Mechanics Research Communications 57 (2014) 49–56

Contents lists available at ScienceDirect

Mechanics Research Communications

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

The effect of microstructural morphology on the elastic, inelastic, and

degradation behaviors of aluminum–alumina composites
Pradeep Gudlur a , Anastasia Muliana a,∗ , Miladin Radovic a,b
a
Mechanical Engineering Department, Texas A&M University, United States
b
Materials Science and Engineering Department, Texas A&M University, United States

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

Article history: Micromechanical models with idealized and simplified shapes of inhomogeneities have been widely
Received 22 November 2013 used to obtain the average (macroscopic) mechanical response of different composite materials. The
Accepted 10 February 2014 main purpose of this study is to examine whether the composites with irregular shapes of inhomo-
Available online 20 February 2014
geneities, such as in the aluminum–alumina (Al–Al2 O3 ) composites, can be approximated by considering
idealized and simplified shapes of inhomogeneities in determining their overall macroscopic mechani-
Keywords:
cal responses. We study the effects of microstructural characteristics, on mechanical behavior (elastic,
Aluminum–alumina composites
inelastic, and degradation) of the constituents, and shapes and distributions of the pores and inclu-
Micro-structures
Elastic
sions (inhomogeneities), and thermal stresses on the overall mechanical properties and response of the
Inelastic Al–Al2 O3 composites. Microstructures of a composite with 20% alumina volume content are constructed
Thermal stress from the microstructural images of the composite obtained by scanning electron microscopy (SEM). The
SEM images of the composite are converted to finite element (FE) meshes, which are used to determine
the overall mechanical response of the Al–Al2 O3 composite. We also construct micromechanics model
by considering circular shapes of the inhomogeneities, while maintaining the same volume contents
and locations of the inhomogeneities as the ones in the micromechanics model with actual shapes of
inhomogeneities. The macroscopic elastic and inelastic responses and stress fields in the constituents
from the micromechanics models with actual and circular shapes of inhomogeneities are compared and
discussed.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction can affect the overall mechanical properties and performance of

the composites.
The overall response of aluminum (Al)–alumina (Al2 O3 ) In order to understand the effects of different microstructural
composites depend on the microstructural characteristics of characteristics and behaviors of different constituents on the
the composites, such as composition and distribution of the overall performance of composites, micromechanical models are
constituents, amount of the porosity, size and shape of the micro- often used. There have been several micromechanical models
constituents, which are strongly influenced by the processing formulated for predicting the overall mechanical properties of
methods. Another factor that could influence the mechanical prop- metal–matrix composites. Most of these micromechanical models
erties and performance of Al–Al2 O3 composites is the residual are derived based on a volume-averaging scheme on idealized com-
stress. The residual stresses can arise due to temperature changes posite microstructures, e.g., Aboudi (1985), Eischen and Torquato
during the composite processing, which in this case are known as (1993), Dvorak and Srinivas (1999), Khan et al. (2011). Several
thermal stresses. High thermal stresses can cause yielding in ductile micromechanical studies (Rasool and Böhm, 2012; Qin et al.,
constituents such as in the aluminum, while high tensile thermal 1999) have considered the effects of particle shapes on the overall
stresses can induce cracking in the brittle constituents such as alu- thermo-mechanical properties of ceramic-metal composites, with
mina. Both plasticdeformation and cracking of micro constituents a relatively low volume content of ceramics (20% or less). They
considered composites with spherical, cubical, octahedral, and
tetragonal shapes, and showed that the particle shapes have rather
insignificant effect on the overall linear thermo-elastic mechan-
ical properties of the composites. Rasool and Böhm (2012) also
∗ Corresponding author. showed that the particle shapes have more effects on the con-
E-mail addresses: [email protected], [email protected] (A. Muliana). duction properties. Composites with sharp edged particles lead to

http://dx.doi.org/10.1016/j.mechrescom.2014.02.003
0093-6413/© 2014 Elsevier Ltd. All rights reserved.
50 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56

