PSI - Issue 68

Stefan Fladischer et al. / Procedia Structural Integrity 68 (2025) 486–492 S. Fladischer et al. / Structural Integrity Procedia 00 (2024) 000–000

491

6

2.00

single

1.0

interacting

convex hull

1.75

fit interacting

fit convex hull

0.9

1.50

1.0

2.00

0.8

0.9

1.25

[ ]

[ ]

1.75

0.8

gle

1.00

1.50

[mm]

0.7

,sin

NPC

NPC

0

0.7

NPC

a

NPC, single

1.25

0.75

0.6

0.6

1.00

[mm]

0

a

0.50

0.5

0.75

2.0

0.5

0.50

single

0.25

1.5

interacting

]

0.0

1.0

0.25

convex hull

0.5

1.0

[mm

0.5

0.4

0.00

1.5

d

0

0

2.0

a

0.00

0.0 0.5 1.0 1.5 2.0 2.5 3.0

a

[ ]

2.5

0

0.0

a)

b)

3.0

d

0

[ ]

a

0

(a) over d 0 a 0

; (b) over d 0 a 0

∆ σ NPC ∆ σ NPC , single

Fig. 6. Interaction e ff ects in terms of relative NPC fatigue limit

and a 0 .

For crack spacings d 0 ≤

a 0 / 2 , the non-propagating coalesced cracks are decisive for the fatigue limit. Noteworthy, the

interacting crack results for any relative crack spacing d 0 / a 0 show little scatter with respect to the crack size a 0 . This is due to the fact, that the crack driving force is closely related to the geometry factor Y , which is fairly invariant to the crack size a 0 . This is especially true for Y 0 , relevant for the formation of non-propagating cracks at small crack extensions. Due to the selected normalization, any crack size e ff ects, that may also be present in the single crack case, are canceled out. Overall, the interaction e ff ect results are in accordance with Murakami (2019) and Åman et al. (2020), regarding the occurrence at relative crack spacings d 0 / a 0 . The experimental results are summarized in Fig. 7a, presenting normalized maximum run-out load levels ∆ σ RO of specimens containing two interacting defects with respect to the corresponding single flaw results ∆ σ RO , single These follow the general trend of simulated interaction e ff ects, while not precisely coinciding with the simulation prediction. Two disparities between the experimental observations and simulation results need to be mentioned. For one, local initiation at the bottom of the flaws with respect to the SLM build direction leads to non-symmetric crack growth paths, cf. Fig. 7b. This deviates from the simulation assumptions of circumferential crack initiation around the additively manufactured voids followed by symmetric crack growth. Additionally, the endured run-out load levels are significantly higher than the predicted by the fracture mechanical assessment under conservative assumptions of ideal crack like flaws, e.g. with ∆ σ RO , single = 120 MPa for single defect specimens of a 0 = 0 . 5 mm, whereas the simulation results lie at ∆ σ NPC , single = 49 MPa. This improves to ∆ σ NPC , single = 74 MPa for the higher ∆ K th , lc of the cyclic R-curve SLM05, as specified in Table 1. The tested crack configurations are at and beyond the upper limit of applicability of the √ area-concept (Tanaka and Akiniwa, 2003; McEvily et al., 2003; Scho¨nbauer et al., 2017). Smaller flaw sizes should lead to better accordance of experimental and simulation results, but the melt pool size of the SLM process imposes a limit on the manufacturability of artificial defects in terms of shape accuracy and notch sharpness. The present study focused on the numerical short crack growth simulation of neighboring defects, in order to determine interaction e ff ects in terms of the fatigue limit of maximum non-propagating crack formation for large range of flaw sizes a 0 and interaction distances d 0 . This work yields novel geometry factors for interacting cracks, that consider the adjacency relationship of interacting pores instead of evaluating only one flaw, e.g. by extremal cross section or based on the convex hull of the studied pores. Interaction e ff ects are evaluated in terms of the degradation of the fatigue limit of interacting defects with respect to the individual one, based on simulations and experimental results. Conservative assumptions of crack-like defects lead to conservative predictions of fatigue limits, but the resulting interactions e ff ects correlate satisfyingly with the observed experimental results. 3.3. Experimental validation 4. Conclusions