In this paper we study the influence of morphological parameters
of composite materials on their effective mechanical properties.
The usage of composite materials for industrial applications
motivated a huge amount of publications on the subject in recent
years: they concern both experimental and modeling results. The
reason for us to address the question is twofold as well: on the
one hand we explore the existing modeling techniques, on the
other hand we have in mind very concrete applications related to
the project in the industry of aeronautics.
The need in modeling for the analysis of composite materials as
in most of the applied domains comes from the fact that experimental
work is usually expensive and difficult to carry out. It is
thus important to develop modeling approaches that are efficient,
reliable and sufficiently flexible so that the outcome can be validated
by an experiment. The strategy that we adopt here is related
to the notions of stochastic homogenization. The key idea is to consider
a sample of a composite material that is sufficiently large to
capture its behavior and compute the macroscopic parameters:mechanical properties it can be for example the Young modulus,
the Poisson ratio or eventually the whole stiffness tensor. To take
into account possible imperfections or random factors one can
average the result for a series of tests representing the same macro
characteristics. The usual technology for this is to generate a series
of samples (representative volume elements) randomly, controlling
though their parameters, perform the computation for each
of them and average the result.
It is now generally accepted that the main characteristic affecting
the effective properties of a composite material is its morphology,
i.e. the combination of geometric characteristics of the
inclusions and their distribution in the supporting matrix. To analyze
the phenomenon one needs thus a tool to generate RVEs capturing
various morphological parameters. We have developed and
implemented such a tool: in [1] we described the algorithms to
produce the RVEs containing spheres (that represent globular inclusions)
and cylinders (that are responsible for fiber-type reinforcements).
We are able to reach the volume fraction of inclusions up
to relatively high values of 50–60%, and in addition we can control
the geometric configuration of a sample as a whole, namely manage
the intersections of inclusions and eventually their distribution.
Moreover in [2] we have extended the method to introduce irregularities
to the shape of inclusions. In this paper we describe the
results of computations carried out with the generated