See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227050677

Constitutive behavior of as-cast AA1050, AA3104, and AA5182

Article  in  Metallurgical and Materials Transactions A · July 2002

DOI: 10.1007/s11661-002-0030-8

CITATIONS READS

40 848

4 authors, including:

Laurens Katgerman
Delft University of Technology
286 PUBLICATIONS   5,804 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Aluminium extrusion View project

Effect of compositions on microstructure and hot tearing susceptibility in direct-chill casting of 7XXX aluminum alloys View project

All content following this page was uploaded by Laurens Katgerman on 27 May 2014.

The user has requested enhancement of the downloaded file.

Constitutive Behavior of As-Cast AA1050, AA3104, and
AA5182
W.M. VAN HAAFTEN, B. MAGNIN, W.H. KOOL, and L. KATGERMAN

Recent thermomechanical modeling to calculate the stress field in industrially direct-chill (DC) cast-
aluminum slabs has been successful, but lack of material data limits the accuracy of these calculations.
Therefore, the constitutive behavior of three aluminum alloys (AA1050, AA3104, and AA5182) was
determined in the as-cast condition using tensile tests at low strain rates and from room temperature
to solidus temperature. The parameters of two constitutive equations, the extended Ludwik equation
and a combination of the Sellars–Tegart equation with a hardening law, were determined. In order
to study the effect of recovery, the constitutive behavior after prestraining at higher temperatures was
also investigated. To evaluate the quantified constitutive equations, tensile tests were performed
simulating the deformation and cooling history experienced by the material during casting. It is
concluded that both constitutive equations perform well, but the combined hardening-Sellars–Tegart
(HST) equation has temperature-independent parameters, which makes it easier to implement in a
DC casting model. Further, the deformation history of the ingot should be taken into account for
accurate stress calculations.

I. INTRODUCTION where s is the stress, K is a material constant (i.e., the stress

at « 5 1 and « 5 1 s21), « is the plastic strain, nL is the
A. DC Casting Models strain-hardening coefficient, « is the strain rate, «0 is a con-
IN the direct-chill (DC) casting process of aluminum, stant taken equal to 1 s21, and mL is the strain-rate sensitivity.
the primary and secondary cooling impose strong thermal It fits stress-strain curves from room temperature to solidus
gradients on the ingot. Stresses and strains caused by both temperature at the strain rates applicable to DC casting. The
solidification shrinkage and thermal contraction may lead temperature dependency of the stress comes into the equation
to distortion of the ingot shape (e.g., butt curl, butt swell, through K, mL , and nL .
or rolling face pull-in) or to hot tearing and cold cracking. For the higher temperature regime of the casting process,
In order to understand these phenomena and optimize the where the material deforms mainly by creep, experimental
casting process accordingly. DC casting simulation models data[3,11– 13] have been fitted to the Garofalo or Sellars–
were developed.[1 –5] Critical input to these models is the Tegart equation:[14,15]

F 1 2G
material behavior under the prevalent conditions, i.e., low nH

1 RT 2
strains, low strain rates, and temperatures ranging from cast- sss 2Q
« 5 A sinh exp [2]
ing temperature to room temperature. Therefore, the thermo- s0
mechanical behavior is governed successively by power-law
creep, power-law breakdown and low temperature plasticity. where A, «0, and nH are material constants, sss is the steady-
Such behavior can be described by many different constitu- state stress, Q is the activation energy, R is the universal
tive equations, varying from equations that result from sim- gas constant, and T is the temperature (K). This equation
ple hardening models to complex equations with internal fits the data from creep tests in the temperature range 400
variables to describe the microstructural evolution of the 8C to 600 8C and the strain-rate range 1026 to 1022 s21 very
alloy. well. Below 400 8C, the Sellars–Tegart law overestimates
the stress, as at low temperatures more strain is required to
reach steady-state conditions.[12] The advantage of Eq. [2]
B. Constitutive Equations over Eq. [1] is that the parameters are temperature independ-
To describe the material behavior in the as-cast condition ent and that the temperature occurs explicitly in the equation.
in the complete temperature range encountered during cast- A convenient way to extend the validity of Eq. [2] to
ing, the extended Ludwik equation has been used,[1,6– 10] lower temperatures is to replace sss by sH and to combine
which is a phenomenological equation of the form it with a simple hardening law in the following way:
mL

12 sH 5 (s0 1 k! a (T )«) ? f (Z ) [3]

«
s 5 K« nL [1]
«0
where k is a constant, and a is a hardening parameter. The
function f (Z ) is defined as follows:
mH

