flow-stress ratios as function of plastic work were
plastic strain curve, i.e. calculated up to necking for the 45° and 90° directions. The ratio
changes rapidly in the low-strain range of the stress–strain curves,
  
raf ¼ r0 þ Q R 1  exp C R epa ð1Þ and thereafter it converges to a nearly constant level. A representa-
tive value of the flow-stress ratio in the range in which it is approx-
where raf is the flow stress in a tensile test in direction a; epa is the imately constant was computed as specified in Table 2. The values
plastic strain in the same direction, and the strain-hardening in the table, which were applied in the subsequent calibration of
parameters r0 , Q R and C R were found by curve fitting. An example the anisotropic yield criterion, are based on tests at strain rate
of the fitting of the Voce hardening rule in a dynamic test is shown 103 s1 .
in Fig. 5.
Representative hardening curves for the four alloys in the ED 2.4. Fracture strains
are plotted in Fig. 6 for strain rates between 103 and 103 s1 .
The curves are all plotted up to incipient necking. The true stress Assuming incompressible plastic deformation and neglecting
at 2% and 5% plastic strains vs. true plastic strain rate are plotted elastic strains, the fracture strain was obtained from measure-
for the four alloys in Figs. 7–9, depicting respectively the extrusion, ments of the original area A0 and fractured area Af of the gauge sec-
45° and 90° direction. Data from all tests are presented in these fig- tion by ef ¼ lnðA0 =Af Þ (e.g. Dieter, 1988). The fractured surfaces
ures, thus, the results from duplicate tests at low to moderate were assumed to have oval shape. The diameters along the long

600 600

N0 10

A B C D E
6080 2060 7100

Fig. 3. Schematic overview of split-Hopkinson tension bar. Measures in millimetre.

3828 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835

500 500
AA6060-T6 AA6082-T6
. .
ε=10-3s-1 ε=10-3s-1
400 400
True stress (MPa)

True stress (MPa)

300 300

200 200

0° 0°
100 45° 100 45°
90° 90°

0 0
0 0.04 0.08 0.12 0 0.04 0.08 0.12
Plastic strain Plastic strain

500 500
AA7003-T6 AA7108-T6
. .
ε=10-3s-1 ε=10-3s-1
400 400
True stress (MPa)

True stress (MPa)

300 300

200 200

0° 0°
100 45° 100 45°
90° 90°

0 0
0 0.04 0.08 0.12 0 0.04 0.08 0.12
Plastic strain Plastic strain

Fig. 4. Representative true stress versus true plastic strain curves showing the plastic anisotropy of the various alloys.

300 axis and the short axis of the ellipse were measured with a micro-
scope. Four measurements were made, two for each fractured sur-
AA6060-T6
. face. The fractured area of a specimen was calculated by
0o, ε ≈300s-1
Af ¼ pD1 D2 =4, where D1 and D2 are respectively the average of
the measured diameters along the long axis and the short axis of
the ellipse for the two fractured surfaces.
True stress (MPa)

200
The results indicated large scatter in the fracture strain at a gi-
ven strain rate, and for some alloys and directions a significantly
higher fracture strain was obtained for the high-rate tests than
for tests at lower rates. The specimens were carefully inspected
and it was found that the specimens tested in the servo-hydraulic
100 test machine fell into two categories. In the first, fracture was af-
fected by the small but sharp notch caused by the knives of the
extensometer. These specimens most often failed by shear fracture
experiment
after slight necking or without any sign of necking at all (Fig. 10a).
fitted
In the second, fracture was not influenced by the extensometer,
and the fracture mode was characteristically of cup-cone type with
0
substantial necking (Fig. 10b). This notch sensitivity seemed to be
0 0.04 0.08 0.12 0.16 0.2
strongest in the ED, and least in the 45° orientation. This is illus-
Plastic strain trated for alloy AA7003-T6 in Fig. 11. It is seen from the figure that
if the specimens that were influenced by the extensometer are dis-
Fig. 5. An example of the fitting of the Voce rule to the measured true stress versus
true plastic strain in the dynamic tests. regarded, the influence of strain rate upon the fracture strain
 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835 3829