higher micro-fields, i.e., stresses, near the particle-matrix inter-

faces, which potentially cause debonding. Williams et al. (2012)
examined the effects particle shapes and interfacial decohesion on
the overall tensile stress-strain responses of ceramic-metal com-
posites with 20% ceramic volume contents. They showed that in
absence of the interfacial decohesion the effect of particle shapes
is quite insignificant when the strain is relatively small, and devi-
ations are observed under a relatively large strain. However, when
interfacial decohesion is considered, the effect of particle shapes
on the overall stress-strain responses are pronounced. There are
a limited number of micromechanics models of composites that
incorporate the detailed microstructural morphologies of the metal
matrix composites. Of these, image based finite element soft-
ware OOF (Langer et al., 2007) is one of the latest techniques to
incorporate detailed microstructural morphologies of the compos-
ite. OOF was used to determine thermal residual stress, for e.g.,
in polycrystalline alumina by Zimmermann et al. (1999), in alu-
mina phase of borosilicate–alumina composite by Cannillo et al.
(2002). Chawla et al. (2004) used OOF to determine Young’s modu-
lus and thermal expansion coefficient of SiC reinforced Al matrix
composites. Gudlur et al. (2012, 2014) conducted a numerical
study to examine the effects of microstructural morphologies,
plastic deformations in the aluminum constituent, and ‘stress-
free’ temperature on the overall elastic modulus, Poisson’s ratio,
and thermal expansion coefficient of the Al–Al2 O3 composites.
There are several advantages of using the micromechanical models
with microstructural details in modeling the overall mechani-
cal response of the composites. First, it allows incorporation of
different distribution of constituents in the composite and their
behaviors. Second, micromechanical models can capture the stress
and strain concentrations (or discontinuities) at various locations
within the microstructures of the composites and also take the Fig. 1. (a) SEM image of A-20 composite sample; (b) four random square micro-
residual stresses into account in predicting the overall response graphs of A-20 composite sample and their corresponding FE microstructural
of the composites. However, they are generally computationally models 1, 2, 3, and 4 (from Gudlur et al. 2014).

expensive, which make them impractical to be incorporated in

performing large scale structural analyses. 2. Micromechanical models
In this study, we examine the effects of shapes of the inhomo-
geneities (inclusions and pores) that are randomly dispersed in the The microstructural models of the Al–Al2 O3 composites with
homogeneous ductile matrix on the overall mechanical response 20% alumina volume content are generated from the SEM micro-
of Al–Al2 O3 composites. The study is done on the composites with structural images, as shown in Fig. 1. The size of the microstructural
20% alumina volume content. The first micromechanical model image is 217 ␮m × 173 ␮m. This micrograph image is divided
is constructed from the microstructural images of the compos- into uniform sub-images (regions). As an example, the image is
ite obtained from the scanning electron microscope (SEM). The divided into 12 uniform regions with 50 ␮m × 50 ␮m. Four different
SEM images of the composite are converted into finite element 50 ␮m × 50 ␮m square regions having different microstructures
(FE) meshes using software OOF2 1 and ABAQUS FE code is used are randomly chosen as representative microstructures of the com-
to analyze the overall mechanical response of the Al–Al2 O3 com- posite. Using software OOF2 , the aluminum matrix, the pores and
posite. Fig. 1 shows the SEM image of 20 vol% composite, where the alumina particles in each of the selected microstructure are
the alumina particles appear light in color, aluminum matrix in determined based on their color contrast. After defining the geom-
gray, and voids/pores appear dark in color. The second microme- etry and boundaries of the different pixel groups, meshes are
chanics model is constructed by considering idealized shapes of the generated and converted to two-dimensional (2D) finite elements.
inhomogeneities, while maintaining the same volume contents and The continuum plane stress element (CPS4) is used in the ABAQUS
locations of the inhomogeneities as the ones in the first microme- FE analyses. These four representative volume elements (RVEs) are
chanics model. The alumina constituent is modeled as linear referred as FE meshes 1, 2, 3, and 4. We first determined the overall
elastic, while the aluminum constituent follows an elastic–plastic linear elastic modulus of the composites from the selected RVEs.
response. The overall response from the detailed microstructural We have also conducted convergence studies in terms of num-
characteristics is compared to the ones determined with ideal- ber of elements used in FE analyses and also the sizes of RVE,
ized microstructures. We also investigate the thermal stress effects in which we varied the RVE sizes by selecting two regions with
using the two micromechanics models discussed above. The pur- 100 ␮m × 100 ␮m. The two RVEs are referred as FE meshes 5 and 6.
pose of this study is to examine the feasibility in using idealized The elastic moduli, determined from different RVE sizes and num-
and further simplified micromechanics model for determining the ber of elements, are compared.
macroscopic mechanical properties of the composites. The reinforcement and pores in Al–Al2 O3 composites, manu-
factured using powder metallurgy method, have irregular shape
as shown in Fig. 1. Analytical and numerical studies on microme-
chanics of composites are often done on idealized microstructures,
i.e., assuming circular or spherical particles dispersed in homo-
1
http://www.nist.gov/mml/ctcms/oof/index.cfm. geneous matrix, in order to reduce complexity in obtaining the
 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56 51

