PSI - Issue 76

Vladimír Mára et al. / Procedia Structural Integrity 76 (2026) 123–130

129

the fact that in the case of T200, there are no changes in the length of cracks compared to noHT, since the Si network dimensions do not change at all during the annealing process (see Fig.2b). In LPBF AlSi10Mg alloy, crack tends to propagate along the MPs, especially in regions with higher porosity and larger defects. Weaker boundaries also creates preferential paths for effective crack growth (Zhu et al., 2024). The highest values of a ECG were measured for T300, where MP boundaries have the highest degree of degradation of eutectic Si network. Although the β -Si network is completely dissolved after T6 treatment, a ECG continues increasing. This can be associated with a higher fraction of defects in the microstructure (1.38 %) of batch No. 3 specimens (to which T6 and T6mod configurations belong), which promote crack propagation. Growth occurs regardless of the size of the killer defect. Value of K Imax is changing based on the applied heat treatment – values are increasing with gradual degradation of β -Si network. For noHT, T240, T200, T300, T6 and T6mod, the values are 17.6; 16.5; 15.7; 18.4; 19.9 and 19.4 MPa·m 1/2 respectively. Values are order of magnitude consistent with work (Xu et al., 2020). Fig.6c illustrates the influence of defect size on the predicted fatigue strength for all specimens. It is evident that predicted fatigue limit is decreasing with increasing defect size. The results also follow the Murakami’s empirical -1/6 slope, which corresponds to the work (Nadot, 2022).

Fig. 6. (a) Classification of killer defects; (b) Relationship between the defect size, heat treatment type and length of early crack growth; (c) Kitagawa diagram showing predicted fatigue strength using Murakami equation for various heat-treatment regimes.

A comparison of the fatigue limits shows good overall agreement between experimental and predicted values (see Figure 7a). The only exception, however, is the as-built state (no HT), which falls outside the 95% prediction band. Figure 7b illustrates the effect of defects and heat treatment on fracture surface formation. Treatments that cause significant disintegration or total dissolution of the β -Si network (T300, T6, and T6mod) are the least susceptible to fracture region size, regardless of defect size. The A CGR /A RFR ratio remains similar, with no changes observed even with killer defects up to 150 µm in size. Conversely, the as-built (no HT) and modified (T200, T240) cellular structures show substantial changes in A CGR /A RFR size. In these cases, the A CGR regions increase with increasing defect size: slightly for noHT and significantly for T200 and T240.

Fig. 7. (a) Comparison between experimental and predicted fatigue limit with prediction band; (b) Change in area of fracture regions based on the heat treatment and critical defect size; (c) Relationship between fatigue crack growth rate da/dN and GND density with confidence band.

The effect of heat treatment on fatigue crack growth rate is shown in Fig. 7c. Celullar Si morphology and high GND density retard crack propagation, while dissolution of the β -Si network and decrease of GND lead to increased