Computers and Structures 88 (2010) 25–35

Contents lists available at ScienceDirect

Computers and Structures

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

Aluminium foams structural modelling

M. De Giorgi *, A. Carofalo, V. Dattoma, R. Nobile, F. Palano
Dipartimento di Ingegneria dell’Innovazione, Università del Salento, Via per Arnesano, 73100 Lecce, Italy

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

Article history: The aim of this work is the development of microstructural numerical models of metallic foams. In par-
Received 24 February 2009 ticular, attention is focused on closed cell foam made of aluminium alloy. By means of a finite elements
Accepted 3 June 2009 code, the material cellular structure was shaped in different ways: firstly the Kelvin cell, with both plane
Available online 30 June 2009
and curved walls; finally an ellipsoidal cell defined by random dimensions, position and orientation has
been adopted as base unit. In order to validate the foam numerical models, static tests were performed to
Keywords: obtain the typical stress–strain curves and then compared with the numerical analysis results.
Aluminium foams
Ó 2009 Elsevier Ltd. All rights reserved.
Mechanical characterisation
FEM
Microstructural model

1. Introduction resting on the whole structure weight. Furthermore, electromag-

netic shielding device transparent to light is sometimes required.
Metallic foams are a new class of materials that turn out to be In these applications, metallic foams seem to have interesting util-
promising in different engineering fields [1–7]. They show a cellu- isation perspective, because their mechanical properties guarantee
lar structure consisting in a solid and a gaseous phase and their the structural auto-supporting.
classification is based on the cells morphology. A certain number The absence of well-established procedure for the structural ver-
of manufacturing processes are currently used and other innova- ification of innovative materials represents a relevant obstacle to
tive methods are being developed in order to obtain a more reliable their effective use, even if they could assure the best performances.
and uniform production. Technological process determines also if The use of components made of metallic foams or their use as a fil-
the morphology of the component is made of open or closed cells. ler for hollow structures, in fact, requires a deeper knowledge of
Metallic foams show an interesting combination of physical and structural behaviour. Designer needs not only experimental data,
mechanical properties that make them particularly versatile: the but also reliable and relatively simple analytical or numerical meth-
low apparent density, for example, allows obtaining a high stiff- ods for calculation of applied stress and for prediction of failure of
ness/specific weight ratio, the presence of cavities and the essential metallic foam components. In perspective, these design tools must
in-homogeneity provide them acoustic and thermal insulation guarantee a reliability level that would be comparable to the ones
properties, besides the possibility to absorb impact loads and to used for traditional materials. Intrinsic inhomogeneities of metallic
damp vibrations. Finally, the metallic structure potentially gives foam represent a serious problem for calculation needed in design
those good electromagnetic shielding properties and inalterability phases. In particular, it would be desirable to establish a link be-
in time. tween microstructural and macroscopic behaviour of this kind of
The combination of these physical properties makes metallic materials. It is beyond doubt that microscopic geometry and prop-
foams able to be competitive in terms of performances and costs erties affect macroscopic structural behaviour, which is essential
in several unusual advanced applications. For example, the realisa- for the design of components, but it is very difficult to derive mac-
tion of structures able to shield efficaciously the electromagnetic roscopic properties from microstructure.
fields is very interesting. Within this context, one of the most fre- The problem of deriving macroscopic from microscopic proper-
quently used solutions consists in holding the volume to be ties is very complex and probably its resolution will lead to partial
shielded in a metallic surface. In some situations, it is required that solutions that will be applicable in well-defined and restricted con-
the capability to support relevant load levels without excessively text. Nevertheless, this step will represent an important advancing
in this research field.
This paper is inserted in this general perspective, since the main
* Corresponding author.
E-mail addresses: [email protected] (M. De Giorgi), [email protected]
aim is to try to evaluate macroscopic mechanical behaviour start-
(A. Carofalo), [email protected] (V. Dattoma), [email protected] (R. Nobile), ing from a micro-structural approach for particular metallic foam.
[email protected] (F. Palano). The purpose is using different numerical methodologies to model a

0045-7949/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruc.2009.06.005
26 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35

