CRUSHING OF ALUMINUM CLOSED CELL FOAMS:
DENSITY AND STRAIN RATE EFFECTS
I.W. Hall, M. Guden* and C.-J. Yu
†
Department of Mechanical Engineering, University of Delaware, Newark, DE 19716 *Izmir Yuksek
Teknoloji Enstitusu, Izmir, Turkey
†
Fraunhofer Center-Delaware, 501 Wyoming Road,
Newark, DE 19716
(Received October 1, 1999)
(Accepted in revised form April 11, 2000)
Introduction
Potential applications of metal foams include light weight cores for sandwich panels, shells and tubes
where the foam can increase the resistance to local buckling, increase the impact resistance, and
improve the energy absorbing capacity of the structure [1,2]. This latter property offers potential uses
in transportation applications where, for example, foam-filling of the hollow sections of automobiles,
such as fenders, may reduce damage and injuries resulting from impact accidents. For this type of
application, aluminum foam is more suitable than a polymeric foam, because it deforms plastically
under impact and with essentially no spring back, preventing further damage [3]. Other important
advantages of using aluminum foams over polymeric foams include high fire resistance and insensitivity to cold and hot weather and humidity [3].
Impact accidents produce loading rates which are higher than those of static or quasi-static rates and
which may significantly alter mechanical response of the materials. Therefore, in designing with
metallic foams as energy absorbing fillers, mechanical properties are needed for strain rates corresponding to those created by impact events. Quasi-static mechanical behavior of metallic foams has
been fairly extensively studied and reported, e.g. [3–5], but data concerning high strain rate mechanical
behavior of these materials are, however, only just becoming available and are rather sparse [6,7]. This
study was initiated, therefore, to study the high strain rate mechanical behavior of a range of metallic
foams, and to compare it with quasi-static behavior and, hence, determine any effect on energy
absorbing capacity. Microscopic observations were also made in order to clarify the deformation
mechanisms involved during crushing of the foam.
Material and Testing Methods
The metal foam investigated in this study was manufactured by the Fraunhofer Center - Delaware, using
a patented powder metallurgical technique [8]. One of the advantages of this method is that it is possible
to manufacture complex structures to near net shapes. The foam was prepared using 6061-Al alloy
powder and TiH2as the foaming agent and was supplied as rectangular plate (50 mm thick and 100 mm
wide) with a typical closed cell structure. The cell size varied somewhat through the thickness and
Scripta mater. 43 (2000) 515–521
www.elsevier.com/locate/scriptamat
1359-6462/00/$–see front matter. © 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved.
PII: S1359-6462(00)00460-7
resulted, therefore, in a slight density gradient. Typical cell sizes were 4 –5mm in the mid-sections and
2–3mm near to the external surfaces.
Quasi-static compression tests were conducted using a displacement controlled Instron machine at
an initial strain rate of 1.5310
23
s
21
. High strain rate tests were conducted using a compression-type
Split Hopkinson Pressure Bar (SHPB), also known as a Kolsky Bar. Briefly, in this technique a
cylindrical specimen is mounted between the long incident and transmitter bars, while a shorter striker
bar is used to impact the end of the incident bar. This impact creates a compressive pulse which travels
down the incident bar and into the specimen which then deforms as it is sandwiched between incident
and transmitter bars.
Conventionally, the strains measured by strain gages on the incident and transmitter bars are then
used to determine strain rate, strain and stress in the specimen using one dimensional elastic wave
analysis [9]. However, tests on foams violate several of the fundamental assumptions of this type of
analysis: for example, the volume does not remain constant during testing and the elastic wave velocity
depends on the density which itself changes during the test. Nevertheless, and bearing in mind that these
are non-standard tests with limitations, it is found that conventionally-reduced SHPB foam data
resemble the stress-strain curves which would be anticipated from quasi-static tests. Consequently, and
also since an acceptable alternative means of data presentation has not yet been developed, the data are
presented as engineering stressvs.engineering strain curves.
High strain rate tests were conducted on the foam at strain rates up to;2.0310
3
s
21
. The strain
rates involved in automobiles during impact accidents are most frequently in the range between;13
10
2
and;1310
3
s
21
and the investigated strain rates are, therefore, relevant to impacts involved in
such accidents.