1 11 2 22
W.M. VAN HAAFTEN, Researcher, is with Corus R,D&T, 1970 CA Z
IJmuiden, The Netherlands. B. MAGNIN, Technical Director Extrusion, f(Z ) 5 min 1, arcsinh [4]
A
is with the Pechiney Centre de Recherche de Voreppe, 38340, Voreppe,
France. W.H. KOOL, Senior Scientist, and L. KATGERMAN, Professor,
are with the Laboratory of Materials, Delft University of Technology, 2628
where Z is the Zener–Hollomon parameter (Z 5 « exp
AL Delft, The Netherlands. Contact e-mail: l.katgerman@tnw.tudeleft.nl (Q/RT )), A is the same material constant as in Eq. [2], and
Manuscript submitted July 5, 2001. mH equals 1/nH. The parameter a represents the percentage

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JULY 2002—1971

 Table I. Literature Data on Parameter Values in Sellars–Tegart Equation (Eq. [2])
s0 Q
Alloy Treatment* A (s21) (MPa) n (kJ/mol) Reference
AA1050 HG 3.9 3 1011 27 3.84 157 23
AA1201 AC 2.8 3 1026 13.6 7.13 400 3
AA3103 AC 1.33 3 1016 31.6 7.94 241 14
AA3004 HG 1.8 3 1012 29.1 3.6 194 23
AA5182 HG 5.8 3 1009 16 1.35 174 23
AA5182 HR 2.64 3 1011 50 5.78 196 13
*HG 5 homogenized, AC 5 as-cast, and HR 5 hot rolled.

of hardening occurring at a certain temperature compared of aluminum alloys under the conditions that prevail during
to the hardening occurring at room temperature under the DC casting. Since the microstructure should resemble the
same load. At high temperature, a 5 0, and at low tempera- microstructure during DC casting as closely as possible, two
ture, a 5 1. Between 250 8C and 400 8C, a changes from alternative procedures are adopted in solid-state testing. The
1 to 0 to provide a smooth transition between hardening- first is that the material is solidified in situ.[8,21] The second
dominated behavior to steady-state behavior. The tempera- is that the material in the as-cast condition is reheated to
ture dependency of a is as follows: the test temperature.[8,10,12,17,22] Nedreberg[10] has performed
tensile tests on AA6063 from room temperature to solidus
1
a5 [5] temperatures. The as-cast specimens were heat treated to
1 1 exp (a0 1 a1T ) redissolve the precipitates and then cooled as fast as possible
in which a0 and a1 are fitting parameters. We will refer to Eqs. to the test temperature. It was concluded that Eq. [1] fits
[3] through [5] as the combined hardening-Sellars–Tegart the data very well over the entire temperature range. Magnin
(HST) equation. et al.[8] have performed tensile tests with an Al-4.5 pct Cu
alloy at strain rates of 1022 to 1023 s21 and from room
temperature to coherency temperature. They compared
C. Determination of Constitutive Data results from specimens cooled from the liquid state, as-cast
In earlier studies, Eq. [1] was applied to a solution heat- specimens heated directly to the test temperature, and as-
treated and aged Al-3 pct Mg alloy.[7] The strain-rate sensitiv- cast specimens cooled to the test temperature after a short
ity, mL , varies from slightly negative (20.01) below 200 8C treatment at 500 8C for 10 seconds. The results from the
to 0.3 at 400 8C, while nL drops from 0.4 at room temperature first and the latter test were similar, and therefore, it was
to 0 at 400 8C. The parameters mL and nL were also deter- decided that the best procedure to imitate the as-cast condi-
mined for an as-cast Al-4.5 pct Cu alloy.[8] For this alloy, tion was cooling to the test temperature after this short heat
mL increases from 0 at 150 8C to 0.2 at 550 8C, and nL treatment. Several other alloys have been investigated in the
decreases from 0.26 at room temperature to 0 around 400 as-cast condition (AA3004,[17] AA3103,[12] AA1201,[3] and
8C. The strain-rate-sensitivity parameter, mL , depends lin- AA2024[22] ), but these studies are limited either to the creep
early on the Mg content at elevated temperature for homoge- regime[3,12,22] or to low temperatures,[17] and the experimental
nized Al-Mg alloys with maximum 6 pct Mg.[16] Park and procedures all differ from one study to the other.
Morris[17] investigated the tensile behavior of AA3004 in The aim of this study is to determine the constitutivebehav-
both as-cast condition and after a H19 treatment. The mate- ior of three non heat-treatable alloys (AA1050, AA3104, and
rial shows an increasing strain-rate sensitivity from 20 8C AA5182) under conditions that resemble those during DC
to 200 8C, but the increase is much stronger for the H19 casting as closely as possible. Tensile testing was performed
material. At the lowest temperatures, strain-rate sensitivity in the as-cast condition from solidus temperature to room
is slightly negative, and serrated flow is observed. These temperature at strain rates from 5 3 1023 to 1 3 1025 s21.
observations are attributed to the Mg atoms in the solid
solution. The steep increase of mL and the disappearance of
II. EXPERIMENTAL
the serrated flow above 150 8C is due to the precipitation
of Mg2Si and the increase in thermal energy so that pinning A. Material and Mechanical Testing
of dislocations is no longer effective.[17]
The Sellars–Tegart equation (Eq. [2]) has been used for The materials for this study were sampled from industrial
a wide range of conditions. Parameter values for both as- rolling slabs.1 The slabs were received in the as-cast condi-
cast material and homogenized and/or hot-rolled material tion and were not heat treated. Tensile tests were carried out
are summarized in Table I.[3,11,12,18] It is observed that the with a GLEEBLE* 1500 (AA3104) and a GLEEBLE 3500
values for Q are much higher than the value of 142 *GLEEBLE is a trademark of Dynamic Systems, Inc., Poestenkill, NY.
kJ/mol for vacancy diffusion.[19,20] Further, the as-cast mate-
rial yields high values for A and n, whereas the hot-rolled (AA1050 and AA5182) thermomechanical simulator. The
material has an intermediate n value.
Whereas many studies deal with the constitutive equa-
tions for aluminum alloys, only a limited amount of 1
AA3104 at Pechiney CRV, France; AA1050 and AA5182 at VAW,
data[3,8,10,12,17,21,22] is available on the constitutive behavior Germany.