specific class of metallic foams in order to evaluate advantages, In this work the first approach has been followed. The use of
limitations and simplicity of use. The metallic foam class that elements like beam or shell, in fact, presupposes an idea of foam
was selected for this study is the closed cell aluminium foam, since structural behaviour at an intermediate scale, being in authors
its high relevance in industrial context. opinion the better compromise nowadays possible to study foam
In the literature, there are some studies regarding the way to structural properties both at microscopic and macroscopic scale.
foresee the foams macroscopic behaviour by cell geometric data. The problems to solve in the micro-structural model definition
The references [8–14] represent some significant examples of of closed cell metallic foams can be resumed as follows:
structures formed by closed cells.
Hanssen et al. [8] present results from an experimental pro- – choice of a cell morphology to be employed as a repetitive unit
gramme for the general validation of constitutive models for alu- for the model construction;
minium foam. The obtained database is applied in order to – definition of geometric dimensions, which characterise the cell
investigate how different existing constitutive models for alumin- form;
ium foam cope with quasi-static and low-velocity dynamic load- – selection of cells assembly criteria;
ing. Compared with the experiments in the validation – metallic foam block numerical model realisation through the
programme, none of the models managed to represent all load con- repetition of several cells in the three directions of the space.
figurations with convincing accuracy. One reason for this is that
fracture of the foam is a very likely local failure mechanism, which In this paper, two different cell morphologies were considered
was not taken into account in any of the models considered. and numerical results were compared to experimental data. In par-
Belingardi et al. in [9] show the results of some finite elements ticular, the model adequacy was evaluated by comparison of the
simulations of a set of experimental compression tests, both static predicted and actual compressive elastic modulus.
and dynamic, in order to verify the quality of the material models.
For the load phase, they found that the numerical results agreed 2. Survey of proposed models
with the experimental ones, both in the static and dynamic cases
for which the strain rate effects are taken into account. Moreover, The first model considered was referred to the ideal Kelvin cells
the authors develop a new material model for the unload phase. (tetrakaidecahedron) with planar walls. This cell morphology has
Roberts and Garboczi [10] have computed the density and been already used by other authors, since it corresponds to the
microstructure dependence of the Young’s modulus and Poisson’s minimum surface energy for a fixed volume [1,15].
ratio for several different isotropic random models based on Voro- Models based on Kelvin cells were various: the most simple
noi tessellations and level-cut Gaussian random fields. The results, considered regular Kelvin cell having planar faces, while more
which are best described by a power law E / qn (1 < n < 2), show complex models were obtained introducing the cells walls curva-
the influence of randomness and isotropy on the properties of ture and thickness variation. These modifications could improve
closed cell cellular materials. the mechanical behaviour approximation. Further refinements of
In [11], Lu et al. analyse the deformation behaviour of two dif- this approach could give a better approximation, but limitation
ferent types of aluminium alloy foam (with closed and semi-open due to the regularity of the resulting cellular structure was not
cells) under tension, compression, shear and hydrostatic pressure. removable and constituted an important starting error. According
The influence of relative foam density, cell structure and cell orien- to other authors [10,16–18] the structure regularity could be con-
tation on the stiffness and strength of foams is studied; the mea- sidered as the origin of the higher model stiffness with respect to
sured dependence of stiffness and strength upon relative foam the real foam behaviour.
density is compared with analytical predictions. Loading paths For this reason a different approach was used to develop a sec-
are compared with predictions from a phenomenological constitu- ond model that considered the real cellular structure irregularity.
tive model. It is found that the deformations of both types of foams In order to reproduce the pores feature and to consider its random
are dominated by cell wall bending, attributed to various process variability, an elementary ellipsoidal feature was chosen as unitary
induced imperfections in the cellular structure. The closed cell cell. The assembly of a fixed number of ellipsoidal cells, having
foam is found to be isotropic, whereas the semi-open cell foam dimensions, position and orientation varying parametrically in a
shows strong anisotropy. random way, could constitute a better approximation of the foam
Another possibility is to use X-ray tomography to produce a real structure.
data file containing all the geometrical information on a relatively
large foam specimen. In this case the model is realised through
brick elements associated to each voxel identified by X-ray scan 3. Foam mechanical characterisation
[12,13]. Numerical results show a large influence of scan spatial
resolution, which influence directly computational time. Finally, Experimental data used for comparison between numerical
an interesting approach is presented in [14], where a spectral anal- models were obtained through compressive tests. These data rep-
ysis carried out on X-ray tomography of open cell foams allows to resented the necessary experimental term of comparison used un-
derive statistical parameters used to build solid finite element til now for the numerical model improvement.
models. The uniaxial compression tests represented the easiest and
Although the variety of approaches that haves been briefly pre- most efficacious way to determine the metallic foams mechanical
sented, all the model could be classified into two different catego- behaviour under static conditions. The resulting data, particularly
ries. The first one tries to identify a simple geometrical feature that the applied load versus the overall strain trend, defined the macro-
can constitute the repetitive cell used to built a foam model [8–11]. scopic behaviour of the foam cellular structure. Then, re com-
The unit cell, which is not necessarily close to the real foam geom- pressive curve was the term of comparison used to evaluate the
etry, is chosen on the basis of energy or morphology consideration studied numerical model efficiency and reliability.
and leads to models generally based on shell elements. On the con- The analysed foam was made of AlSi10Mg aluminium alloy and
trary, the efforts of the other works are concentrated to obtain an produced by Alulight Company. Table 1 resumes the general prop-
exact reproduction of foam geometry by means of experimental erties of this aluminium foam. The material was available in forms
techniques [12–14]. In this case, solid models are generally of 10-mm-thick sheets that were cut to obtain the specimens. It
obtained. was observed that sheets showed a central region with a foam
 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35 27

