Reliability Calculation and Analysis of Influencing Factors for Girth Welds in High-Grade Steel Pipelines Based on HMC-SS
Abstract
Fracture failure of girth welds in high-grade steel pipelines poses a critical threat to pipeline integrity. Leveraging enhanced digitalization in pipeline engineering, a statistical database has been developed to support reliability analysis based on actual operational data. This study utilizes real project data to analyze the failure probability and key influencing factors of girth welds containing crack defects, thereby providing theoretical support for safety design and risk management. To overcome the conservatism of traditional deterministic methods, a probabilistic reliability model was established, incorporating a modified PRCI-CRES ultimate tensile strain criterion. Addressing the inefficiency of standard Monte Carlo (MC) simulation in high-dimensional low-probability contexts, an efficient Hamiltonian Monte Carlo-Subset Simulation (HMC-SS) strategy was introduced. Results show that HMC-SS improves computational efficiency by 99.95% over MC, with only 0.90% relative error. Key findings include: crack depth has the strongest influence – variation from 0.92 mm to 3.68 mm, which increases failure probability by 103 times; the strength matching coefficient is dominant, and higher values reduce failure risk; strain demand exhibits a positive correlation with failure probability and couples with material properties. It is concluded that high- or equal-strength material matching should be emphasized in welding, and reliability-informed design should account for multi-parameter interactions to ensure global safety.