1972—VOLUME 33A, JULY 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 1—Specimen shape.

Fig. 3—Imposed temperature (???) and stress (–––) path during the continu-
ous cooling test.

AA3104 slab. Specimens from different locations across the

slab were stretched at a strain rate of 2 3 1024 s21 at 50
Fig. 2—Schematic view of experimental setup. 8C. The second test investigated the effect of the duration
of a short heat treatment at 550 8C with AA1050 and AA3104
and at 480 8C with AA5182 to redissolve small precipitates
specimens were taken with the tensile direction parallel to and to annihilate dislocation networks developed during
the casting direction. The specimen (Figure 1) was secured cooling and shrinkage of the ingot. The heat treatment was
between two water-cooled jaws of which one was moving given in situ in the thermomechanical simulator. Tensile
(Figure 2). Heating of the specimens during the heat treat- behavior at 50 8C was determined without treatment and
ments and during the tests was done via Joule heating in after this treatment at 550 8C or 480 8C for 10 and 60
the GLEEBLE. Temperature was measured with a chromel- seconds, followed by standard cooling in the GLEEBLE
alumel thermocouple that was spot welded on the specimen. (,50 8C/s). Based on this test, all following tests (numbered
The copper wedges quickly remove the heat from the outer 3 through 6) started with a heat treatment at 550 8C (AA1050
parts of the specimen so that deformation remains restricted and AA3104) or 480 8C (AA5182) for 10 seconds, after
to the middle part. Most experiments were performed in which, the specimen was cooled (,50 8C/s) to the testing
stroke control. Only the tests at the highest temperatures temperature.
and the continuous-cooling tests were carried out in load In the third test, the effect of different cooling rates was
control. True plastic strain, «, was measured via a dilatome- investigated in AA3104 by comparing the tensile behavior
ter, which monitored the diameter of the specimen. Assum- at 50 8C after cooling from 550 8C at 5 8C/s or 0.5 8C/s.
ing constant volume during plastic deformation, the true These rates simulate the cooling histories near the edge and
plastic strain is given by « 5 22 ln (D/D0), with D the in the core of the slab, respectively.
diameter during the test and D0 the initial diameter. True
stress is defined by the force measured by a load cell, divided
by the cross-sectional area of the specimen during the test. C. Tests for the Determination of Constitutive Behavior
Stress and temperature fields in the specimen are assumed to
produce a uniaxial stress state in the middle of the specimen, Different aspects of the constitutive behavior of the alloys
considering modeling results of Mo et al.[23] Depending on were investigated with the following tests.
the experiment, different loading histories were applied. In In the fourth test, the effect of deformation history was
all tests deformation was kept below 3 pct strain. Strain studied in AA3104 and AA5182. The specimens were 1 to
rates were determined from the true plastic strain and, there- 2 pct deformed at 200 8C, 300 8C, 400 8C, or 500 8C,
fore, from the recorded diameter reduction of the specimen. subsequently cooled (,50 8C/s), and then further deformed
Since this reduction is influenced by the sample volume at 50 8C. These tests were compared with tests with the
deformed, there is not a one-to-one correspondence between same temperature-time history but without performing the
jaw velocity and strain rate. Strain rates varied from 5 3 first deformation.
1023 to 1 3 1025 s21, which are realistic values for DC In the fifth test, tensile tests between 50 8C and the solidus
casting. During the tests, time, temperature, specimen diame- temperature were carried out with strain-rate changes during
ter, and load were recorded. All tests were carried out two the test to determine the constitutive behavior of the alloys.
or three times. Since the maximum deviation in stress was After 1 to 2 pct strain, the strain rate was reduced to charac-
65 pct or less, average curves are shown. terize the strain-rate sensitivity of the material.
In the sixth test, AA5182 specimens were strained and
cooled at the same time (Figure 3), similarly to the strain
B. Initial Tests
and temperature path during DC casting. The temperature
Before the constitutive behavior of the alloys was investi- path was taken from in situ thermocouple measurements
gated, some initial tests were carried out. In the first test, during a full-scale casting experiment.[24] The applied load
the effect of specimen location was investigated for the was imposed such that the plastic strain continuously