300 400

AA6060-T6 AA6082-T6
0o 0o
360
250
True stress (MPa)

True stress (MPa)

320

200

. .
ε≈900s-1 280 ε≈900s-1
. .
ε≈300s -1 ε≈300s-1
. .
150 ε=1s -1 ε=1s-1
. .
ε=10 s -1 -1 240 ε=10-1s-1
. .
ε=10 s -2 -1 ε=10-2s-1
. .
ε=10-3s-1 ε=10-3s-1
100 200
0 0.04 0.08 0.12 0.16 0 0.02 0.04 0.06 0.08
Plastic strain Plastic strain

500 600

AA7003-T6 AA7108-T6
0o 0o

500
True stress (MPa)

True stress (MPa)

400

400

. .
ε≈900s-1 ε≈900s-1
300 . .
ε≈300s-1 ε≈300s-1
. .
ε=1s-1 300 ε=1s-1
. .
ε=10-1s-1 ε=10-1s-1
. .
ε=10-2s-1 ε=10-2s-1
. .
ε=10 s
-3 -1
ε=10-3s-1
200 200
0 0.04 0.08 0.12 0 0.04 0.08 0.12
Plastic strain Plastic strain

Fig. 6. Representative true stress versus true plastic strain curves showing the strain-rate sensitivity of the various alloys.

seems to be small. This was generally the case. However, the small 1.054 and 0.799, respectively, which suggest a slightly higher duc-
rate sensitivity could not be established for all combinations of al- tility at this rate than under the dynamic SHTB conditions, see
loy and direction, because in some cases the fracture mode of all Table 3. The standard deviation in these two test series was respec-
specimens at low to moderate strain rates was influenced by the tively 0.076 and 0.023, which is somewhat less than the values
extensometer. reported in Table 3.
The average fracture strain and its standard deviation from the
SHTB tests in the strain rate range 100–1000 s1 , which were all 3. Constitutive modelling
performed without extensometer, are compiled in Table 3. It is
seen that the fracture strain is strongly direction dependent. For In this section, the acquired data are used to determine the
the recrystallized AA6060 alloy the fracture strain is a great deal parameters of a thermo-viscoplastic constitutive model that ac-
lower in the 45° direction than in the two other directions. The counts for strain hardening, rate effects and thermal softening
trend is opposite for the non-recrystallized alloys, and it is also ob- due to adiabatic heating. The model adopts the anisotropic yield
served that the fracture strain in the 45° direction is about the criterion Yld91 proposed by Barlat et al. (1991) to describe the
same for all four alloys. However, it is clear that AA6060, which plastic anisotropy of the aluminium alloys. We have chosen a rel-
is the softest alloy, also has the highest overall ductility in tension. atively simple anisotropic yield criterion since the amount of test
It is finally noted that the measured fracture strains are rather con- data related to plastic anisotropy is rather limited.
sistent as illustrated by the standard deviations.
The possible scatter was evaluated by performing two test ser- 3.1. Thermo-viscoplastic constitutive relation
ies, each involving 10 samples, on AA6060-T6 in the 45° direction
and AA7108-T6 in the 90° direction. These specimens were not The constitutive relation, which has several similarities with the
equipped with any extensometer, and the tests were carried out constitutive relation proposed by Johnson and Cook (1983), reads
at strain rate 0.1 s1 . The average fracture strain was found to be
3830 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835

300 500

400
Flow stress (MPa)

Flow stress (MPa)

200
300

200
100

AA6060-T6 AA6082-T6
0° 100 0°
5% true strain 5% true strain
2% true strain 2% true strain

0 0
0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000
Plastic strain rate (s-1) Plastic strain rate (s-1)

500 500

400 400
Flow stress (MPa)

Flow stress (MPa)

300 300

200 200

AA7003-T6 AA7108-T6
100 0° 100 0°
5% true strain 5% true strain
2% true strain 2% true strain