Fig. 2. FE meshed model #3 of A-20 composite sample with (a) actual shape and (b) circular shape of reinforcement and pores; stress ( 11 ) contour for (c) actual shape and
(d) circular shape of reinforcement and pores due to thermal stress.

overall/effective responses of the composites. We then study the aluminum follows the deformation theory of plasticity (Hencky–
effect of idealizing the shapes of the inclusions (particles and pores) Ilyushin):
on the effective response of the composites. Using the image anal-
ysis software ImageJ, the area of the actual shaped reinforcements p Sij kk
and pores are calculated along with their centroids (see Gudlur, εij = εeij + εij + εth ; εeij = + ı ; εth = ˛ıij (T − Tref )
ij 2G 9K ij ij
2013), for each of the FE meshed models of the A-20 sample ⎧
shown in Fig. 1b. Using ABAQUS Python script, circular shaped ⎨ 3 ¯ Sij , F = 0 (2)
p
εij = 2 ¯
reinforcements and pores, with the areas same as those of their
⎩
corresponding actual shaped reinforcement and pores, are substi- 0, F<0
tuted in their places at their respective centroids. Fig. 2(a and b)
illustrates the FE mesh 3 with actual and circular inhomogeneities. 
p p p
Next, we used the micromechanics models to study the inelastic The equivalent plastic strain is ¯ = (2/3)εij εij , where εij is
deformation and degradation behaviors in the composites, incor- the
 component of the plastic strain, and the effective stress is ¯ =
porating the thermal stress effect. The alumina is modeled as linear (3/2)Sij Sij . It is assumed that the material is rate-independent
elastic while the aluminum is assumed to exhibit an elastic–plastic p
and the yield criterion is given as F = ¯ − y (εij ). The aim of this
deformation. Both aluminum and alumina are modeled as isotropic
study is to examine the effect of microstructures, i.e., porosity, local
materials undergoing small deformation gradients. The linear elas-
yielding, thermal stress, distribution of the alumina particles in the
tic constitutive model for the alumina is:
homogeneous aluminum matrix, and shapes of inhomogeneities,
on the overall linear elastic moduli2 and response of the Al–Al2 O3
Sij kk composites. It is noted that due to the plane stress assumption the
εij = εeij + εth = + ı + ˛ıij (T − Tref ) (1) out of plane strain component arises from imposing zero out of
ij 2G 9K ij
plane stress condition.