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JULY 2002—1973

 Table II. Elemental Concentrations (Weight Percent) of
AA1050, AA3104, and AA5182 by XRF Analysis

Element AA1050 AA3104 AA5182

Mg ,0.01 1.35 3.60
Mn ,0.01 1.04 0.16
Si 0.07 0.42 0.21
Fe 0.33 0.34 0.26
Cu ,0.01 0.14 ,0.01
Zn 0.01 0.08 0.01
Ti ,0.01 0.02 0.02

increased to a final value of 0.02. The results of this so-

Fig. 4—0.2 pct yield stress vs distance from the rolling face for AA3104.
called continuous-cooling test[12] are used to evaluate the
predicting potential of the constitutive equations determined
in the fifth test. Since the measured strain is calculated
from the change in diameter of the specimen, it consists of
contributions from both the elastic and plastic deformation
and the thermal contraction. The latter contribution was
derived in a separate experiment, where the same tempera-
ture path was imposed while keeping the stress equal to
zero. While the thermal contribution corresponds well to
calculations, using a linear expansion coefficient (23.2 3
1026/ 8C), the experimental data were used for the correction.

D. Fitting Procedure
The fitting procedure was restricted to the plastic part and
before fitting, the data were filtered by an adjacent averaging
procedure to reduce the statistical noise on the strain signal
(a)
and to improve the determination of the strain rate during
the test. The data were fitted to the extendedLudwik equation
(Eq. [1]) and the HST equation (Eqs. [3] to [5]). As the
strain hardening becomes zero at high temperatures, nL was
set to zero above 350 8C in the extended Ludwik equation
to facilitate the fitting procedure. Also, as the strain-rate
sensitivity is negligible at low temperatures, mL was set to
zero below 200 8C. At higher temperatures, mL was deduced
from the strain-rate jump performed at a strain of about 0.01.

III. RESULTS
A. Material
The chemical composition of the three alloys, determined (b)
by X-ray fluorescence analysis, is given in Table II. Speci-
men AA5182 has a Mg content of 3.6 wt pct, whereas the
composition limit is 4.0 to 5.0 wt pct.

B. Initial Tests
The effect of specimen location in the slab was investi-
gated for AA3104. The results show that the 0.2 pct yield
stress decreases considerably with increasing distance from
the rolling face (Figure 4). Specimens near the edge yield
at ,115 MPa, whereas specimens at the middle of the slab
yield at ,80 MPa.
A short, initial heat treatment at 550 8C for AA1050 and
AA3104 and at 480 8C for AA5182 strongly influences the (c)
tensile strength (Figure 5). The 0.2 pct yield strength is Fig. 5—Tensile tests of (a) AA1050, (b) AA3104, and (c) AA5182 meas-
reduced by about 10 to 40 MPa for the different alloys. A ured at 50 8C without heat treatment (–––) and after a heat treatment for
heating time of 10 or 60 seconds gives the same result. It 10 s (---) or 60 s (???).

1974—VOLUME 33A, JULY 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

 Fig. 8—The effect of prestrain at 200 8C on the hardening behavior of
Fig. 6—Tensile behavior of AA3104 at 50 8C after cooling from 550 8C AA5182. Prestrain of 0.025 (???), 0.01 (---), and 0.005 (- - -). No
at 0.5 8C/s (---) or 5 8C/s (–––). prestrain (–––).

behavior has been investigated for AA5182 at 200 8C (Figure

8). Clearly, more prestrain leads to a higher yield stress and
less strain hardening.

