Article révisé par les pairs
Résumé : This contribution addresses the multiscale computational modelling of closed cell metal- lic foams by means of an integrated Representative Volume Element (RVE) generation and computation strategy. The microstructural geometry is computationally generated by con- trolling relevant fine scale features such as the distribution of cell sizes, the spatial orga- nization of cell sizes and that of cell wall thicknesses and curvatures. The number of faces per cell and of edges per face are also set to comply with the experimentally observed values. The computational generation of the RVE is built on three ingredients: (i) a ran- dom close inclusions packing algorithm based on random sequential addition assisted by neighbour distance control, (ii) a distance field-based shape tessellation (morphing) that allows reproducing cell wall curvatures and varying cell wall thicknesses from the inclu- sions packing, (iii) a close control of the shape of the cells. The RVE geometry is thus de- scribed using implicit functions, thereby allowing a seamless transition towards a recently developed mesh generation technique for heterogeneous microstructures represented by such implicit functions, enabling simulations in standard softwares. This controlled gener- ation methodology is illustrated based on experimental data available in literature for mor- phological indicators relevant to the foam mechanical behaviour. A qualitative and quanti- tative agreement between FE results and experimental data is obtained for the mechanical response of a commercially available ALPORAS foam. The individual contribution of each microstructural feature (size distributions, wall thickness and curvatures) to the average behaviour of closed cell foams is assessed through FE computations on increasingly com- plex geometries.