where the linearized strain is defined as εij = 1/2(ui,j + uj,i ), ui is

the scalar component of the displacement vector, Sij is the scalar
2
component of the deviatoric stress,  kk is the volumetric stress, While the macroscopic response of the composite might show a linear effec-
G and K are the shear and bulk moduli, respectively, that depend tive stress–strain relation and an elastic response under a sufficiently small overall
strain, the stress field inside the composite microstructure could vary significantly
on the elastic modulus E and Poisson’s effect , ıij is the Kro- and the stress concentrations and/or discontinuities could occur at various loca-
necker delta, T and Tref are the current and reference temperatures, tions within the microstructure of the composite. The localized stresses within the
respectively, and ˛ is the thermal expansion coefficient. The microstructure could lead to local yielding and/or fracture.
52 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56

In order to incorporate degradation in the ceramics caused by Table 1

Area fraction of alumina (%AF), porosity (%pores), elastic modulus, and Poisson’s
relatively high localized tensile stresses/strains, we consider the
ratio of the FE models generated from composite with 20% volume fraction (%VF) of
following rate degradation model for the ceramics: alumina. For comparison, results of the experiment from the nondestructive testing,
resonant ultrasound spectroscopy (RUS), are also provided in the table.
ḋ = (1 − d)f (εij ) (3)
Model %AF of alumina %Pores E (GPa) 
where f is the general functions of strain. The elastic property, i.e., FE-1 20.2 0.4 71.23a /70.82b 0.397/0/402
elastic modulus, of the ceramics changes with the degradation, FE-2 21.7 0.1 72.57/72.15 0.393/0.399
E(d). It is noted that this study does not aim at rigorously model FE-3 20.9 0.7 69.44/69.06 0.390/0.395
FE-4 21.2 0.0 75.53/75.08 0.393/0.395
the degradation in the composite constituents, instead we want FE-5 21.7 0.3 71.43 0.405
to briefly examine the overall macroscopic response of composites FE-6 19.1 0.8 70.03 0.396
in case of some localized degradation occurs in the constituents. RUS 22.6 0.5 75.15 0.310
The elastic–plastic deformation and degradation model in Eqs. (2) a
Using coarse meshes.
and (3) are implemented in user material subroutine (UMAT) of b
Using finer meshes.
ABAQUS FE.
Table 2
Effective elastic properties of composites with 20% VF of alumina, without residual
3. Numerical results (thermal) stress effect.

