Analysis of collapse response of offshore pipes under external pressure and bending considering sizing effect

Offshore steel pipelines are being increasingly utilized to transport gas and oil over long distances, resulting in a higher need for reliable assessments of their failure response to combined external pressure and bending throughout all phases, from installation to operation. Electric resistance welded (ERW) pipes, produced through a cold-forming and welding process, are commonly used. In this research, the manufacturing process of ERW pipe was simulated though numerical analysis using ABAQUS. Subsequently, the formed pipe was utilized in numerical collapse analysis to investigate the collapse behavior of pipe subjected to combined loads, considering the stress-strain history induced by the forming process. The improvement of pipe collapse performance was assessed through the sizing process by varying sizing ratios and pipe geometries. The results show that the collapse limit rises with increasing sizing ratios, attributed to the strain hardening introduced during sizing process. Additionally, a stacked ensemble learning (SEL) model was established to estimate the failure pressure of steel pipes under combined loading. The effectiveness of developed SEL model was assessed against seven fundamental ensemble machine learning models, with the SEL demonstrating superior accuracy. Lastly, Shapley Additive Explanations method was adopted to evaluate the influence of input parameters on the prediction outcomes.