Reliability assessment of dented pipelines transporting hydrogen by artificial neural network

The transportation of hydrogen energy through existing aged pipelines is a global concern, yet research on reliability assessment is limited. This work utilizes theoretical analysis, finite element (FE) simulation and an Artificial Neural Network (ANN) model to establish an assessment framework for the reliability of X52 dented pipelines transporting gaseous hydrogen. The degradation of the fracture toughness related to the hydrogen concentration was considered as the failure pattern of the pipeline. Denting and hydrogen diffusion processes were simulated using FE method. With the increasing indentation displacement, the simulation results indicated that the maximum hydrogen concentration shifted circumferentially away from the dent center of the pipelines. Using FE data from this study and the literature, an optimizable ANN model was trained through hyperparameter optimization. This led to the establishment of a reliability assessment framework for dented pipelines degraded by hydrogen, through the determinations of failure probabilities.