Model E (GPa) 
The FE simulation is carried out in two steps. The first step is pre-
scribing a uniform temperature change from 200 ◦ C, which is con- FE-1 83.28 0.313
sidered as the stress-free temperature3 based on our previous work, FE-2 86.67 0.311
FE-3 83.70 0.314
Gudlur et al. (2014), to 25 ◦ C and examining the residual (remain- FE-4 88.05 0.315
ing thermal) stresses. The thermal expansion coefficients for the Mori–Tanaka 87.00 0.330
aluminum and alumina at room temperature are 28 × 10−6 /◦ C and
4.6 × 10−6 /◦ C, respectively. The second step is simulating a uni-
axial mechanical boundary condition to determine the uniaxial The uniaxial compressive stress is simulated by prescribing a
elastic moduli and stress–strain response of the composite. The uniaxial displacement ı0 = 2.5 × 10−3 mm, which corresponds to
elastic moduli of the aluminum and alumina at room tempera- the axial strain 5%4 followed by removal of the strain.
ture are 65 GPa and 370 GPa, respectively. The Poisson’s ratios of
the aluminum and alumina are 0.35 and 0.22, respectively. The 3.1. Linear elastic properties
above thermal expansion and elastic properties of the aluminum
are obtained from Gudlur et al. (2012). The stress–strain response of The linear responses from the FE analyses are first studied and
the aluminum obtained from literature Moosbrugger (2002) is used compared to the ones obtained from the experiment from the
to model the plastic deformation of aluminum in the numerical nondestructive testing, resonant ultrasound spectroscopy (RUS).
analyses and the yield stress of aluminum is taken as 33 MPa. Table 1 presents the values of Elastic moduli and Poisson’s ratios
Consider a FE microstructure of a square region with a side from the six FE meshes considered. The first four FE meshes (FE-
length L placed in the 2D Cartesian coordinate system with the 1, FE-2, FE-3, and FE-4) have 50 ␮m × 50 ␮m square region, while
bottom left corner of the microstructure is at the origin. The FE-5 and FE-5 have 100 ␮m × 100 ␮m square region. We first per-
microstructure is subjected to the following displacement bound- formed a convergence study in terms of number of elements (mesh
ary conditions: sizes) used in the FE analyses for the first four FE meshes, as shown
in Table 1. The average size of the element in the FE analyses
u1 (0, x2 ) = u2 (x1 , L) = 0.0, u1 (L, x2 ) = C and u2 (x1 , 0) = ıo (4) with coarse elements is 3.27 × 10−4 mm × 3.27 × 10−4 mm, while
the one with finer elements is 1.96 × 10−4 mm × 1.96 × 10−4 mm. In
where u1 and u2 are the displacements in the x1 - and x2 -directions, the above FE analyses the effect of thermal stresses on the effective
respectively. The linear elastic modulus is determined by prescrib- elastic moduli and Poisson’s ratios of composites is also considered
ing a relatively small uniaxial compressive stress ¯ 22 (x1 , 0) = 0 = by prescribing a uniform temperature change from 200 ◦ C to 25 ◦ C
1 MPa and monitoring the corresponding overall (average) strain prior to determining the elastic properties. We concluded that the
ε̄22 . By prescribing a relatively small stress we can minimize effects coarse elements are capable in capturing the overall properties of
of the possible plastic deformation due to mechanical loading. The the composites while significantly reduce the computational cost.
corresponding transverse displacement along the boundary of the The rest of the FE analyses are performed using coarse elements.
microstructure is constrained to be uniform. The effective elastic It is also seen that the smaller RVE size is capable in represent-
modulus Ē and Poisson effect ¯ are defined as: ing the heterogeneities in composites. For comparison, we also
determined the overall elastic properties of composites without
ı0 ¯ 22
ε̄22 = ; Ē = considering the thermal (residual) stress effect. Table 2 lists the
L ε̄22
(5) effective elastic modulus and Poisson’s ratios from four FE meshes
C ¯ 22 and from Mori–Tanaka micromechanics model. It is seen that the
ε̄11 = ; =-
L ε̄22 elastic moduli without the residual stress effect are about 12 GPa
higher than the ones with the residual stress effect. It is also noted
that the effective elastic properties from the four FE analyses are
3
The composite processing involves curing at 600 ◦ C followed by slow cooling closed to the ones of Mori–Tanaka model with 20% alumina volume
down to room temperature (about 7 h). During the cooling process, the aluminum contents, in which an idealized microstructure of a composite with
experiences stress relaxation at elevated temperatures which reduces the pre- a solid sphere embedded in homogeneous matrix is considered.
stresses in the constituents due to the previous processing conditions. It might not
be possible to eliminate the entire pre-stress by the stress-relaxation; however
we consider the temperature at which the pre-stress is minimum as our zero-
4
stress temperature. Further continuing the cooling process to room temperature This is an engineering strain defined as the overall axial displacement divided
will eventually increase the residual-stresses. by the original length of the specimen.
 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56 53

80 with actual parƟcle shape In the microstructure with actual shapes of inhomogeneities, the
with circular parƟcles maximum tensile stress is around 300 MPa, while the maximum
ElasƟc Modulus (GPa)

75 compressive stress is around 900 MPa, which are observed in sev-

eral small regions near the sharp edges. In the microstructure with
70 idealized shapes of inhomogeneities, the maximum tensile stress
is around 65 MPa, and the maximum compressive stress is around
65 −250 MPa, which are also observed in several small regions near
the interfaces. The effective elastic moduli from the two micro-
structures are shown in Fig. 3 with slight differences in the moduli
60
between the two microstructural models. Based on this analysis,
1 2 3 4
FE meshed model # we observe that the effect of shapes of inhomogeneities within
the microstructures on the overall elastic moduli of composites is
Fig. 3. Linear elastic moduli from micromechanics with different shapes of inho- insignificant.
mogeneities.