0 0
0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000
Plastic strain rate (s-1) Plastic strain rate (s-1)

Fig. 7. Flow stress versus plastic strain rate for the various alloys in extrusion direction (ED).

r ¼ ½r0 þ Q R ð1  expðC R eÞÞ½1 þ e_  Cv ½1  T M  ð4Þ 3.2. Anisotropic yield criterion

where r  is the effective stress, r0 is the yield stress, Q R and C R are The effective stress is defined by the anisotropic yield criterion
hardening constants, e is the effective plastic strain, e_  ¼ e_ =e_ 0 is a Yld91 of Barlat et al. (1991), viz.
non-dimensional strain rate, C v and e_ 0 are parameters governing
the rate sensitivity of the material, M defines thermal softening jS1  S2 jm þ jS2  S3 jm þ jS3  S1 jm ¼ 2r
m ð7Þ
and T  is the homologous temperature where S1 ; S2 and S3 are the principal values of the stress tensor S,
which in matrix representation reads
T  T0
T ¼ ð5Þ 2 cCbB 3
Tm  T0 3
hH gG
6 aAcC
fF 7
S ¼ 4 hH 3 5 ð8Þ
Here, T 0 and T m are the room temperature and the melting temper-
bBaA
ature, respectively. Since we are considering loading at high strain gG fF 3
rates, it is reasonable to assume adiabatic conditions. In this case,
In Eq. (8), the variables A, B, C, F, G and H are defined by the compo-
the rate of temperature change is computed as
nents of the Cauchy stress tensor r as
r e_ A ¼ ry  rz ; B ¼ rz  rx ; C ¼ rx  ry ;
T_ ¼ v ð6Þ
qC p F ¼ ryz ; G ¼ rzx ; H ¼ rxy ð9Þ

where q is the density, C p is the specific heat capacity at constant where the x-axis is aligned with the extrusion direction (ED), the
pressure, and v is the Taylor–Quinney empirical constant which y-axis is aligned with the in-plane transverse direction, and the
specifies the fraction of plastic work converted to heat. A typical z-axis is aligned with the thickness direction of the profile.
value of v for metals is 0.90. The six parameters a, b, c, f, g and h define the plastic anisotropy
 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835 3831

500 500

400 400
Flow stress (MPa)

Flow stress (MPa)

300 300

200 200

AA6060-T6 AA6082-T6
100 45° 100 45°
5% true plastic strain 5% true plastic strain
2% true plastic strain 2% true plastic strain

0 0
0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000
Plastic strain rate (s-1) Plastic strain rate (s-1)

500 500

400 400
Flow stress (MPa)

Flow stress (MPa)

300 300

200 200

AA7003-T6 AA7108-T6
100 45° 100 45°
5% true plastic strain 5% true plastic strain
2% true plastic strain 2% true plastic strain

0 0
0.001 0.01 0.1 1 10 100 1000 10000 0.001 0.01 0.1 1 10 100 1000 10000
Plastic strain rate (s-1) Plastic strain rate (s-1)

Fig. 8. Flow stress versus plastic strain rate for the various alloys in the 45° direction.

