Eventos Anais de eventos
MECSOL 2022
8th International Symposium on Solid Mechanics
An implementation of 3D RVE-based computational homogenization considering an elastic-perfectly plastic matrix and spherical voids
Submission Author:
Carlos Alberto da Maia , RS
Co-Authors:
Carlos Alberto da Maia, Andrey Brezolin, Tiago dos Santos, Rodrigo Rossi
Presenter: Tiago dos Santos
doi://10.26678/ABCM.MECSOL2022.MSL22-0040
Abstract
This work investigates the strength of porous materials using a finite element homogenization procedure based on three-dimensional Representative Volume Elements (RVE). The RVE consists of an aggregate composed of randomly distributed non-overlapping spherical voids embedded in an elastic-perfectly plastic matrix, obeying the von Mises criterion. In order to obtain distinct macroscopic stress states, mixed boundary conditions are imposed on the RVE. Using such conditions, it is possible to reach different macroscopic stress triaxialities by varying specific loading parameters. The macroscopic stress tensor is considered to be the average of its microscopic counterpart taken for a stabilized RVE response, which is found in terms of an asymptotic homogenized response. The calculations were carried out using the commercial finite element software Abaqus and the results were post-processed using a Python script, which calculates the stress averages based on information taken at each integration point in the RVE. The simulations are performed employing different initial void volume fractions using multiple pores and single pores, with different boundary conditions, which provide distinct macroscopic stress states. The obtained numerical results allow constructing macroscopic yield surfaces in the deviatoric-hydrostatic stress space depending on the initial void volume fraction and to compare the difference between multiple and single pores in an RVE with in same void volume fraction. The results showed that, for a low void volume fraction, a single pore can represent well a multiple pore model, while for larger void volume fractions, a single pore model presents deviations. The results were also compared with the Gurson and Gurson-Tvergaard-Needleman (GTN) models. It was noted that the Gurson model overestimates the results, while the results of the (GTN) model were close to the computational calculation.
Keywords
plasticity, Representative Volume Element (RVE), computational homogenization, porous metals