Table 1 – finally a third region in which the material showed a stiffness

Mechanical properties of AlSi10Mg base alloy. increment due to the remarkable material densification and to
Aluminium Solidus density Elastic modulus Yield strength Poisson the strain hardening phenomena.
alloy qS (kg/m3) ES (GPa) ry (MPa) modulus
AlSi10Mg 2860 69 250 0.33 It is also important to notice that aluminium foam showed a dif-
ferent behaviour between the first load application and the subse-
quent ones: in fact, the material seemed to undergo an initial
density quite uniform, while external zones were characterised by arrangement, which gave rise to a different elastic modulus when
higher values of density. It was probably due to the different initial comparing the first and the following loading cycles. This fact is
distribution of the gas particles in the matrix during manufacturing generally explained invoking an initial rearrangement and local-
process. Therefore, specimens having different density were easily ised plastic strain of cell walls that is realised in the first compres-
obtained from the same sheet. Mechanical experiments were sion phases and responsible of the subsequent drastic Young’s
planned in the following way: modulus increment. The Young modulus measured in the load cy-
cles subsequent to the first one is assumed to be the most impor-
 initial tests were carried out to establish the influence of the tant constitutive parameter of the metallic foam.
specimen dimension on mechanical properties: 40  40 mm As a consequence, compressive test were executed into two
(D40), 70  70 mm (D70), 100  100 mm (D100) specimens steps:
having a similar density were considered;
 extensive experimental tests were then carried out on D40 spec- – a first loading step up to about 75% of plateau stress: this step
imens having different density, with the aim to highlight the leaded to the determination of the first cycle Young modulus
influence of apparent density on the stress–strain curve; these (Eload), followed by an unloading phase that allowed to deter-
compressive tests were executed into two steps, as described mine a characteristic Young modulus (Eunload);
in the following, in order to evaluate constitutive behaviour of – a second loading step up to foam densification.
aluminium foam;
Fig. 1b represents the trends obtained by testing three samples
Tests were carried out on a servo-hydraulic MTS 810 machine having different densities and allows drawing attention to the den-
having 100 kN load capacity equipped with two planar plates, sity influence on the foam mechanical behaviour. The curves
according to the indication reported in [1]. Since the crosshead showed that, when the specimen density decreased, also the elastic
speed was fixed at 0.005 mm/s and foam thickness was 10 mm, modulus, the compressive strength and the curve slope reduced,
the initial compressive deformation rate of the test was while the strain corresponding to the densification beginning
0.5  103 mm/mm s. Strain was calculated dividing the cross- raised.
head displacement by the specimen thickness. Fig. 2 reports the linear interpolation of experimental data of
As a first result, the specimen dimensions seemed to be not rel- the loading and unloading phases that was used for the evaluation
evant with regard to mechanical foam properties, as it was evident of the compressive Young’s modulus Eload and Eunload, respectively.
by comparing the plots reported in Fig. 1a and obtained on speci- Data reported in Table 2 allowed to evaluate the difference existing
mens D40, D70 and D100 having different dimensions and similar in the load and unload phases. The stress–strain curve reported in
densities. On the basis of this observation, further tests were car- Fig. 2 could be taken as reference in order to describe the general
ried out on specimen D40. behaviour of the examined foam and it was achieved imposing
Stress–strain curves deduced from the experimental tests were firstly two subsequent loading cycles and then loading the sample
similar to those reported in the literature [1–5,17,19] and showed until its compacting.
the existence of three different and characteristic zones: Testing some specimens with only one loading cycle, we noted
that the compacting phase trend was not influenced by the pres-
– a first part, corresponding to little deformations, in which the ence of a possible unloading step. The reference loading curve em-
material presented an elastic behaviour and the cells walls chan- ployed for the numerical results verification was then obtained
ged their curvature; with a first loading phase until 75% of plateau stress and a follow-
– a second very wide zone, characterised by a plateau with a ing unloading step. In fact, such a curve was repeatable and inde-
rather constant stress, phase in which the cells underwent phe- pendent from initial arrangement phenomenon, which is
nomena of local instability, yielding and rupture; characteristic of cellular materials.

Fig. 1. Uniaxial compression tests: (a) effect of specimen size; (b) effect of specimen density.
28 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35

Fig. 2. Loading and unloading behaviour for the determination of elastic modulus.

Table 2
Experimental elastic modulus in loading and unloading phases.