of the material, and are determined from mechanical tests as ex- equal to unity. The resulting material parameters are summarized
plained in the next subsection. in Table 4.
The parameters of the anisotropic yield criterion were deter-
mined based on the average flow-stress ratios r a and the R-values,
which are defined by
3.3. Parameter identification
e_ pw
The constitutive relation, see Eq. (4), consists of three factors Ra ¼ ð10Þ
e_ pt
governing strain hardening, strain-rate sensitivity and temperature
softening. The parameters for each phenomenon were calibrated where e_ pw and e_ pt are the plastic strain rates in the width and thick-
separately. First, the hardening parameters r0 ; Q R and C R were ness direction of the tensile specimen. The R-values were not
determined from the reference hardening curve in the ED at strain determined in the present experimental investigation. Instead, the
rate 103 s1 . Although addressing a dynamic test, Fig. 4 clearly R-values for AA6060, AA6082, AA7003 and AA7108 in T6 temper
shows that the Voce law applied in the first factor of Eq. (4) is able were respectively taken from Jensen (2005), Wang (2006), Achani
to represent also the stress–strain curves at strain rate 103 s1 in (2006) and Kokkula (2005), and are summarized in Table 5. Since
Fig. 5. With these parameters given, the parameters C v and e_ 0 gov- the R-values of an alloy may vary from one extruded profile to an-
erning the strain-rate sensitivity of the material were determined other, the accuracy of these data for the actual rectangular profile is
from flow-stress values at 2% plastic strain for strain rates between somewhat uncertain.
103 and 103 s1 . Again the data for the ED were used. An example The yield condition was then calibrated by minimizing the
of the resulting strain-rate part of the model, resulting from this residual
identification process, is shown in Fig. 12. Owing to lack of tests
 2  2
at elevated temperature, it was assumed that the flow stress de- R ¼ ðRexp  Rmod Þ2 þ Rexp mod
þ Rexp mod
0 0 45  R45 90  R90 ð11Þ
creases linearly with increasing temperature and thus M was set
3832 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835

500 500

400 400
Flow stress (MPa)

Flow stress (MPa)

300 300

200 200

AA6060-T6 AA6082-T6
100 90° 100 90°
5% true plastic strain 5% true plastic strain
2% true platic strain 2% true plastic strain

0 0
0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000
Plastic strain rate (s-1) Plastic strain rate (s-1)

500 500

400 400
Flow stress (MPa)

Flow stress (MPa)

300 300

200 200

AA7003-T6 AA7108-T6
100
90° 100 90°
5% true plastic strain 5% true plastic strain
2% true plastic strain 2% true plastic strain

0 0
0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000
Plastic strain rate (s-1) Plastic strain rate (s-1)

Fig. 9. Flow stress versus plastic strain rate for the various alloys in the 90° direction.

Table 2 subject to the constraints: r moda ¼ r exp

a for a ¼ 0 ; 45 ; 90 . The
Flow-stress ratios for the four alloys at strain rate 103 s1 . superscripts ‘‘exp” and ‘‘mod” refer to experimental and model data.
Alloy r0 r 45 r 90
The flow-stress ratios r mod
a and R-values Rmoda can be obtained from
the constitutive model, using the yield function and the associated
AA6060-T6 1.0 0.988 1.079
AA6082-T6 1.0 0.919 0.975
flow rule. The exponent m in the yield function is usually given the
AA7003-T6 1.0 0.841 0.923 value 8 for FCC materials such as aluminium (see e.g. Barlat and
AA7108-T6 1.0 0.870 0.954 Lian, 1989). The parameters of the yield criterion are compiled in

Fig. 10. Fracture modes: (a) shear fracture and (b) cup-cone fracture.
 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835 3833

1.2 1.2
AA7003-T6-0o AA7003-T6-45o
2
2 2
2 2 2 2 22
2 2 2 2
2 2

0.8 0.8
Fracture strain

Fracture strain
2
3
33 3
3 3
33
3
4
0.4 0.4
4
4
4
4
4
1 necking 1 necking
4 2 necking+slight shear 2 necking+slight shear
4
3 slight necking+shear 3 slight necking+shear
4 shear 4 shear
0 0
0.001 0.01 0.1 1 10 100 1000 10000 0.001 0.01 0.1 1 10 100 1000 10000
Plastic strain rate (s-1) Plastic strain rate (s-1)

Fig. 11. Fracture strain versus strain rate for alloy AA7003-T6 in the 0° and 45°. The symbols ‘‘N” and ‘‘M” indicate fracture influenced by extensometer and fracture not
influenced by extensometer, respectively.

