Efficient computational modelling of closed cell metallic foams using a morphologically controlled shell geometry

•A shell-based RVE generation methodology is proposed for closed cell metallic foams.•The surface geometry is extracted from an implicitly defined 3D geometry.•A parametric study using shell FE models is performed and compared to 3D FE results.•The accuracy and efficiency of the shell description is compared to the 3D model.•The shell model is used to reproduce a foam compression test up to densification. [Display omitted] This contribution addresses the finite element modelling of closed cell metallic foams using Representative Volume Elements (RVEs) based on shell geometries directly extracted from implicitly defined 3D geometries. 3D RVEs of closed cell foam materials are produced by means of a generation strategy allowing a close morphological control reproducing fine scale geometrical features incorporating cell size, cell wall thickness and cell wall curvature distributions. The strategy is built on three computational ingredients: (i) a random packing algorithm based on random sequential addition assisted by neighbour distance control, (ii) a distance field-based shape tessellation (morphing) that allows incorporating cell wall curvatures and varying cell wall thicknesses and (iii) a close control on the shape of the cells. In order to decrease the computational cost of a full 3D finite element model, an original approach is proposed to produce a shell-based geometry directly from 3D information. Extracting the shell geometry from the implicitly defined 3D geometry based on the zero level of distance fields that would represent the cell walls is computationally impossible. Therefore, a novel robust procedure is proposed using careful cutting operations on distance fields for this purpose. The effect of the different microstructural geometrical features of interest on the average mechanical behaviour of the foam is investigated using shell-based finite element analyses. The computational cost and the accuracy of the proposed shell models are then assessed by comparing their results to full 3D simulations. The macroscopic behaviour of the generated shell-based model under compressive loading is then assessed up to the densification stage (including contact), and compared qualitatively with experiments from the literature. The macroscopic behaviour of the shell-based model is explained by linking it to cell/wall level deformation mechanisms.