The cylindrical foam specimens were 18mm in diameter and 12mm tall. Some scatter was observed
in the data (see Fig. 1 below) but was not reduced by the use of larger samples. The diameter of the
samples was slightly less than that of the bars (19mm) and, due to the small Poisson’s ratio of the foam,
specimen diameters during deformation never exceeded the bar diameter within the studied strain levels
up to;60% strain. The relatively large diameter of the SHPB used was advantageous, since larger
volumes of foam could be tested, thus reducing statistical scatter somewhat. By conducting tests on
identically sized specimens in the Instron, any significant effect of specimen volume on the measured
flow stress was avoided.
Flow stresses reported below were determined at 10% strain since a) severe initial stress drops were
noted in many of the tests and b) an inherent feature of the SHPB apparatus is that flow stress values
Figure 1. Quasi-static eng. stress vs. eng. strain curves for samples of two different nominal densities.
ALUMINUM FOAMS 516 Vol. 43, No. 6
at small strains, such as the yield strain, can not be determined due to the absence of stress equilibrium
in the early stages of the test.
Foam blocks with average densities of 0.34, 0.47, 0.57 and 0.82gcm
23
were used. Specimens were
core-drilled through the thickness direction of the foam blocks and then individually weighed and
measured in order to calculate the actual and relative densities before testing. The relative density,r*,
is calculated as:
r*5
rf
rm
3100% (1)
where f and m refer to the foam and precursor (matrix) alloy respectively. The average relative densities
of the four foam blocks were, therefore, 12.6%, 17.4%, 21.1% and 30.4% respectively.
Results
Typical quasi-static compressive engineering stress-strain curves from 3 samples of each of the lowest
and highest density foams are shown in Fig. 1. It is seen that the flow stress is a strong function of
relative density and the curves show the typical shape which may be divided into three regions, 1) linear
elastic region, 2) collapse region and finally, 3) a densification region in which the properties approach
those of fully-dense alloy. It is also seen that, for the low density foams, a load drop occurs after the
initial elastic extension while the higher density ones exhibited no load drop. Also, stress oscillations
occur in the collapse region of low density samples while the higher density ones showed an essentially
monotonic increase in stress throughout the tests. The flow stress in the collapse region increased as the
relative density increased, as expected, but the rate of densification was also higher in the denser
samples. Strains in excess of 60% were easily achieved in all cases.
Foams were then tested in the SHPB and Fig. 2 shows typical stress-strain curves for foams of three
different densities tested at 2310
3
s
21
. Again, the flow stress values of the foam are seen to be a strong
function of relative density although it is also noted that the areas under the stress strain curves, which
scale with the energy absorbed, do not scale linearly with the density. The respective areas under the
curves were calculated between 5% and 25% strain, the latter being the maximum strain obtainable at
the lowest strain rate in the SHPB, and the data are plotted as a function of density in Fig. 3. Fig. 4
shows data for foams, with individually measured densities of 0.52gcm
23
, tested over a range of strain
rates from quasi-static to 2310
3
s
21
and up to;25% strain. Even when minor variations in density
Figure 2. Stress vs. strain curves at 2310
3
s
21
for samples of three different (individually measured) densities.
ALUMINUM FOAMS 517 Vol. 43, No. 6
are removed, there is still no unequivocal effect of strain rate within this strain range and the areas under
the curves, were approximately equal. The absence of marked strain rate sensitivity is in agreement with
other observations for this and other aluminum alloys [10,11].
The sequence of deformation events was recorded during quasi-static compression testing and
several frames of a typical record are presented in Fig. 5. The time at each frame is marked and the
compression loading axis was horizontal. Frame 1, 0s, shows an undeformed region containing several
cells. As soon as compression starts the thinnest sections of cell walls approximately parallel to the
compression axis start to buckle, 20s. In the later stages of deformation, frames at 90s and 120s, cell
walls normal to the compression axis tear due to the induced tensile strains. It may also be noted that
deformation is highly inhomogeneous and that cells to the left are completely undeformed while
adjacent ones are almost totally closed up.
Microscopic examination of cell walls showed clear evidence of the original powder particles, Fig.