3.2. Inelastic and degradation behaviors

From the above analyses one can observe that the existence of resid-
ual (thermal) stresses in the constituents significantly influence the The FE microstructures with actual and circular shapes of
macroscopic elastic properties of the composites. The axial stress inhomogeneities are used to simulate the uniaxial compressive
contours due to the thermal stress and small uniaxial compressive responses of the composites loaded beyond the yield limit of the
stress ¯ 22 (x1 , 0) = 0 = 1 MPa for FE-3 is illustrated in Fig. 2 left bot- materials. The effect of thermal stresses on the effective uniaxial
tom. There are several localized stresses in the aluminum, mainly compressive response of the composites is also considered by
from the thermal stresses, that are greater than the yield limit of prescribing a uniform temperature change from 200 ◦ C to 25 ◦ C
the aluminum, thus inducing local plastic deformations. These local prior to prescribing the mechanical loading. Fig. 4 depicts the
plastic deformations influence the overall elastic constant of the stress–strain response of the composites with different shapes of
composites. inhomogeneities. Once again it is seen that the shapes of inhomo-
Next, we compare the elastic moduli obtained from the FE with geneities only slightly influence the macroscopic inelastic response
actual and idealized shapes of inhomogeneities. Fig. 2(a and b) of the composites. The corresponding stress contours in the trans-
illustrates an example of composite microstructure (from FE-3) verse direction ( 11 ) at the applied maximum strain (ε̄22 = 5%) are
with actual and idealized (circular) inhomogeneities. Fig. 2(c and d) shown in Fig. 5. It is noted that under the prescribed boundary
compares the stress contours due to the thermal (residual) stress conditions, the surfaces with normal vectors n = e1 and n = −e1
and small uniaxial compressive stress. As expected, the composite directions are under traction free (¯ 11 = 0); however the hetero-
model with actual shapes of inhomogeneities experiences higher geneities in the composite body leads to nonzero transverse stress
stresses than the one with idealized (circular) inhomogeneities in the constituents. It is also seen that the sharp edges in the FE
due to the sharp edges in the actual shapes of inhomogeneities. meshes with actual shapes of inhomogeneities lead to higher stress

Fig. 4. The effect of shapes of inhomogeneities on the overall inelastic response of composites under uniaxial compression.
54 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56

Fig. 5. Stress contour in the transverse direction ( 11 ) of FE meshed model #3 at maximum applied strain (ε22 = 5%). Left: actual shapes of inhomogeneities and right: circular
shapes of inhomogeneities.

Fig. 6. Uniaxial tensile response of composites with inelastic and degradation behaviors for (a) FE-1 and (b) FE-3.

fields due to the effect of stress concentration. In this study, the dis- stress and degradation fields in FE meshed model #1. As seen in
tribution of stress fields in the composite models with actual and the figures, the composite with actual shapes of inhomogeneities
circular microstructures only slightly influences the macroscopic shows stress and degradation fields with higher magnitude as com-
inelastic response of the composites. pared to the composite with circular shapes of inhomogeneities,
We now consider the possibility of degradation in the brittle especially in the brittle inclusions. The degrading elastic properties
inclusions due to existence of tensile stresses. For this purpose, of the inclusions do not have significant effect in this composite sys-
we define the degradation parameter, according to Eq. (3) and tem, as the response is dominated by the ductile matrix behaviors.
the elastic modulus of the alumina decrease with the degradation This might also be because of the relatively low content of the brittle
E = E0 (1 − d), where E0 is the elastic modulus of alumina in absence alumina (20%). It is noted that sharp edges of the inhomogeneities
of degradation. It is assumed that degradation occurs when the in the microstructures with actual shapes of inhomogeneities lead
normal stresses in the alumina are under tension. The following to high localized stresses (see contours in Figs. 2, 5 and 7). In real-
function is considered to numerically study the effect of degrada- ity, high localized tensile stresses would cause cracks to initiate and
tion: propagate in the composite bodies. When the effect of cracking in
 the composite on the overall macroscopic mechanical performance
f (εij ) = ˛ ε211 + ε222 + ε233 ; ε11 > 0, or ε22 > 0, or ε33 > 0 of the composites is considered, perhaps it is necessary to consider
composites with actual shapes of inhomogeneities (see Williams
(6) et al., 2012).