Specimen Specimen Maximum stress Density Eload Eunload

number type smax (MPa) (kg/m3) (MPa) (MPa)
1 D70 3.9 423.2 109 Loading only Fig. 3. Tetrakaidecahedral geometry used in the Kelvin cell model.
2 D100 4.7 473.7 78
3 D100 4.9 445.3 111
4 D40 4.1 433.1 103 [18]. In this work, the authors affirm also that the cell size is inde-
5 D40 12.5 771.2 243
pendent from direction. The used foam properties are summarised
6 D40 5.0 432.5 146 577
7 D40 7.8 546.8 229 436 in Table 3.
8 D40 4.9 432.9 155 433 A sample load of 1 MPa was used to perform a mesh sensitivity
9 D40 4.4 435.6 152 481 analysis, considering that the model was used to determine macro-
10 D40 9.6 568.2 256 453
scopic properties of the foam. In particular the evaluation of the
11 D70 4.8 412.3 97 215
12 D40 4.2 555.6 222 493 upper surface cell displacement was directly used to calculate
13 D70 5.1 418.1 181 522 foam Young modulus. For this reason, the Von Mises stress in the
central node of the upper face and the mean displacement of the
central upper face were considered in the sensitivity analysis.On
4. Numerical analysis the basis of the results resumed in Fig. 4, authors decided to divide
each cell edge in four parts. In this manner, the unit cell consisted
4.1. Kelvin cell model in 352 elements. The global model was obtained replicating the
unit cell in all the space directions. In order to optimise the solu-
A simple microstructural model reproducing a cellular material tion efficiency, the model presented nine cells on the basis and four
can be based on the regular space repetition of a unit cell with along the thickness. The uninfluence of the cells number on the
proper geometry and dimensions. The tetrakaidecahedron or Kel- macroscopic mechanical properties of the cellular structure was
vin cell is the most utilised in literature as unit cell to model the preliminarily verified through a model having a number of 16 cells
foam material [1–2,15], since it is the lowest surface energy unit on the basis.
cell known consisting of a single polyhedron. It is defined by six Boundary conditions were applied on the lower face, imposing a
planar square faces and eight hexagonal faces that could be planar, zero value to z-direction displacement and rigid body boundaries;
with a mean curvature equal to zero (Fig. 3), or convex, with a po- pressure load was applied to nodes of the square faces on the
sitive mean curvature. upper surface and it was set to 4 MPa.
In order to simulate the actual foam structure, FEM 3D model Fig. 5 shows the node displacements in the load direction and a
based on the Kelvin unit cell was refined in the course of the work limited edge effect due to absence of adjacent cells can be ob-
introducing the following improvements: served. However, the displacement of the middle node of the cen-
tral face is considered as representative of the global behaviour of
1. planar faces model the foam. Foam macroscopic strain, calculated as displacement/
2. non-zero faces curvature model thickness ratio, was recorded at each load increment to dis-
3. thickness faces variability play the stress–strain curve.
Young’s modulus value of this model resulted unacceptably
FEM analysis was performed using the Ansys software adopting higher than experimental one. This discrepancy was due to the
a discretisation in shell element named Shell93 having eight nodes excessive stiffness associated with the idealised structure, which
and a quadratic formulation. The study was performed in the large resulted unrealistically ordered. In fact, the real structure presents
displacements range. The material model was isotropic with elastic imperfections originated by the production process and that are
perfectly plastic stress–strain behaviour. The material properties, typical of the foam.
reported in Table 1, were the actual mechanical properties mea- The case study based on Kelvin cell was refined trying to reduce
sured directly on Alulight foam by micro-hardness characterisation its inadequacy: the face curvature of the unit cell was then intro-
[17]. duced in the model. The face curvature can be calculated from
The unit cell apparent density was assumed equal to the actual measurement of the chord length L and the triangular area A indi-
density of the foam, of about 500 kg/m3. On the contrary, the actual cated in Fig. 6 [17]. Considering the angle h between the normals at
cell size was assumed based on a statistical analysis reported in the end of the cell wall, it can obtain the relationship:
 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35 29

Table 3
Geometrical properties and results of Kelvin cell models.

Average cell Average Edge length Density Wall thickness Edge thickness Edge volume Young’s modulus
dimension (mm) curvature L/2R L (mm) (kg/m3) tf (mm) te (mm) fraction U (MPa)
Alulight (experimental) 418.1–555.6 – – – 433–577
Planar faces model 500 0.1360 0.1360 – 3287
Curved faces model 2.6 0.37 0.9192 500 0.1178 0.1178 – 1944
432.5 0.113 0.113 – 1265
Curved faces with variable 432.5 0.0973 0.2350 50 2340
thickness model 0.0876 0.2465 55 2198
0.0778 0.2575 60 1947
0.0681 0.2680 65 1714

could be explained considering that the curved wall was less stiff
than the planar face; moreover, cell faces were characterised by a
lower thickness that involved a further decrease of the stiffness.
In the more realistic model with curved walls, the effect of the
density on the Young’s modulus has been analysed. The compari-
son of models with planar faces and curved faces having the same
density showed that the stiffness of the two models were compa-
rable if the density was higher than 560 kg/m3. On the contrary, if
the density approached 500 kg/m3 or less the difference of the two
models became very significant. For example, in the interest den-
sity range of 432.5–500 kg/m3, because of the introduction of the
walls curvature curved face model stiffness decreased at about
60% of the planar cell model.
The last refinement of the model consisted in the introduction
of the wall thickness variability. In the real structure, in fact, during
the foaming process the liquid foam drains toward the cell edges
causing a thickening of these. According to [2], the solid fraction
at the edge results:

t 2e
/¼ ð4Þ
te þ Znf t f l
where n  is the average number of the edges of each face, Zf the num-
ber of faces converging in each edge, te and tf the edge and the face
thickness, respectively, and l the edge length.
The relative foam density is:
 