Table 3 Table 4
Average logarithmic strain at fracture from SHTB tests with strain rates between 100 Material parameters for the four alloys.
and 1000 s1 for each alloy and tensile direction. The standard deviation is given in
parenthesis. Alloy r0 (MPa) Q R (MPa) CR e_ 0 ðs1 Þ Cv M

AA6060-T6 196.1 51.2 24.71 0.001 0.0038 1.0

Alloy Logarithmic strain at fracture, ef
AA6082-T6 310.2 62.7 24.31 0.001 0.0015 1.0
0° 45° 90° AA7003-T6 344.2 115.7 15.19 0.095 0.0116 1.0
AA7108-T6 354.4 101.1 19.91 0.001 0.0093 1.0
AA6060-T6 1.530 (0.106) 0.856 (0.095) 1.422 (0.105)
AA6082-T6 0.707 (0.148) 0.901 (0.057) 0.638 (0.127)
AA7003-T6 0.592 (0.045) 0.928 (0.099) 0.785 (0.054)
AA7108-T6 0.579 (0.055) 0.949 (0.077) 0.606 (0.057)
Table 5
Experimental and calculated (in parenthesis) R-values for the four alloys. The
experimental values are taken from the literature and are obtained for another set of
extruded profiles than those involved in the current study.
500
Alloy R-values
R0 R45 R90
AA6060-T6 0.48 (0.51) 0.25 (0.90) 1.38 (0.78)
400
AA6082-T6 0.37 (0.70) 1.19 (1.09) 0.87 (0.61)
AA7003-T6 0.25 (0.94) 2.11 (1.69) 0.96 (0.58)
True stress σ2 (MPa)

AA7108-T6 0.27 (0.43) 0.67 (1.00) 1.16 (0.34)

300

Table 6
Parameters for YLD91 for the four alloys.
200
Alloy m a b c f g h
AA6060-T6 8 0.959 1.098 0.894 1.0 1.0 1.006
AA7108-T6 AA6082-T6 8 1.098 1.049 0.949 1.0 1.0 1.093
100
2% true strain AA7003-T6 8 1.170 1.009 0.991 1.0 1.0 1.218
AA7108-T6 8 1.194 1.106 0.885 1.0 1.0 1.149
Tests
Model
0
0.001 0.01 0.1 1 10 100 1000
surfaces differs significantly from the von Mises ellipsoid. The accu-
Plastic strain rate (s-1) racy of the yield condition can be evaluated by comparing the cal-
culated and experimental R-values. This is done in Table 5 in
Fig. 12. True stress at 2% plastic strain vs. plastic strain-rate from tests and material
model for AA7108-T6. which the calculated R-values are given in parenthesis. Clearly,
the yield function is not able to represent the variation of the R-va-
lue with the orientation of the tensile test. However, with the avail-
Table 6. The yield surfaces for AA7003-T6 and AA7108-T6 are de- able experimental data it was not found feasible to use an
picted in Fig. 13 for a state of plane stress ðrz ¼ ryz ¼ rzx ¼ 0Þ. anisotropic yield criterion with more parameters. One possible can-
The contours in the figure represent levels of constant normalized didate would be Yld2003-18p (Barlat et al., 2005), but this yield cri-
shear stress with a contour distance of 0:03r0 . It is evident that ow- terion has 18 parameters that need to be determined from a more
ing to the high exponent and the anisotropy, the shape of the yield comprehensive set of experiments or crystal plasticity calculations.
3834 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835

1.5 1.5
AA7003-T6 AA7108-T6

1 1

0.5 0.5

σy σy 0
0
σ0 σ0
-0.5 -0.5

-1 -1

-1.5 -1.5
-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
σx/σ0 σx/σ0
Fig. 13. Yield surface for AA7003-T6 and AA7108-T6 in a state of plane stress ðrz ¼ ryz ¼ rzx ¼ 0Þ.

