Computational micromechanics of heterogeneous materials

Metal Matrix Composites (MMCs) of the type Al/SiC or Al/Al2O3 contain significant residual stresses due to different thermal expansion coefficients from the metal and ceramic constituents. They are believed to influence the mechanical properties of these materials to some extent - including changes in their failure behaviour. In this contribution, a physically based micromechanical approach is applied in order to study the influence of residual stresses on local as well as global properties of MMCs. A representative microstructural cut-out of an Al/10%SiC-composite is carefully meshed with finite elements in order to take phase boundaries into account. This mesovolume possesses all characteristic features of the material, such as volume fraction, distribution characteristics as well as the shape of the particles. The deformation behaviour of this microstructure is analysed under large compressive external loading up to strains of about 10%. In addition, the failure behaviour is modelled using Rice&Tracey's damage parameter which was previously shown to model microstructural failure to a good approximation. It is found that although residual stresses do have some impact on failure initiation in the microstructure, strains due to external loading are much more of importance in this respect. In order to illuminate the influence of particle shape and arrangement, artificial two-dimensional microstructures are analyzed as well. It is found that real irregular particle shapes are much more prone to fracture as compared to artificial regular shapes and that particle allignments are not beneficial with respect to failure aspects. The results are given with respect to the maximum value and the distribution of the damage parameter. it is found that in most cases analyzed, damage follows the pattern of plastic deformation and is much less influenced by hydrostatic stresses than expected. Thus, damage nucleates between clusters of particles where shear deformation is concentrated in the matrix.