PSI - Issue 76

Daniele Rigon et al. / Procedia Structural Integrity 76 (2026) 35–42

36

characteristics such as surface roughness, internal porosity, and residual stresses, all of which can significantly degrade fatigue performance Leuders et al. (2013); Tammas-Williams et al. (2015); Torries et al. (2017); Yadollahi et al. (2017); Zerbst et al. (2021). In the as-built condition, PBF-LB / M parts often exhibit high surface roughness with partially fused particles, micronotches and waviness (Diaz (2019)), making fatigue strength prediction challenging. These surface features serve as natural crack initiation sites and therefore fatigue life is typically governed by early-stage crack nucleation and short crack growth mechanisms (du Plessis and Beretta (2020)). Previous studies have used various approaches to characterize and model the fatigue behavior of as-built spec imens. Among these, models based on fracture mechanics have shown good potential to rationalize the e ff ect of manufacturing-induced defects on fatigue limits (Nakatani et al. (2019); Barricelli et al. (2023); du Plessis and Beretta (2020). In particular, Rigon et al. (2024) proposed a method grounded in fracture mechanics principles to estimate the fatigue limit of PBF-LB / M Ti6Al4V specimens in the as-built state, introducing a semicircular surface crack approach based on areal roughness measurements combined with extreme value statistics (EVS). The EVS enable the possibility of estimating the deepest micro-notch of the net-shape or as-built surface condition in a reference area (which can be the highest loaded region of the component), through a block maxima sampling of a limited portion of the interest area (see Rigon et al. (2024)). In the previous work (Rigon et al. (2024)) the evaluation of the critical equivalent surface crack led to satisfying fatigue limit estimations through the Atzori-Lazzarin-Meneghetti (ALM) diagram Atzori et al. (2003, 2005) of Ti6Al4V specimens in as-built surface condition. It is worth noticing that in Rigon et al. (2024): i) the demonstration that the crack was initiated due to the roughness and not because of a subsurface defect was performed by analyzing the fracture surfaces of the specimens (thus a destructive analysis) and ii) the estimation of the deepest notch in the entire gauge section of the specimens was done simply using EVS and it was not verified by identifying the actual deepest notch of the as-built surface, due to the unavailability of the areal measurement of the entire gauge area of the specimens. The present paper presents a preliminary attempt to overcome the latter issue by using an experimental measurement setup able to evaluate the deepest notch from optical profilometry of surfaces texture of cylindrical specimens, subsequently compared with an estimation through EVS applied to set of data having di ff erent sample size. In addition, a preliminary characterization of the internal defects interacting with the surface roughness was performed using X-ray computed tomography (CT), with the objective of formalizing a procedure to estimate the fatigue limits of a batch of samples with as-built surfaces in future work. 2. Theoretical framework In this section, the strictly necessary theoretical background of EVS applied to PBF-LB / M as-built surface is reported. The interested reader is referred to Rigon et al. (2024) for the complete framework. A widely adopted model to predict the fatigue limit of components with notches, defects, or cracks in di ff erent sizes and scales is the one proposed by Atzori, Lazzarin, and Meneghetti (Atzori et al. (2003, 2005)). Since the analysis deals with defects and non-propagating cracks, the fatigue limit for a given load ratio R, expressed in terms of remote stress range ( ∆ σ g , th ), can be estimated in the normalized ALM diagram by the following equation:

Fig. 1. (a) Specimen’s geometry and (b) microstructure of the Ti6Al4V after heat treatment