4. Discussion and conclusions in Eq. (6) is the area under the true stress vs. plastic strain curve,
which from Fig. 5 is seen to be of order 450 MPa  0:1. Applying
Tensile tests on four aluminium alloys relevant for application the standard values of q ¼ 2700 kg=m3 and C p ¼ 910 J=kg K for
in automotive crash components were carried out over a wide aluminium, the temperature rise is found to be about 15 °C during
range of strain rate, and the stress–strain behaviour and plastic the part of the test which is relevant for the parameter identifica-
anisotropy were established. Two tests techniques were used to tion. Earlier investigations suggest that the mechanical properties
obtain the data. In the low to medium strain-rate range a servo- are only slightly different at room temperature and 100 °C (Clausen
hydraulic test machine was applied, while a split-Hopkinson ten- et al., 2004; Børvik et al., 2005). Thus, adiabatic heating of the spec-
sion bar was used for high strain rates. imen seems to be a minor source of error.
The tests show that AA6060-T6 and AA6082-T6 exhibit only
slight sensitivity to the strain rate, and could probably be modelled
Acknowledgement
as rate-insensitive with good accuracy. AA7003-T6 and AA7108-T6
show a marked sensitivity to strain rate, which should be included
The present work was supported by the Research Council of
in simulations. This is in agreement with earlier observations by
Norway through the strategic university programme ‘‘Design of
Oosterkamp et al. (1999) and Reyes et al. (2006) on similar alloys.
Crashworthy Light Structures”.
The fibrous, non-recrystallized alloys AA6082-T6, AA7003-T6
and AA7108-T6 have a marked strength anisotropy with the lowest
strength in the 45° direction with respect to the extrusion direc- References
tion. The strength in the 90° direction is somewhat lower than in
the extrusion direction. The strength anisotropy for the recrystal- Abbadi, M., Hähner, P., Zeghloul, A., 2002. On the characteristics of Portevin–Le
Chatelier bands in aluminum alloy 5182 under stress-controlled and strain-
lized alloy AA6060-T6 is small. The dynamic fracture strain varied controlled tensile testing. Mechanical Science and Engineering A 337, 194–201.
with alloy and direction. For the AA6060-T6 alloy the fracture Achani, D., 2006. Constitutive models of elastoplasticity and fracture for aluminium
strain was low in the 45° direction and high in the 0° and 90° direc- alloys under strain path change. Ph.D. Thesis, Department of Structural
Engineering, Norwegian University of Science and Technology.
tions. For the other alloys, the fracture strain was significantly
Albertini, C., Montagnani, M., 1976. Wave propagation effects in dynamic loading.
higher in the 45° direction. It is believed that this difference is Nuclear Engineering and Design 37, 115–124.
linked to the different microstructure (texture and grain structure) Albertini, C., Montagnani, M., 1977. Dynamic material properties of several steels
of the alloys, and is a topic for further investigation in future work. for fast breeder reactor safety analysis. Report EUR 5787 EN, Applied Mechanics
Division, Joint Research Centre, Ispra, Italy.
A possible source of error is the application of two different test Barlat, F., Lian, J., 1989. Plastic behaviour and stretchability of sheet metals. Part I: A
devices. The measured data for the alloys presented in Figs. 6–9 yield function for orthotropic sheets under plane stress conditions.
indicate, however, that the stress–strain behaviour is consistent International Journal of Plasticity 13, 385–401.
Barlat, F., Lege, D.J., Brem, J.C., 1991. A six-component yield function for anisotropic
in tests performed with the servo-hydraulic machine and the SHTB. materials. International Journal of Plasticity 7, 693–712.
Also, earlier investigations reported by Clausen et al. (2004) and Barlat, F., Aretz, H., Yoon, J.W., Karabin, M.E., Brem, J.C., Dick, R.E., 2005. Linear
Børvik et al. (2005) reveal that the results from the SHTB fit well transformation based anisotropic yield functions. International Journal of
Plasticity 21, 1009–1039.
to the tendency observed from the tests in the servo-hydraulic Børvik, T., Clausen, A.H., Eriksson, M., Berstad, T., Hopperstad, O.S., Langseth, M.,
machine. 2005. Experimental and numerical study on the perforation of AA6005-T6
A question arising in connection with dynamic material testing panels. International Journal of Impact Engineering 32, 35–64.
Clausen, A.H., Børvik, T., Hopperstad, O.S., Benallal, A., 2004. Flow and fracture
is temperature rise. Most of the plastic work in the specimen is characteristics of aluminium alloy AA5083-H116 as function of strain rate,
converted to heat, and this heat does not have time to be trans- temperature and triaxiality. Mechanical Science and Engineering A 364, 260–
ferred away from the coupon during the duration of a SHTB test. 272.
Dieter, G.E., 1988. Mechanical Metallurgy. McGraw-Hill, New York.
A large temperature increase will of course soften the material
El-Magd, E., Abouridouane, M., 2006. Characterization, modelling and simulation of
and provide hardening behaviour not representative for the as- deformation and fracture behaviour of the light-weight wrought alloys under
sumed room temperature conditions. There was not performed high strain rate loading. International Journal of Impact Engineering 32, 741–
any temperature measurement during the SHTB tests. On the other 758.
Field, J.E., Walley, S.M., Proud, W.G., Goldrein, H.T., Siviour, C.R., 2004. Review of
hand, Eq. (6) can provide an estimate of the temperature increase experimental techniques for high rate deformation and shock studies.
during the hardening phase up to onset of necking. The nominator International Journal of Impact Engineering 30, 725–775.
 Y. Chen et al. / International Journal of Solids and Structures 46 (2009) 3825–3835 3835