where ˛ is a constant. We study response of the composite under

uniaxial tensile deformation, incorporating the thermal stress 4. Conclusions
effect, in order to examine the tensile degradation in the alumina.
Fig. 6 shows the effective (macroscopic) stress–strain responses We have examined the effects of microstructural character-
from FE meshed models #1 and #3 with actual and circular shapes istics, of the constituents, and shapes and distributions of the
of inhomogeneities. The cooling down process results in shrinking pores and inclusions (inhomogeneities), and thermal stresses on
of the composites, as shown by the negative initial strains. Again, for the overall (macroscopic) response i.e., properties and mechani-
the composite system considered in this study, the microstructural cal behaviors (elastic, inelastic, and degradation) of the Al–Al2 O3
morphologies do not have a significant role in the overall macro- composites with 20% Al2 O3 volume contents. The studies were
scopic mechanical response of the composites. Fig. 7 illustrates the performed by considering linear elastic response of the brittle
 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56 55

Fig. 7. Transverse stress and degradation contours from FE meshed model #1 with actual (top) and idealized (bottom) inhomogeneities.

alumina and inelastic response of the aluminum matrix. Further- materials. The effect of degradation in the brittle alumina under
more, the effect of the degradation in the brittle alumina due tension was also investigated. Once again, it was observed that the
to tensile stresses/strains, which reduces the elastic modulus of composites with actual shapes of inhomogeneities experience
the alumina, on the overall stress–strain response of the com- higher magnitude of stress and degradation fields as compared to
posite was examined. Two composite microstructures with actual the ones with circular shapes of inhomogeneities. However, the
and idealized shapes of the inhomogeneities were considered. The macroscopic stress–strain responses from the two composites are
stress fields in the two microstructures due to prescribed thermal very close, which shows the insignificant overall effect of the shapes
and mechanical stimuli were compared and their effects on the of inhomogeneities on the macroscopic mechanical responses. The
overall macroscopic mechanical response of the composites were reason might be due to a relatively low volume content of inhomo-
examined. Micromechanics models with idealized and simplified geneities (around 20% volume contents of the composites), in which
shapes of inhomogeneities have been widely used to obtain the the macroscopic response is dominated by the matrix behavior. It is
average (macroscopic) response of composites due to their rela- also noted that this study does not consider fracture in the compos-
tively low costs in providing analytical and numerical solutions. ites due to severe stresses localized in the constituents that might
The main purpose of this study is to examine whether the com- lead to crack initiation and propagation.
posites with irregular shapes of inhomogeneities, such as in the
Al–Al2 O3 composites, can be approximated by considering ideal-
Acknowledgements
ized and simplified shapes of inhomogeneities in determining their
overall macroscopic mechanical responses.
This research is sponsored by the National Science Foundation,
We found that the linear elastic modulus of the Al–Al2 O3 com-
under the grant CMMI-1030836 and Air Force Office of Scientific
posite, ignoring the effect of thermal stresses, can be well predicted
Research AFOSR FA 9550-10-1-0002.
by the Mori–Tanaka model with a solid spherical alumina inclusion.
We also found that the linear elastic properties of composites with