q f n t2e 1 tf
¼ þ ð5Þ
qS C 4 2Z f l2 2 l
where C4 is the constant related to the cell volume and f the number
of faces per cell. For the most foam, Zf = 3, n  5, f  14 and C4  10,
the face and edge thickness related to the chosen solid fraction
Fig. 4. Mesh sensitivity analysis of Kelvin cell model. could be obtained with good accuracy [2]:
 
  q t2 tf
4A ¼ 1:2 2e þ 0:7 ð6Þ
h ¼ 4 tan1 ð1Þ qS l l
L2 tf q
¼ 1:4ð1  /Þ ð7Þ
and the normalised cell curvature, L/2R, results [17]: l qS
    12
L h te 1 q
¼ sin ð2Þ ¼ 0:93ð/Þ2 ð8Þ
2R 2 l qS
In order to keep constant the density value, the face thickness was In order to assign a different thickness value to edge and face, the
recalculated referring to the following equation: model was divided in different areas. Fig. 7 shows the mesh per-
formed on the model with thickening edges. Cell dimensions were
AC  t C ¼ A  t ð3Þ
the same of the previous model but the mesh resulted denser and
where AC and tC were, respectively, the area and the thickness of the consisting in 792 elements per cell. The U effect was evaluated sim-
curved wall, A and t were, respectively, the area and the thickness of ulating the compressive test using different U values and an applied
the planar wall. Table 3 reports the geometric parameters of the load of 4 MPa.
unit cell with curved walls. Results are reported in Fig. 8, where Young’s modulus was dia-
The curved walls model presented a less evident edge effect re- grammed versus solid fraction parameter U. Elastic modulus was
spect to the planar walls one. Analysing the data reported in Table generally higher respect to the model with the curved face only.
3, it could be observed that the new model showed stiffness 40% Probably, the greater tendency to instability of the cell walls, due
lower than the initial model with the same density. This behaviour to thickness face decrease, was compensate by the stiffening
30 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35

Fig. 5. Map of the z-displacement for the Kelvin cell planar faces model.

Fig. 8. Dependence of elastic modulus against solid fraction U.

Fig. 6. Parameters of a curved cell wall.

originated by solid concentration at the edge, at least for the U val-

ues considered in this analysis. Therefore, considering the solid
concentration, it resulted a useless complication.

4.2. Ellipsoidal cell model

The Kelvin cell model, discussed in the previous section, re-

sulted inadequate to describe the foam structure. In particular,
Young’s modulus of the foam was overestimated and the typical
plateau region of the foam compression curve was not reproduced
by Kelvin cell model. This difference was originated by the regular
disposition of Kelvin cell into the model, which was far from the
real aspect of the metallic foam. For this reason, the efforts were
focalised to obtain a model that would be able to conjugate the
simplicity of a geometrical elementary cell and the possibility to
build a random assembly. In order to define a schematisation with
highly random cell morphology, authors’ attention focused on a
simple ellipsoidal unit cell, having the same length for two of the
three axes. It was possible and very easy to build a parametric
model of this elementary cell (Fig. 9) by the definition of the fol-
Fig. 7. Mesh of Kelvin cell model with variable thickness cell face. lowing parameters:
 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35 31

