PSI - Issue 68

Giovanni Chianese et al. / Procedia Structural Integrity 68 (2025) 1245–1251 Chianese et al. / Structural Integrity Procedia 00 (2025) 000–000

1246

2

Nomenclature a

crack length measured on the lateral surface

B

thickness of the C(T) specimen

BFS back-face strain CNN convolutional neural network FE finite elements ROI region of interest W width of the C(T) specimen α aspect ratio of the crack front β

crack length normalized with respect to W

ε 1 , ε 2

local strains measured with strain gauge 1 and 2

ε X , ε Y strain fields along the X and Y direction

1. Introduction Fatigue loads can drive propagation of cracks that may occur in mechanical components and subsystems leading to fracture (Chandawanich and Sharpe (1979); Stephens et al. (2000); Leonetti (2021)). Experiments are carried out to evaluate crack propagation rate in metals (De Iorio et al. (2012)), and international standards provide procedures for analysis of data to monitor crack length. In this context, indirect monitoring of the crack length is a topic of interest, because it does not require analyses (by microscope and/or crack gauges) of the zone near the crack tip. In this regard, standard and studies indicate procedures to monitor crack length as a function of compliance, which is calculated from crack mouth opening displacement (ASTM E399, 2024) and back-face strain (BFS) (ASTM E647, 2024). In studies by Newman Jr et al. (2011), and Riddel and Piascik (1998), authors established relations between BFS- compliance and the crack length based on 2D finite element (FE) analyses and assuming a straight crack front. However, preliminary 3D FE analyses carried out in the present work showed that normalized back-face compliance deviates from relations proposed in these studies when elliptical crack fronts are considered. The occurrence of elliptical crack front is also termed as crack tunneling, and is a key factor in the framework of the damage tolerance approach (Carpinteri (1993); Jeong et al. (1997); Carpinteri et al. (2013); Pucillo et al. (2019)). In Fig. 1, results from 3-D FE simulations with straight crack front (red solid line) show good agreement with the relation provided by Newman Jr et al. (2011) (black solid line); on the other hand, dashed lines, which report results from simulations with increasing crack tunnelling, show a deviation in terms of normalized back face compliance. This preliminary analysis showed that the specimen compliance depends on both the crack length and the crack front geometry indicating that monitoring the crack length based on one measure of the compliance only is not feasible. In this regard, the standard ASTM E647 prescribes the use of corrective factors during processing of data collected during the test when crack tunnelling exceeds 5%.

Fig. 1. Normalized back-face compliance calculated with 2D and 3D FE analyses in specimen with crack tunnelling conditions and straight cracks.

For all these reasons, this study aims to propose a novel technique for in-situ and real-time simultaneous monitoring of the crack growth and tunnelling condition. As use of one scalar measurement is not feasible to