C. Thermomechanical Behavior
Tensile tests were carried out between 50 8C and the
solidus temperature. The resulting stress-strain curves for
AA1050 are shown in Figure 9(a). The behavior between
50 8C and 200 8C is very similar, and the strain hardening
decreases only slightly with increasing temperature. At 250
8C and above, hardening is strongly reduced. Above 350
8C, hardening is almost absent, and the material shows
Fig. 7—The effect of prestrain on the yield stress at 50 8C. Alloy AA3104 steady-state behavior already at low strain levels. At 500 8C
(M ) was prestrained 1 pct (300 8C, 500 8C) or 2 pct (200 8C, 400 8C). and above, horizontal lines are shown because these tests
AA5182 (n ) was prestrained 1 pct at all temperatures. were load controlled, i.e., creep tests. The corresponding
strain vs time graphs indicated steady-state behavior. Alloys
AA3104 and AA5182 roughly show a similar behavior (Fig-
ures 9(c) and (e)), but the overall stress levels lie much
is also found that specimens from different locations yield higher, and they keep their strength until 300 8C.
very similar results after the heat treatment. This allows Figures 9(b), (d), and (f) show the continuation of the
that, after such a heat treatment, the specimens can be used tensile tests after the strain-rate decrease. For alloy AA1050,
irrespective of their location in the slab. no effect is observed from 50 8C to 200 8C. Above 200 8C,
To study the role of possible precipitation, tensile tests the decrease in strain rate is followed by a gradual decrease
were done at 50 8C, after cooling at 5 8C/s and 0.5 8C/s of the stress level to a new steady-state value. Both the
from 550 8C. Figure 6 shows for AA3104 the stress-strain duration of the transient and the relative change in stress
curves after cooling at different rates. The curves are very level increase with increasing temperature. At 500 8C to 620
similar, indicating that the difference in cooling rate has no 8C, the stress level decreases immediately because these
influence on the tensile behavior of the material over the tests were load controlled. For AA3104, no strain-rate sensi-
range of cooling rates expected during DC casting. tivity is observed below 250 8C. At 300 8C and above, the
The effect of deformation history on yield strength at 50 stress level decreases slowly as was also observed in
8C was studied for alloys AA3104 and AA5182. Figure 7 AA1050. In contrast to the other alloys, the hardening
shows the yield strength after prestraining the specimens at increases after the strain rate decrease at 250 8C. This is
temperatures ranging from 200 8C to 500 8C. The yield attributed to dynamic strain aging in which, at low strain
strength with the same temperature history but without rate, the dislocations are captured by solute atoms of which
prestraining is also given. For both alloys, the results of the speed is limited by their diffusion rate.[25,26] For alloy
prestraining at 400 8C, prestraining at 500 8C, and no AA5182, the strain-rate sensitivity is apparent at 200 8C and
prestraining are comparable, which means that above 400 shows similar behavior to AA1050 at higher temperatures.
8C possible strain hardening is fully recovered. Prestraining Figure 10 shows the total strain in the continuous-cooling
at 200 8C hardens the material significantly. After prestrain- test and the thermal contribution derived from a separate
ing at 300 8C, the material shows intermediate behavior. experiment. By subtracting these two values, the plastic
The effect of the amount of prestrain on the hardening strain is obtained.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JULY 2002—1975

 (a) (b)

(c) (d )

(e) (f)
Fig. 9—Tensile curves of AA1050 (a, b), AA3104 (c, d), and AA5182 (e f ). Graphs on the left-hand side show the tensile test to the point at which the
strain rate was decreased. The strain rate, which varies from test to test, is 5 3 1023 to 1 3 1024 s21. Graphs on the right-hand side show behavior after
the strain rate decrease to 6 3 1024 to 1 3 1025 s21.

D. Constitutive Description IV. DISCUSSION

The tensile curves (Figure 9) were fitted to the extended A. Effect of Specimen Location, Heat Treatment, and
Ludwik equation (Eq. [1]) and the HST equation (Eqs. [3] Cooling Rate
to [5]). The parameters of the extended Ludwik equation
(K, mL , and nL) are shown in Table III and Figure 11. With During the initial tests with AA3104, it is observed that
increasing temperature, mL increases, while K and nL in the as-cast condition the strength of the material depends
decrease. The strain-rate-sensitivity parameter, mL , of on the position in the slab (Figure 4) and that these differ-
AA1050 shows some irregularities at high temperature. ences are erased after a short heat treatment (Figure 5). This
These are due to a relatively large experimental error caused indicates that the differences in strength cannot be due to
by the low stress levels and consequently by the large uncer- macrosegregation, as diffusion to homogenizethe segregated
tainty in the difference between the stress levels before and elements would take much longer. Further, cooling at differ-
after the strain-rate jump. ent rates after the heat treatment does not influence the
The parameters of the HST equation are shown in strength of the material (Figure 6). This indicates that the
Table IV. differences in strength cannot be due to differences in either

