PSI - Issue 68

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

314

5

Fig. 3. Force vs. crack opening displacement for FP1

Therefore, the incorporation of artificial intelligence within simulation will be demonstrated below, as it reduced the computation time from 30 days down to 6 hours. The next section will go into detail regarding the optimization process and the way AI was utilized in order to hasten pipe failure prediction.

2.2. Prediction of failure of the pipeline using ANN approach The application of ANN has a wide area in different fields to increase the output and improve quality. In our work, the integration of ANNs into our process allowed us to make significant optimizations of the time necessary for predictions regarding pipe failure. In the above section, we have shown that GTN model gives good predictions of pipeline failure, but needed 30 days of computation. It should be noted that to save this computation time, only one-quarter of the specimen was simulated taking the advantage of axisymmetry. For more complex and sophisticated equipment, the time would be much longer than three days, which again becomes impracticable. This shows the rise in demand for yet another alternative approach to bridge this gap and accelerate the process of predicting the failure of equipment. When ANNs were adopted, the time used was greatly reduced, hence efficient and feasible.

NT Specimen modelling

The simulation and geometry of the NT specimen are indicated in Fig.4; the reason behind choosing the NT specimen is the short time of computation, which leads us to generate a significant database to train the neural network. To find the GTN parameters using ANN, we decided to make 60 simulations of the NT specimen Fig. 5.

Fig. 4. Simulation of NT specimen