Length of major axis r was related, but not equal, to the half of
the average cell dimension. Variability range of r was assumed to
be 2–3 mm. In this manner the single ellipsoidal cell has dimen-
sions that were twice with the respect to the mean pore dimension
of the foam. When ellipsoidal cell were assembled to build the
model, interceptions that occurred between cells produced pores.
It means that dimension would be near the half of ellipsoidal cell
dimension.
Ratio of major/minor semi-axis t was chosen in the range 1–2.
Lower limit of t = 1 corresponded to a spherical shape. For the
choice of upper limit, some trials suggested that values of ratio
higher than 2 originated some troubles in the discretisation of
ellipsoid cells in elements.
Finally, orientation angles h, u of major semi-axis varied in the
range 0–180° in order to obtain all the possible orientation in the
space.
The complete definition of the model required also the choice of
two other parameters that were not completely independent.
These parameters were the number of cells used to build the model
and the thickness associated to cell walls that was considered con-
stant for all the cells. The number of cells had to be sufficient to
regularly fill the model and not excessive in order to obtain pores
having dimension similar to foam morphology. On the other hand,
the number of cells determined also the amount of areas that con-
stitutes the model. Since the volume of material involved was pro-
Fig. 9. Geometrical parameters of ellipsoidal cell unit.
portional to the foam density, an increase of cell number requires a
decrease of the cell wall thickness. Also in this case, several trails
were considered and the best solution for the reference cubic vol-
 xc, yc, zc: cartesian coordinates of the ellipsoid centre; ume 10 mm side was obtained considering a number of 35 cells
 r: length of the major semi-axis; and a corresponding thickness of 0.11 mm, which was not far from
 t: ratio of major/minor semi-axis; reality.
 h, u: orientation angles of major semi-axis respect to a global Finally, since the foam was delimited by two solid surfaces
coordinates system; (Fig. 10), two planes corresponding to the upper and lower surfaces
of the cubic volume of the foam were added.
By this way, the dimension and the position of the elementary On the basis of the previous observations, the ellipsoidal cell
cell could be easily defined using an algorithm for the generation model has been built following this operative steps:
of set of random real number. The choose of a probability density
function for the random parameter generation is the object of a – definition of a parameter set assigning to each factor a random
large number of papers in the literature. Readers can refer to fun- value in their range variability: a random uniform distribution,
damental work of Kaminski [20] for an overview about probabilis- provided by a Matlab routine, was used for this purpose; the
tic approach in computational mechanics. For the purpose of this parameter set generated are reported in Table 4, with the indi-
study, an algorithm based on a random uniform distribution into cations of the material volume and the foam density corre-
the parameter range variability was considered adequate. Uniform sponding to the amount of all the ellipsoidal cells;
random distribution is characterised by a mean value and a vari- – modelling of a single ellipsoid corresponding to each parameter
ance equal respectively to: set: a CAD parametric software, in particular PRO-EÒ WildFire
aþb 2.0 modeller, used the input data to generate the ellipsoid sur-
mean value ¼ ð9Þ face feature in a neutral format that is easily recognised by sev-
2
eral FEM software;
ðb  aÞ2
variance ¼ ð10Þ
12
where a and b are the interval minimum and maximum values.
It is important to observe that the random generation of dimen-
sion, position and orientation of the different cells produces in
most cases the co-penetration of the cells. In this manner, pores
of various shape, which will exalts the irregularity of the model,
are obtained.
Range variability parameter could be defined on the basis of
simple considerations.
Cartesian coordinates of the ellipsoid centre were a direct con-
sequence of the dimension chosen for the numerical model: in or-
der to reduce number of nodes and elements of the model to an
acceptable dimension, the choice was to model a metallic foam re-
gion constituted by a 10 mm side cube. Therefore, variability range
of xc, yc, zc was fixed to 1 to 11 mm in order to consider also ellip-
soid having centre beyond the cubic region but a consistent surface
interior to model. Fig. 10. Aluminium foam specimen.
32 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35

Table 4
Geometrical parameters used for the generation of ellipsoidal cell model.

Thickness (mm) Parameters

0.11 xc yc zc r t h / Volume (mm3)
Cell number [1;11] [1;11] [1;11] [2;3] [1;2] [0;180] [0;180]
1 0.11 5.09 1.78 2.86 1.59 14.56 81.31 4.303
2 0.27 6.76 5.48 2.38 1.21 139.9 109.77 5.105
3 0.55 9.15 6.43 2.59 1.3 162.92 10.69 5.257
4 0.56 5.78 0.98 2.6 1.47 96.08 56.85 4.144
5 0.74 1.84 9.29 2.39 1.23 19.65 139.09 4.982
6 1.62 9.41 9.87 2.67 1.84 148.65 125.36 2.792
7 1.79 7.28 4.03 2 1.19 60.86 22.56 3.694
8 1.89 2.8 4.72 2.7 1.23 52.92 23.43 6.393
9 2.3 1.49 4.81 2.58 1.17 134.34 16.62 6.439
10 2.36 4.04 3.2 2.75 1.23 1.86 1.41 6.637
11 2.93 3.84 4.11 2.35 1.44 120.22 76.16 3.512
12 3.19 9.62 7.38 2.24 1.31 108.62 118 3.846
13 3.69 2.88 1.44 2.02 1.92 124.66 130.13 1.448
14 3.78 9.99 2.37 2.73 1.43 127.31 95.62 4.838
15 4.18 0.78 4.76 2.33 1.18 140.43 19.59 5.139
16 4.19 9.62 3.91 2.4 1.9 100.2 113.72 2.106
17 4.24 9.99 6.73 2.89 1.98 131.35 22.77 2.834
18 4.34 1.42 8.6 2.68 1.44 133.43 24.17 4.594
19 4.73 6.56 2.06 2.4 1.11 140.65 17.75 6.171
20 4.76 0.27 4.07 2.7 1.26 8.72 25.56 6.092
21 4.98 4.55 5.61 2.59 1.41 11.09 30.29 4.469
22 5.03 1.86 8.31 2 1.59 132.18 35.32 2.069
23 5.14 8.8 9.03 2.99 1.26 94.7 57.15 7.501
24 5.16 4.15 2.54 2.49 1.6 109.42 56.96 3.202
25 5.65 4.91 6.61 2.47 1.71 123.07 39.16 2.758
26 5.66 3.77 3.28 2.66 1.22 87.34 45.19 6.303
27 5.68 2.83 7.65 2.64 1.12 51.84 160.73 7.365
28 5.78 1.49 7.83 2.96 1.3 143.81 126.58 6.903
29 5.83 8.17 5.04 2.54 1.32 98.92 100.03 4.900
30 6.16 7.75 8.66 2.37 1.42 71.37 33.2 3.675
31 6.4 1.19 6.19 2.83 1.51 132.29 38.17 4.669
32 6.44 1.81 1.11 2.24 1.09 15.93 13.92 5.556
33 6.5 4.9 1.25 2.73 1.26 18.87 164.48 6.231
34 6.64 5.65 5.07 2.3 1.8 21.15 127.21 2.151
35 6.8 2.91 3.13 2.84 1.03 60.77 100.4 10.107
Total volume (mm3) 168.2
Foam density (kg/m3) 481.0

