PSI - Issue 68

Xingling Luo et al. / Procedia Structural Integrity 68 (2025) 694–700

697

4

Xingling Luo et al. / Structural Integrity Procedia 00 (2025) 000–000

2. Methodology 2.1. Microstructure-based modelling

In this section, the microstructure-based modelling details are summarised. All the models were developed in Abaqus/Explicit, with their dimensions based on a statistical analysis of geometrical parameters of graphite particles in CGI (Table 2). The volume fraction of graphite particles was calculated with ImageJ. Models with single and double inclusions used the following two assumptions: (i) the metallic matrix was assigned the effective CGI properties, and (ii) the volume fraction of graphite was around 8.3%, chosen as the average volume fraction of graphite in Table 2. Table 2. Minimum and maximum values of geometrical parameters of graphite particles in CGI (Palkanoglou et al., 2022). Volume fraction of graphite (%) Perimeter (μm) Area (μm 2 ) Major axis (μm) Minor axis (μm) 5.2-11.37 3.54-315.88 0.99-6086.96 0.602-67.96 0.52-28.51 The effect of the shape of graphite inclusions has a significant impact on its fracture behaviour. To study this fracture behaviour systematically, it is crucial to develop a model that separates the effects of individual graphite inclusions, thus eliminating the influence of their interaction (Luo et al., 2024b). Hence, the effect of graphite was investigated first for a single graphite inclusion. After that, the effect of the shape and spacing between graphite particles was analysed. However, it was also important to generate both real and simplified models of areas with many inclusions based on CGI micrographs that could more accurately predict crack paths. In the end, Python scripts were used to create a set of random-inclusion models. The size of the square domain in single- and two-particle models was 100 μm, while for dimensions of real and simplified models were 420 μm and 340 μm. The square domain in random particle models was 500 μm. The tensile displacement loading equivalent to 1% strain was applied to all models. Periodic boundary conditions (PBCs) at the cell’s edges were implemented in all the models, except for the random inclusion case (it was found that PBCs had little effect on these models).

Fig. 2. Summary of graphite morphology used in models.

The CZM method was applied in the first two unit-cell models, whereas the Johnson-Cook (JC) damage model