PSI - Issue 68

Chahboub Yassine et al. / Procedia Structural Integrity 68 (2025) 310–317 CHAHBOUB Yassine/ Structural Integrity Procedia 00 (2025) 000–000

311

2

by the operators of the plant, mainly to avoid unwanted fractures that may lead to disastrous scenarios for human life and the environment. Nuclear power plants need all components in good working conditions to ensure high performance, but also that they be safe. In practice, even small-scale pipeline leakages can cause much more impact on the general efficiency and safety of these installations. The GTN model is one of the most widely implemented models over the last five decades in the study of pipeline ductility as applied to nuclear purposes. Nevertheless, there are scant studies involving the application of advanced optimization techniques-ANNs-for the prediction of pipeline failure in the nuclear sector. While most of these studies directly predict failure, a few have sought to probe the model GTN combined with state-of-the-art technologies like ANN or hybrid swarm optimization. To this end, A. On the other hand, Alswaidan et al. (2016), Taslim D. Shikalgar et al. (2020), and E. J. Pérez-Pérez et al. (2021) showed that it had to be established because such integration might increase predictive accuracy and reliability. The paper is aimed at performing a statistical comparison between conventional direct methods and artificial neural network approaches for making evident potential benefits that could be involved in using artificial intelligence for improved predictions about pipeline failure within the nuclear field. 1.1. Introduction to GTN model The idea of Gurson (1975) is based on the model of Rice et al. (1969); Gurson suggested a micromechanical model based on a fracture mechanics perspective, such as the critical evolution of voids; Gurson developed an approximate yield criterion for ductile materials, by proposing a continuum theory for ductile failure based on void nucleation and void growth and by showing the important role of hydrostatic stress in plastic yield and void growth, the yield criterion developed by Gurson took as a basis the idealization of the material matrix as being perfectly rigid and obeying the Von-Mises yield criterion. The Gurson model only shows the growth stage of material failure. It needs to be expanded to include the nucleation and coalescence stages. So Tveegard and Needlman (1986) could make modifications to the model and introduce new parameters, as shown in the modified model of the Gurson model, which gave birth to the GTN model; the definition of the new parameters will be defined in this section : The model is defined as : (1) In which q 1 refers to the material constant, and trσ is the sum of principal stresses; in addition to this, the parameter σ M is the equivalent flow stress and the new parameter f* is the ratio of voids effective volume to the material volume ratio defined as follows: (2) (3) Where f is the voids volume ratio, f c is the voids volume ratio at the beginning of nucleation, and f f is the voids volume ratio when the fracture occurs. σ M is the equivalent flow stress, and it is obtained from the following work hardening relation: (4) In which n is the strain-hardening exponent, and Ꜫ M is the equivalent plastic strain. ( ) ! ! ! " " ! ! #$%& " ! ! " " #$ % & % & ! ! " ! ! # # $ % = + & + ' ( ) * ( ) ( ) ! ! " " " #" " " ! = " ( ) ( ) ( )( ) ! !" ! ! ! ! " ! # " " " " " " $" " " " " ! " = + " # " ( ) ! ! "# "# $ $ $ % % ! " ! " ! # $ = + % & % & ' (