– import of all the ellipsoidal cells in the chosen FEM code by neu- ellipsoidal cells did not allow meshing the model without admit-
tral .iges format and subsequent assembly in FEM assembly ting the presence of degenerate elements.
module (Fig. 11a): since ellipsoidal cell model was more com- Material model was isotropic with elastic perfectly plastic
plex than Kelvin cell model, the possibility to use parallel com- stress–strain behaviour. The material properties were the same
puting resources advised authors to prefer Abaqus software in of the previous model (Table 1). Boundary conditions were applied
this case; on the lower face, imposing a zero value to z-direction displace-
– generation of planes delimiting the cubic volume of the model: ment and rigid body boundaries; load was applied to the upper
only the top and the bottom face of the cube are considered surface as a 10 MPa pressure load distributed in three load steps
material rigid planes (Fig. 11b), while the remaining were used to overcome convergence troubles.
as trimmer planes (Fig. 11c); Displacement map in Fig. 12b was quite uniform in the z-direc-
– trim and elimination of the surface regions that reverted outside tion and the upper nodes presented the highest values of displace-
the cube (Fig. 11d): this operation reduced the amount of sur- ment since they were in proximity of the loaded surface.
faces that constitutes the model, therefore volume material Displacement distribution that has been obtained in the model re-
and foam density reported in Table 4 decreased; however the flected the expected behaviour, confirming the approach validity.
use of a wall thickness of 0.11 mm allowed to obtain a model The ratio of upper surface displacement to model thickness al-
characterised by a final density of 438 kg/m3 that is very close lowed to calculate the global strain of the foam and, then, the
to experimental mean value; stress–strain curve was obtained dividing the displacement by
– application of boundary condition and pressure load on the model thickness (Fig. 13). In order to carry out a reliable compari-
lower and upper plane, respectively. son of numerical and experimental behaviour, not only the vari-
ability of foam density but also the scatter of experimental data
Fig. 12a reports the assembly of the ellipsoidal cells and the ri- must be considered. For this reason, mean values of elastic modu-
gid surfaces. Fig. 12b, which reports the z-displacement map, al- lus and plateau stress of the specimens, having the same apparent
lows to evaluate the discretisation utilised for the mesh. In density of the numerical model, have been calculated and used for
particular, the elements were dominant quadratic shell S8R, having a direct comparison (Table 5). It is important to observe that this
eight nodes and reduced integration. The mesh consisted in model was capable to predict the first stage of the stress strain
133,428 nodes and 93,228 elements. The geometrical complexity curve plateau. Besides, experimental and numerical maximum
of the structure in correspondence of intersection line between stress were practically coincident, while a difference still existed
 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35 33

Fig. 11. Model assembly steps: (a) import of ellipsoidal cells, (b) generation of upper and lower planes, (c) generation of trimmer planes and (d) removing of the exceeding
surfaces.

Fig. 12. (a) Random model based on ellipsoidal cells. (b) Mesh and z-displacements map.

in the elastic modulus. This difference, that was however the best of internal contact region is created. Due to its complexity, it was
approximation that has been obtained, seemed to be quite large, not possible to accomplish this phenomenon into the model. As a
but the direct comparison of numerical and experimental stress– consequence, model accuracy rapidly decreased and finally conver-
strain curve revealed a good agreement. In particular, the graph gence problems appeared
in Fig. 14, where both the numerical curve and the experimental Ellipsoidal cell model could be also used to evaluate the influ-
one of Specimen 6 are plotted, suggests that a simple comparison ence on the numerical results when limited changes of apparent
of elastic modulus values could be not adequate to evaluate the density were considered. It was possible, in fact, to consider the
real accuracy of the model that is instead evident through the di- same cell geometry but different values of wall cell thickness. This
rect comparison of the two curves. change produced a variability of the foam apparent density, with-
However, this model presented some limitation during the sim- out the need to change the whole model. Four different values of
ulation of compaction phase at strain higher than 3.5%. In these apparent density have been considered in the range
physical conditions the cells begin to collapse and a large number 438–550 kg/m3, obtaining the stress–strain curves reported in
34 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35

Fig. 15. As expected, density increase produced an increase of elas-

tic modulus and plateau stress, even if there is not a direct
relationship between them. Further and minor changes appeared
in the shape of the curves.

