Numerical Investigation on Crack Propagation of Three-Dimensional Pipeline Under Seismic Wave Loading

Cracks in transport pipelines significantly impact their structural safety, making pipeline fracture damage a major concern. While experimental tests can study crack growth, their setup and costs are substantial. To address this, our work involves parametric analyses to investigate the effects of initial crack length, load types and sizes, and pipe wall thickness on the crack tip stress field. The finite element model is established according to the actual operation, and the crack defects of the pipe are established by the extended finite element method (XFEM). The crack tip variation is studied by extracting the von Mises stress and strain at the crack tip. When the crack propagates, the stress field at the crack tip will fluctuate in a certain range. The relationship between stress field fluctuation and crack propagation was studied by changing the XFEM crack growth for comparative analysis. Utilizing 429 numerical results obtained from a parametric numerical model implemented with Python and Abaqus, we establish a back propagation (BP) neural network prediction model to forecast the stress field at the crack tip. The relative errors between the numerical results and the predictions by our model remain below 5%. In conclusion, our proposed approach for analyzing pipeline fracture under multiple loads proves valuable for pipeline safety assessment, and it offers useful insights for evaluating the suitability of transportation pipelines.