1976—VOLUME 33A, JULY 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

the amount or the kind of precipitation. Therefore, we attrib- The a parameter in the HST equation (Eqs. [3] through
ute the differences in strength found across the slab to the [5]) also gives the hardening behavior as a function of tem-
buildup of strain hardening, which is the strongest at the perature. With the a0 and a1 parameters given in Table IV
outside of the slab because here cooling is much faster. and Eq. [5], a is calculated (Figure 12). This figure fully
Strain hardening is indeed annihilated very fast at high tem- confirms, for AA3104 and AA5182, the absence of harden-
perature by recovery processes. ing above 400 8C, the temperature dependency of hardening
between 200 8C and 400 8C, and the strongly reduced temper-
ature dependency of hardening below 200 8C. For alloy
B. Effect of Deformation History AA1050, the temperature dependency of the hardening
extends from room temperature to 450 8C.
The results of tests 4 and 5 (Figures 7 through 9) can be
Fjaer and Mo[1] have developed a mathematical model to
summarized as follows. The yield stress at 50 8C is not
calculate the thermal stresses and strains during DC casting.
influenced by prestraining at 400 8C and above; between
To implement the constitutive behavior of the alloy, they
200 8C and 400 8C, the yield stress increases with decreasing
have applied a step function in which the amount of harden-
prestrain temperature, and below 200 8C, prestraining has a
ing depends on the amount of strain below a certain tempera-
similar effect on hardening, almost independent of tempera-
ture and is zero above this temperature. The transition
ture. Furthermore, the yield stress increases, and the strain-
temperature was 700 K, determined from experiments in
hardening rate decreases with increasing prestrain at lower
Reference 10. This temperature corresponds well to our
temperatures. These observations are explained by the fact
measurements, which demonstrates that hardening is negligi-
that at the highest temperatures recovery is so fast that the
ble above 400 8C. However, our measurements also show
effect of hardening is annihilated. Below 400 8C, recovery is
that between 400 8C and 200 8C hardening depends not only
reduced, and the effect of prestraining becomes increasingly
on strain but also on temperature (Figure 12). Therefore,
important until 200 8C. Below 200 8C, recovery is less
Eq. [5] provides a better description of the hardening behav-
important, and the temperature dependency of the strain
ior with respect to temperature.
hardening is reduced.

C. Constitutive Behavior
Figure 13 shows the measured stress-strain curves and
calculated curves according to the extendedLudwik equation
(Eq. [1]) and the HST equation (Eqs. [3] through [5]). The
measured stress-strain curves are well approximated by the
extended Ludwik equation. The HST equation describes, for
low and high temperatures, the thermomechanical behavior
of the three alloys well, although less well than the extended
Ludwik equation. The reason is that the HST equation fits
all curves simultaneously, while the extended Ludwik equa-
tion is optimized for each temperature individually.The HST
equation has some difficulties in describing the experimental
data in the intermediate temperature regime, i.e., where both
strain hardening occurs and where the stress is strain-rate
dependent. This is mainly apparent for AA5182 at 300 8C.
The advantage of the use of the HST equation instead of
Fig. 10—Total strain (–––) and thermal contribution (???) during the contin- the extended Ludwik equation is the temperature independ-
uous cooling test. ence of the parameters applied.

Table III. Parameters of the Extended Ludwik Equation (Eq. [1]) for AA1050, AA3104, and AA5182

AA1050 AA3104 AA5182

T (8C) K nL mL K nL mL T (8C) K nL mL
50 119 0.388 0 290 0.250 0 50 315 0.241 0
100 96 0.346 0 273 0.265 0 100 308 0.218 0
150 83 0.312 0 290 0.272 0 150 258 0.195 0
200 64 0.265 0 268 0.247 0 200 256 0.167 0.015
250 49 0.200 0.019 260 0.237 0 250 228 0.137 0.032
300 33 0.125 0.042 250 0.137 0.073 300 229 0.004 0.130
350 19 0.025 0.079 240 0.039 0.179 370 214 0 0.220
400 16 0 0.124 117 0 0.209 420 221 0 0.302
450 13 0 0.145 102 0 0.249 470 221 0 0.352
500 8 0 0.128 107 0 0.270 — — — —
550 6 0 0.124 — — — — — — —
600 6 0 0.151 — — — — — — —
620 11 0 0.231 — — — — — — —

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JULY 2002—1977

 Concerning the fitting parameters found, it is observed increases in the order AA1050, AA3104, and AA5182. This
that the rheological parameters, K, nL , and mL of the extended is attributed to the solute-drag deformation mechanism
Ludwik equation (Table III and Figure 11) vary with temper- caused by the presence of solute Mg[27] of which the concen-
ature in a similar manner to those for other aluminum tration also increases in the order AA1050, AA3104, and
alloys.[7,8,10,27] At high temperature, the strain-rate sensitivity AA5182. In AA3104, Mn in solid solution may also increase
the strain-rate sensitivity.[28]
Concerning the parameters of the HST equation, it is
noted that above 350 8C, a becomes zero, and the HST
equation simplifies to the Sellars–Tegart equation. There-
fore, the found parameters A, s0, mH, and Q in Table IV
can be compared directly to literature values of the Sellars–
Tegart equation (Table I). It is observed that the parameters
found in this study for as-cast material correspond better to
data reported for cast and homogenized material[18] than for
as-cast material.[3,12] However, it should be kept in mind that
there is quite some scatter in the data and that published
data for as-cast material are very limited.
At high temperature and low stress levels (i.e., at s /s0
, 0.8), the HST equation reduces to a power-law equation.
The same is valid for the extended Ludwik equation at
high temperature because nL becomes zero. So under these
(a)
circumstances, it is expected that mL equals mH and that
K/« m0 L equals s0 /AmH exp (mHQ/RT ). Comparing the mL
values (Table III) with the mH values (Table IV), it is indeed
observed that these parameters approach each other above
350 8C. The same observation is made when comparing the
values of K/« m0 L (Table III) with the values of the expression
s0 /AmH exp (mHQ/RT ), calculated with the parameters given
in Table IV.