6; the walls were made up of many particles sintered together confirming that complete melting had not
occurred. These surfaces also exhibited particles which were identified by EDS measurements as being
rich in Ti (arrowed in Fig. 6). The
CRUSHING OF ALUMINUM CLOSED CELL FOAMS:
DENSITY AND STRAIN RATE EFFECTS
I.W. Hall, M. Guden* and C.-J. Yu
†
Department of Mechanical Engineering, University of Delaware, Newark, DE 19716 *Izmir Yuksek
Teknoloji Enstitusu, Izmir, Turkey
†
Fraunhofer Center-Delaware, 501 Wyoming Road,
Newark, DE 19716
(Received October 1, 1999)
(Accepted in revised form April 11, 2000)
Introduction
Potential applications of metal foams include light weight cores for sandwich panels, shells and tubes
where the foam can increase the resistance to local buckling, increase the impact resistance, and
improve the energy absorbing capacity of the structure [1,2]. This latter property offers potential uses
in transportation applications where, for example, foam-filling of the hollow sections of automobiles,
such as fenders, may reduce damage and injuries resulting from impact accidents. For this type of
application, aluminum foam is more suitable than a polymeric foam, because it deforms plastically
under impact and with essentially no spring back, preventing further damage [3]. Other important
advantages of using aluminum foams over polymeric foams include high fire resistance and insensitivity to cold and hot weather and humidity [3].
Impact accidents produce loading rates which are higher than those of static or quasi-static rates and
which may significantly alter mechanical response of the materials. Therefore, in designing with
metallic foams as energy absorbing fillers, mechanical properties are needed for strain rates corresponding to those created by impact events. Quasi-static mechanical behavior of metallic foams has
been fairly extensively studied and reported, e.g. [3–5], but data concerning high strain rate mechanical
behavior of these materials are, however, only just becoming available and are rather sparse [6,7]. This
study was initiated, therefore, to study the high strain rate mechanical behavior of a range of metallic
foams, and to compare it with quasi-static behavior and, hence, determine any effect on energy
absorbing capacity. Microscopic observations were also made in order to clarify the deformation
mechanisms involved during crushing of the foam.
Material and Testing Methods
The metal foam investigated in this study was manufactured by the Fraunhofer Center - Delaware, using
a patented powder metallurgical technique [8]. One of the advantages of this method is that it is possible
to manufacture complex structures to near net shapes. The foam was prepared using 6061-Al alloy
powder and TiH2as the foaming agent and was supplied as rectangular plate (50 mm thick and 100 mm
wide) with a typical closed cell structure. The cell size varied somewhat through the thickness and
Scripta mater. 43 (2000) 515–521
www.elsevier.com/locate/scriptamat
1359-6462/00/$–see front matter. © 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved.
PII: S1359-6462(00)00460-7
resulted, therefore, in a slight density gradient. Typical cell sizes were 4 –5mm in the mid-sections and
2–3mm near to the external surfaces.
Quasi-static compression tests were conducted using a displacement controlled Instron machine at
an initial strain rate of 1.5310
23
s
21
. High strain rate tests were conducted using a compression-type
Split Hopkinson Pressure Bar (SHPB), also known as a Kolsky Bar. Briefly, in this technique a
cylindrical specimen is mounted between the long incident and transmitter bars, while a shorter striker
bar is used to impact the end of the incident bar. This impact creates a compressive pulse which travels
down the incident bar and into the specimen which then deforms as it is sandwiched between incident
and transmitter bars.
Conventionally, the strains measured by strain gages on the incident and transmitter bars are then
used to determine strain rate, strain and stress in the specimen using one dimensional elastic wave
analysis [9]. However, tests on foams violate several of the fundamental assumptions of this type of
analysis: for example, the volume does not remain constant during testing and the elastic wave velocity
depends on the density which itself changes during the test. Nevertheless, and bearing in mind that these
are non-standard tests with limitations, it is found that conventionally-reduced SHPB foam data
resemble the stress-strain curves which would be anticipated from quasi-static tests. Consequently, and
also since an acceptable alternative means of data presentation has not yet been developed, the data are
presented as engineering stressvs.engineering strain curves.
High strain rate tests were conducted on the foam at strain rates up to;2.0310
3
s
21
. The strain
rates involved in automobiles during impact accidents are most frequently in the range between;13
10
2
and;1310
3
s
21
and the investigated strain rates are, therefore, relevant to impacts involved in
such accidents.
