PSI - Issue 25

4

Aleksandr Shalimov et al. / Procedia Structural Integrity 25 (2020) 386–393 Author name / Structural Integrity Procedia 00 (2019) 000–000

389

a

b

Fig. 2. (a) Mesh for RVE with pores volume fraction 25%; (b) Mesh for RVE with pores volume fraction 50%

3. Fracture modelling results Several RVEs of different volume fraction and size were generated for analysis. The FE model was transferred to ABAQUS and tensile mechanical loading along the Y axis was set during computations. To avoid undesired border effect, the load was applied to both XZ-plane surfaces in the opposite direction. The general aluminum elastic properties were taken: Young’s module 70000 E  MPa, Poisson coefficient 0.33   . The strength properties were considered equal for the whole geometry, therefore, only morphological configuration had influence on fracture of the individual parts of microstructure. The isotropic maximal stress criterium was used for all six components of the stress tensor, with max 250   MPa as the ligament strength constant (Zhou et al., 2005). The damage model was implemented with UMAT subroutine using the following damage accumulation model, with 0.9 D  upon reaching the maximum stress value: ( ) D  C   (2) 22 D in the cross section of the RVE with 50% volume fraction. The stress concentrators in the studied models were observed in the thinnest ligaments and leaded to initiation and propagation of damage into more dense parts of the structure. All components of the damage tensor ij D were controlled and it was noted that even thought the load was applied along the 2-direction (Y axis), the structural randomness resulted in uneven accumulation of damage in different directions as well. Fig. 4a and 4b demonstrate field of damage 22 D for RVE with volume fraction 50% and two different dimensions: 5x5x5 mm and 10x10x10 mm, respectively, obtained at the loading displacements value of 0.035 mm. Fig. 3 shows several consequent stages of damage accumulation