PSI - Issue 68

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

1247

3

uniquely solve a problem depending on more than one parameter, use of additional measurements, either local and full field, is investigated with two different approaches. 3-D FE analyses enabled collection of a dataset which was used to test the above-mentioned approaches. 2. Materials and Methods 2.1. Proposed methodology The problem is formally stated in (1), where ε measured indicates one or more strain measurements, α and β are normalized values respectively related to the crack length and the crack front aspect ratio. ! $ = ( !"#$%&"' ) ; = ( # ; β : crack front aspect ratio (1) The problem stated in (1) was addressed by investigating two approaches: 1. fitting functions, where the independent variables are two local strains (in real cases, they are measured by means of electrical strain gauges); 2. CNNs, which can process full field data (in real cases, they represent measurements acquired with full field experimental techniques, such as the digital image correlation). For the purposes of this study, input data for both approaches, namely the two local strains and the strain fields, were obtained by means of FE analyses that reproduced a variety of geometrical conditions. Dedicated Subsections in the following describe details about the implementation of these approaches. 2.2. Simulations with FE analyses A C(T) specimen was modelled for 3-D linear elastic analyses in ANSYS WORKBENCH 2023 R1. Geometric dimensions were defined according to the ASTM standard E 399 with width W = 26 mm and thickness B = 6.2 mm, as shown in Fig 2 (a). The global coordinate system was defined so that the specimen is symmetric with respect to the horizontal X-Z plane and the vertical X-Y plane. The Z-axis passes through fictious vertexes of the notch at the lateral surface. A 0.01 mm displacement in the Y direction was imposed to the axis of the cylindrical surface of the top hole, whereas the bottom one was fixed. Material properties of a generic steel are considered in this analysis, i.e. the elastic modulus and Poisson ratio, respectively equal to 200 GPa and 0.3.

Fig. 2. Geometric dimensions of the C(T)-specimen model with applied boundary conditions (a); parametric geometry of the elliptic crack (b); paths SG1 and SG2, and lateral surface S defined to scope the output of the simulations (c).

The crack geometry was modelled by means of an elliptic surface which lies on the X-Z plane. The center of the ellipse lies on the X-axis and its position depended only on the parameter a center , which is the distance along the X direction from the axis of the holes, see Fig. 2 (b). The orientation of the ellipse is such that the two semi-axis are referred to as R depth and R width and were parallel to the X and Z-axis, respectively. Different geometries of the crack front resulted in different aspect ratio β = R depth /R width . Table 1 reports values of β considered in this study. These values were chosen in order to obtain a quasi-straight crack front, and crack front