Harding, J., Wood, E.O., Campbell, J.D., 1960. Tensile testing of materials at Nicholas, T., 1981. Tensile testing of materials at high rates of strain. Experimental
impact rates of strain. Journal of Mechanical Engineering and Science 2, Mechanics 21, 177–185.
88–96. Oosterkamp, L.D., Ivankovic, A., Venizelos, G., 1999. High strain rate properties of
Jensen, Ø., 2005. Behaviour of aluminium extrusion subjected to axial loading. selected aluminium alloys. Mechanical Science and Engineering A 278, 225–
Dr.ing Thesis, Department of Structural Engineering, Norwegian University of 235.
Science and Technology. Reyes, Aa., Hopperstad, O.S., Lademo, O.-G., Langseth, M., 2006. Modeling of
Johnson, G.R., Cook, W.H., 1983. A constitutive model and data for metals subjected textured aluminum alloys used in a bumper system: material tests and
to large strains, high strain rates and high temperatures. In: Proceedings of the characterization. Computational Materials Science 37, 246–268.
7th International Symposium on Ballistics, Hague, pp. 541–547. Smerd, R., Winkler, S., Salisbury, C., Worswick, M., Lloyd, D., Finn, M., 2005. High
Kokkula, S., 2005. Bumper beam-longitudinal system subjected to offset impact strain rate tensile testing of automotive aluminum alloy sheet. International
loading. Ph.D. Thesis, Department of Structural Engineering, Norwegian Journal of Impact Engineering 32, 541–560.
University of Science and Technology. Staab, G.H., Gilat, A., 1991. A direct-tension split Hopkinson bar for high strain-rate
Kolsky, H., 1963. Stress Waves in Solids. Dover, New York. testing. Experimental Mechanics 31, 232–235.
Li, S., Engler, O., Van Houtte, P., 2005. Plastic anisotropy and texture evolution Thakur, A., Nemat-Nasser, S., Vecchio, K.S., 1996. Dynamic Bauschinger effect. Acta
during tensile testing of extruded aluminium profiles. Model. Simulat. Mater. Materiala 44, 2797–2807.
Sci. Eng. 13, 783–795. Verleysen, P., Degrieck, J., 2004. Experimental investigation of the deformation of
Lindholm, U.S., Yeakley, L.M., 1968. High strain-rate testing: tension and Hopkinson bar specimens. International Journal of Impact Engineering 30, 239–
compression. Experimental Mechanics 8, 1–9. 253.
Naka, T., Yoshida, F., 1999. Deep drawability of type 5083 aluminium–magnesium Wang, T., 2006. Modelling of welded thin-walled aluminium structures. Ph.D.
alloy sheet under various conditions of temperature and forming speed. Journal Thesis, Department of Structural Engineering, Norwegian University of Science
of Materials Processing Technology (89/90), 19–23. and Technology.