The cylindrical foam specimens were 18mm in diameter and 12mm tall. Some scatter was observed
in the data (see Fig. 1 below) but was not reduced by the use of larger samples. The diameter of the
samples was slightly less than that of the bars (19mm) and, due to the small Poisson’s ratio of the foam,
specimen diameters during deformation never exceeded the bar diameter within the studied strain levels
up to;60% strain. The relatively large diameter of the SHPB used was advantageous, since larger
volumes of foam could be tested, thus reducing statistical scatter somewhat. By conducting tests on
identically sized specimens in the Instron, any significant effect of specimen volume on the measured
flow stress was avoided.
Flow stresses reported below were determined at 10% strain since a) severe initial stress drops were
noted in many of the tests and b) an inherent feature of the SHPB apparatus is that flow stress values
Figure 1. Quasi-static eng. stress vs. eng. strain curves for samples of two different nominal densities.
ALUMINUM FOAMS 516 Vol. 43, No. 6
at small strains, such as the yield strain, can not be determined due to the absence of stress equilibrium
in the early stages of the test.
Foam blocks with average densities of 0.34, 0.47, 0.57 and 0.82gcm
23
were used. Specimens were
core-drilled through the thickness direction of the foam blocks and then individually weighed and
measured in order to calculate the actual and relative densities before testing. The relative density,r*,
is calculated as:
r*5
rf
rm
3100% (1)
where f and m refer to the foam and precursor (matrix) alloy respectively. The average relative densities
of the four foam blocks were, therefore, 12.6%, 17.4%, 21.1% and 30.4% respectively.
Results
Typical quasi-static compressive engineering stress-strain curves from 3 samples of each of the lowest
and highest density foams are shown in Fig. 1. It is seen that the flow stress is a strong function of
relative density and the curves show the typical shape which may be divided into three regions, 1) linear
elastic region, 2) collapse region and finally, 3) a densification region in which the properties approach
those of fully-dense alloy. It is also seen that, for the low density foams, a load drop occurs after the
initial elastic extension while the higher density ones exhibited no load drop. Also, stress oscillations
occur in the collapse region of low density samples while the higher density ones showed an essentially
monotonic increase in stress throughout the tests. The flow stress in the collapse region increased as the
relative density increased, as expected, but the rate of densification was also higher in the denser
samples. Strains in excess of 60% were easily achieved in all cases.
Foams were then tested in the SHPB and Fig. 2 shows typical stress-strain curves for foams of three
different densities tested at 2310
3
s
21
. Again, the flow stress values of the foam are seen to be a strong
function of relative density although it is also noted that the areas under the stress strain curves, which
scale with the energy absorbed, do not scale linearly with the density. The respective areas under the
curves were calculated between 5% and 25% strain, the latter being the maximum strain obtainable at
the lowest strain rate in the SHPB, and the data are plotted as a function of density in Fig. 3. Fig. 4
shows data for foams, with individually measured densities of 0.52gcm
23
, tested over a range of strain
rates from quasi-static to 2310
3
s
21
and up to;25% strain. Even when minor variations in density
Figure 2. Stress vs. strain curves at 2310
3
s
21
for samples of three different (individually measured) densities.
ALUMINUM FOAMS 517 Vol. 43, No. 6
are removed, there is still no unequivocal effect of strain rate within this strain range and the areas under
the curves, were approximately equal. The absence of marked strain rate sensitivity is in agreement with
other observations for this and other aluminum alloys [10,11].
The sequence of deformation events was recorded during quasi-static compression testing and
several frames of a typical record are presented in Fig. 5. The time at each frame is marked and the
compression loading axis was horizontal. Frame 1, 0s, shows an undeformed region containing several
cells. As soon as compression starts the thinnest sections of cell walls approximately parallel to the
compression axis start to buckle, 20s. In the later stages of deformation, frames at 90s and 120s, cell
walls normal to the compression axis tear due to the induced tensile strains. It may also be noted that
deformation is highly inhomogeneous and that cells to the left are completely undeformed while
adjacent ones are almost totally closed up.
Microscopic examination of cell walls showed clear evidence of the original powder particles, Fig.
6; the walls were made up of many particles sintered together confirming that complete melting had not
occurred. These surfaces also exhibited particles which were identified by EDS measurements as being
rich in Ti (arrowed in Fig. 6). The
正在翻譯中..