actual and circular inhomogeneities are very close. This shows the References
insensitivity of the irregular shapes of the inhomogenieties on the
Aboudi, J., 1985. Inelastic behavior of metal–matrix composites at elevated temper-
effective linear elastic modulus of the studied composites. How- ature. International Journal of Plasticity 1 (4), 359–372.
ever, the effect of thermal stresses, due to high temperature pro- Cannillo, V., Leonelli, C., Boccaccini, A.R., 2002. Numerical models for thermal resid-
cessing, on the macroscopic elastic properties is quite significant. ual stresses in Al2 O3 platelets/borosilicate glass matrix composites. Materials
Science and Engineering A 323 (1–2), 246–250.
This is because there are some locations in the composite micro-
Chawla, N., Ganesh, V., Wunsch, B., 2004. Three-dimensional (3D) microstructure
structures that experience high thermal stresses and cause yielding visualization and finite element modeling of the mechanical behavior of SiC
in the aluminum matrix. The composites with actual shapes of particle reinforced aluminum composites. Scripta Materialia 51 (2), 161–165.
Dvorak, G.J., Srinivas, M.V., 1999. New estimates of overall properties of heteroge-
inhomogeneities experiences more severe stresses within the con-
neous solids. Journal of the Mechanics and Physics of Solids 47 (4), 899–920.
stituents as compared to the ones with circular shapes of inhomo- Eischen, J., Torquato, S., 1993. Determining elastic behavior of composites by the
geneities. High stresses localized in the sharp edges are noticeable, boundary element method. Journal of Applied Physics 74 (1), 159–170.
which might cause fracture in the composites. The present study, Gudlur, P., Forness, A., Lentz, J., Radovic, M., Muliana, A., 2012. Thermal and mechan-
ical properties of Al/Al2 O3 composites at elevated temperatures. Materials
however, does not consider the possible fracture in the composites. Science and Engineering A 531, 18–27.
In absence of fracture, the actual and circular shapes of inhomo- Gudlur, P., Muliana, A., Radovic, M., 2014. Thermo-mechanical properties of
geneities gave close predictions of the linear elastic properties. aluminum-alumina composites based on its microstructural characteristics.
Composite Part B 58C, 534–543.
Next, we studied the uniaxial response of the composites Gudlur, P., (PhD Dissertation) 2013. Experimental and Numerical Studies of
under compression and tension beyond the yield limit of the Aluminum–Alumina Composites. Texas A&M University.
56 P. Gudlur et al. / Mechanics Research Communications 57 (2014) 49–56

Khan, K.A., Barello, R., Muliana, A.H., Lévesque, M., 2011. Coupled heat conduction Williams, J.J., Segurado, J., Llorca, J., Chawla, N., 2012. Three dimensional (3D)
and thermal stress analyses in particulate composites. Mechanics of Materials microstructure-based modeling of interfacial decohesion in particle reinforced
43 (10), 608–625. metal matrix composites. Materials Science and Engineering A 557, 113–118.
Langer, S., García, R.E., Reid, A., 2007. The OOF2 Manual: Revision 3.2 for OOF2 Version Qin, S., Chen, C., Zhang, G., Wang, W., Wang, Z., 1999. The effect of particle shape on
2.0 beta 8. National Institute of Standards and Technology (NIST). ductility of SiCp reinforced 6061 Al matrix composites. Materials Science and
Moosbrugger, C., 2002. Atlas of Stress–Strain Curves, 2nd ed. ASM International. Engineering A 272, 363–370.
Rasool, A., Böhm, H.J., 2012. Effects of particle shape on the macroscopic and micro- Zimmermann, A., Fuller Jr., E.R., Rödel, J., 1999. Residual stress distributions in ceram-
scopic linear behaviors of particle reinforced composites. International Journal ics. Journal of the American Ceramic Society 82 (11), 3155–3160.
of Engineering Science 58, 21–34.