5. Conclusions

A comparative study of different metallic foams micro-struc-

tural modelling was carried out. Two different approaches were
considered, which differed for the cell morphology.
The easiest model was based on the regular repetition of a Kel-
vin cell. This cell has been largely used [1,15] because it corre-
sponds to the minimum surface energy for a constant volume.
The simple planar walls model was refined introducing the walls
Fig. 13. Numerical stress–strain curve for the model with ellipsoidal cells.
curvature and material thickening along the faces edges. The cur-
vature introduction remarkably improved the initial elastic region
approximation, but the model was still too rigid with respect to the
experimental data. Moreover, it was impossible to recognise the
Table 5 plateau region characteristic of the foam compression curve. Then,
Comparison of experimental and numerical results of ellipsoidal cell model. when the solid volume fraction concentrated in the edges ex-
Specimen Specimen Maximum stress Density Eunload ceeded the 60%, the wall thickness variation further improved
number type smax (MPa) (kg/m3) (MPa) the approximation, but not in a significant way.
6 D40 5.0 432.5 577 The comparison with the experimental data obtained for com-
8 D40 4.9 432.9 433 mercial AlSi10Mg foam produced by Alulight showed the excessive
9 D40 4.6 435.6 481 stiffness of the Kelvin cell models. The elastic modulus calculated
Experimental mean values 4.8 433.7 497
through the last model resulted more than the double respect to
Ellipsoidal cell model 5.3 438 906
the experimentally one.
The best results were obtained with the model realised by
assembling the ellipsoidal cells having random dimensions, orien-
tation and position. The model was able to reproduce also the pla-
teau region, where cells collapsed and densification phenomena
became evident. Even if the elastic modulus was overestimated,
agreement between experimental and numerical stress–strain
curves was excellent.

References

[1] Ashby MF, Evans AG, Fleck NA, Gibson LJ, Hutchinson JW, Wadley HNG. Metal
foams: a design guide. Butterworth-Heinemann Publications; 2000.
[2] Gibson LJ, Ashby MF. Cellular solids: structures and properties. Cambridge
University Press; 1997.
[3] Banhart J. Manufacture characterisation and application of cellular metals and
metal foams. Prog Mater Sci 2001;46:559–632.
[4] Fusheng H, Zhengang Z. The mechanical behaviour of foamed aluminum. J
Mater Sci 1999;34:291–9.
[5] Kovacik J, Tobolka P, Simancik F, Banhart J, Ashby MF, Fleck NA. Metal foams
Fig. 14. Comparison between numerical and experimental stress–strain curves. and foam metal structures. In: Proc Int Conf Metfoam’99. Germany: MIT Verlag
Bremen; 14–16 June 1999.
[6] Deshpande VS, Fleck NA. Isotropic constitutive models for metallic foams. J
Mech Phys Solids 2000;48:1253–83.
[7] McCullough KYG, Fleck NA, Ashby MF. Uniaxial stress–strain behaviour of
aluminium alloy foams. Acta Mater 1999;47(8):2323–30.
[8] Hanssen AG, Hopperstad OS, Langseth M, Ilstad H. Validation of constitutive
models applicable to aluminium foams. Int J Mech Sci 2002;44:
359–406.
[9] Belingardi G, Capittini F, Martella P. Modellazione numerica del
comportamento a compressione di schiume strutturali. Salerno, Italy: XXXII
Conv. Nazionale AIAS; 2003.
[10] Roberts AP, Garboczi EJ. Elastic moduli of model random three dimensional
closed-cell cellular solids. Acta Mater 2001;49:189–97.
[11] Lu TJ, Ong JM. Characterization of close-celled cellular aluminium alloys. J
Mater Sci 2001;36:2773–86.
[12] Maire E, Fazekas A, Salvo L, Dendievel R, Youssef S, Cloetens P, et al. X-ray
tomography applied to the characterization of cellular materials. Related finite
element modeling problems. Compos Sci Technol 2003;63:2431–43.
[13] Youssef S, Maire E, Gaertner R. Finite element modelling of the actual structure
of cellular materials determined by X-ray tomography. Acta Mater
2005;53:719–30.
[14] Laschet G, Kashko T, Angel S, Scheele J, Nickel R, Bleck W, et al. Microstructure
based model for permeability predictions of open-cell metallic foams via
homogenization. Mater Sci Eng A 2008;472:214–26.
[15] Simone AE, Gibson LJ. Effect of solid distribution on the stiffness and strength
Fig. 15. Initial behaviour of stress–strain curves for different density values. of aluminium foam. Acta Mater 1998;46:2139–50.
 M. De Giorgi et al. / Computers and Structures 88 (2010) 25–35 35

[16] Bart-Smith H, Hutchinson JW, Fleck NA, Evans AG. Influence of imperfections [19] Dattoma V, Nobile R, Panella FW, Tafuro R. Comportamento meccanico di
on the performance of metal foam core sandwich panels. Int J Solid Struct pannelli sandwich in schiuma di alluminio. Milan, Italy: XXXIV Convegno
2002;39:4999–5012. Nazionale dell’Associazione Italiana per l’Analisi delle Sollecitazioni (AIAS);
[17] Andrews E, Sanders W, Gibson LJ. Compressive and tensile behaviour of 2005.
aluminium foam. Mater Sci Eng 1999;A270:113–24. [20] Kaminski MM. Computational mechanics of composite materials. London:
[18] Simone AE, Gibson LJ. The effect of cell face curvature and corrugations on the Springer-Verlag; 2005.
stiffness and strength of metallic foam. Acta Mater 1998;46:3929–35.