D. Continuous-Cooling Test
Figure 14 shows the imposed and calculated stresses of
the continuous-cooling test of AA5182. For the calculation,
the stresses were derived from the plastic strain and strain
rate measured during the continuous cooling test, using Eq.
(b)

(c)
Fig. 11—Ludwik parameters vs temperature: (a) K, (b) nL , and (c) mL for Fig. 12—The temperature dependence of a: AA1050 (- - -), AA3104 (???),
AA1050 (c ), AA3104 (M ), and AA5182 (m ). and AA5182 (–––).

Table IV. Parameters of Eqs. [3] through [5] for AA1050, AA3104, and AA5182
Alloy s0 (MPa) k (MPa) a0 a1 (K21) A (s21) Q (kJ/mol) mH
10
AA1050 10.1 127 26.75 1.46 3 10 22
3.02 3 10 156 0.169
AA3104 48.0 400 238.6 6.85 3 1022 6.0 3 109 155 0.217
AA5182 52.0 481 218.1 3.53 3 1022 2.65 3 107 125 0.291

1978—VOLUME 33A, JULY 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

 (a)

Fig. 14—Comparison of imposed stress and calculated stresses in continu-

ous cooling test of AA5182. Imposed stress (–––), calculated stress with
Eq. [1] (- - -), and calculated stress with Eqs. [3] through [5] (???).

Farup and Mo[13] have done similar tests with AA3103.

Specimens were subjected to different cooling and deforma-
tion histories (i.e., to simulate different locations in the
ingot), and they compared the measured stresses with predic-
tions from a previously determined creep law.[12] They found
that above 400 8C, the correspondence is good but that
below this temperature the calculated stresses increasingly
(b) overestimate the measured stresses. They relate this differ-
ence to work hardening, which becomes increasingly
important at decreasing temperatures and which is not taken
into account in the creep law. Our calculations indicate that
by taking into account strain hardening, the correspondence
of the calculated and measured data is indeed improved
(Figure 14).

V. CONCLUSIONS
The constitutive behavior of three aluminum alloys
(AA1050, AA3104, and AA5182) was determined in the
as-cast condition by tensile testing.
The following conclusions were made.
(c)
Fig. 13—Comparison of measured data (–––) with extended Ludwik 1. The constitutive behavior of three commercial aluminum
equation (- - -) and HST equation (???) for (a) AA1050, (b) AA3104, and alloys (AA1050, AA3104, and AA5182) in the as-cast
(c) AA5182. Strain rate decrease is indicated with vertical line.
condition is well described from room temperature to
solidus temperature by both the extended Ludwik equa-
tion and an adapted form of the Sellars–Tegart equation.
[1] and Eqs. [3] through [5] with the parameters determined Although the extended Ludwik equation describes the
during the isothermal tests (Table III and IV). Above 270 data better, the adapted form of the Sellars–Tegart equa-
8C, the stress predicted by the extended Ludwik equation tion is easier to implement in DC casting models because
(Eq. [1]) corresponds very well to the measured stress. Below the temperature appears explicitly in this equation.
270 8C, the measured stress is underestimated by up to 15 2. Differences in strength found across the slab are due to
pct. The HST equation (Eqs. [3] through [5]) underestimates the buildup of strain hardening, which is the strongest at
the measured stress below 370 8C and gives results similar the outside of the slab due to faster cooling.
to the extended Ludwik equation below 270 8C. Despite 3. Above 400 8C, recovery is so fast that strain hardening
the underestimation in the lower temperature region, it is is absent. Below 400 8C, the strain hardening and defor-
concluded that especially the extended Ludwik equation mation history becomes increasingly important until 200
describes the data of the continuous-cooling test reasonably 8C. Below 200 8C, recovery is negligible, and hardening
well and that it can thus be applied in DC casting models. is independent of temperature.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JULY 2002—1979

