A predictive model for the ultimate burst pressure of hydrogen-blended natural gas pipelines considering the spatial non-uniformity of material properties

Hydrogen-blended natural gas transportation via existing pipelines represents an effective approach for large-scale, efficient hydrogen delivery. Existing natural gas pipelines have various defects, which can exacerbate hydrogen embrittlement. Currently, there are few models predicting the ultimate burst pressure of hydrogen-blended pipelines with corrosion defects, and existing models do not take into account the spatial non-uniformity of pipeline steel material properties. Therefore, this study considers the spatial non-uniformity of mechanical properties in X52 pipeline steel caused by the concentration gradient of hydrogen atoms in the pipeline steel. Finite element analysis (FEA) was used to calculate the ultimate burst pressure of the pipeline under different hydrogen partial pressures and geometric parameters of corrosion defects. The results obtained were used to train a genetic algorithm - back propagation (GA-BP) neural network model and modify existing models for ultimate burst pressure. The results indicate that within the hydrogen partial pressure range of 0 to 2 MPa, the change in ultimate burst pressure caused by hydrogen embrittlement is within 3%, suggesting low sensitivity of X52 steel to hydrogen embrittlement. It is meaningful to consider the spatial non-uniformity of the material caused by hydrogen diffusion.