 ACKNOWLEDGMENTS 10. M.L. Nedreberg: Ph.D. Thesis, University of Oslo, Oslo, Norway,
1991.
This research was carried out as part of the EMPACT 11. M.A. Wells, I.V. Samarasekera, E.B. Hawbolt, J.K. Brimacombe, and
Brite-Euram project (BRPR-CT95-0112). Funding by the R. Bolingbroke: in Light Metals Processing and Applications, C.
Bickert and M. Bouchard, eds., Quebec City, PQ, Canada, 1993, pp.
European Community is greatly acknowledged. We thank 483-93.
Professor A. Mo (SINTEF Materials Technology, Oslo, 12. I. Farup, J.-M. Drezet, A. Mo, and T. Iveland: J. Thermal Stresses,
Norway) and Dr. J.-M. Drezet (Ecole PolytechniqueFédérale 2000, vol. 23, pp. 47-58.
de Lausanne, Switzerland) for fruitful discussions. We very 13. I. Farup and A. Mo: J. Thermal Stresses, 2000, vol. 23, pp. 71-89.
much appreciate the use of the experimental equipment at 14. F. Garofalo: Trans. TMS-AIME, 1963, vol. 227, pp. 351-55.
15. J.J. Jonas, C.M. Sellars, and W.J.McG. Tegart: Met. Rev., 1969, vol.
Hydro Aluminium, R&D (Sunndalsøra, Norway) and thank 14, pp. 1-24.
Mr. T. Iveland (Hydro Aluminium a.s., Sunndalsøra, 16. R.A. Ayres: Metall. Trans. A, 1979, vol. 10A, pp. 849-54.
Norway) for technical assistance. The supply of material by 17. D. Park and J.G. Morris: Metall. Mater. Trans. A, 1994, vol. 25A, pp.
VAW, Bonn, Germany (AA1050, AA5182), and Pechiney 357-64.
18. T. Sheppard and A. Jackson: Mater. Sci. Technol., 1997, vol. 13, pp.
CRV, Voreppe, France (AA3104), is greatly appreciated. 203-09.
19. H. Oikawa and T.G. Langdon: in Creep Behaviour of Crystalline
Solids, B. Wilshire and R.W. Evans, eds., Pineridge Press, Swansea,
REFERENCES United Kingdom, 1985, pp. 33-82.
20. H.J. Frost and M.F. Ashby: Deformation-Mechanism Maps, Pergamon
1. H.G. Fjaer and A. Mo: Metall. Trans. B., 1990, vol. 21B, pp. Press, Oxford, United Kingdom, 1982.
1049-61. 21. L. Ohm and S. Engler: Giessereiforschung, 1990, vol. 42, pp. 131-62.
2. B. Magnin, L. Katgerman, and B. Hannart: in MCWASP VII, M. Cross 22. J.A. Spittle, S.G.R. Brown, J.D. James, and R.W. Evans: in Physical
and J. Campbell, eds., TMS, Warrendale, PA, 1995, pp. 303-10. Simulation of Casting, Hot Rolling and Welding, National Research
3. J.-M. Drezet and M. Rappaz: Metall. Mater. Trans. A, 1996, vol. 27A, Institute for Metals, Tokyo, 1997, pp. 81-91.
pp. 3214-25. 23. A. Mo, I. Farup, and J.-M. Drezet: ESAFORM Conf. on Material
4. G.-U. Grün, A. Buchholz, and D. Mortensen: in Light Metals 2000, Forming, J.L. Chenot, J.F. Agassant, P. Montmitonnet, B. Vergnes,
R.D. Peterson, ed., TMS, Warrendale, PA, 2000, pp. 573-78. and N. Billon, eds., Sophia Antipolis, France, 1998, pp. 1-4.
5. J.-M. Drezet, A. Burghardt, H.G. Fjaer, and B. Magnin: Mater. Sci. 24. J.M. Rabenberg, I.J. Opstelten, J.C. Storm, and J.-M. Drezet: in Contin-
Forum, 2000, vols. 329–330, pp. 493-99. uous Casting, K. Ehrke and W. Schneider, eds., Wiley-VCH, Wein-
6. P. Ludwik: Elemente der Technologischen Mechanik, Springer-Verlag heim, Germany, 2000, pp. 71-76.
OHG, Berlin, 1909. 25. A. van den Beukel: Phys. Status Solidi A, 1975, vol. 30A, pp. 197-206.
7. A.M. Hammad and K.K. Ramadan: Z. Metallkd., 1989, vol. 80, pp. 26. P.G. McCormick: Acta Metall., 1988, vol. 36, pp. 3061-67.
431-38. 27. E.M. Taleff, P.J. Nevland, and P.E. Krajewski: in Creep Behaviour of
8. B. Magnin, L. Maenner, L. Katgerman, and S. Engler: Mater. Sci. Advanced Materials for the 21st Century, R.S. Mishra, A.K.
Forum, 1996, vols. 217–222, pp. 1209-14. Mukherjee, and K.L. Murty, eds., TMS, Warrendale, PA, 1999, pp.
9. W.M. van Haaften, B. Magnin, W.H. Kool, and L. Katgerman: in 349-58.
Light Metals 1999, C.E. Eckert, ed., TMS, Warrendale, PA, 1999, pp. 28. L.F. Mondolfo: Aluminium Alloys, Structure & Properties, Butter-
829-33. worth & Co, London, 1976.

1980—VOLUME 33A, JULY 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

View publication stats