Objective Girth weld failures pose a significant threat to the intrinsic safety of high-grade steel pipelines. With the ongoing digitalization of pipeline construction and operation and maintenance in China, a statistical database of basic parameters applicable to high-grade steel pipelines has been established gradually. This database provides critical data support for the reliability analysis of girth welds in these pipelines. Based on statistical results obtained under real-world pipeline conditions, an in-depth reliability analysis was conducted on girth welds in high-grade steel pipelines that contain crack defects, aiming to offer a more accurate theoretical foundation for the safety evaluation of pipelines.

Methods To overcome the conservative tendencies of traditional single-factor deterministic design and evaluation methods, a reliability analysis model was established for girth welds in high-grade steel pipelines exhibiting crack defects. Using the maximum likelihood estimation method, parameter estimations and Kolmogorov-Smirnov (K-S) tests were performed on the data of pipes and welding materials sourced from field sampling inspections on mechanical properties for the China−Russia East-route Natural Gas Pipeline, which helped identify input distributions. Considering the large strain loading conditions of the pipeline, an improved PRCI-CRES ultimate tensile strain prediction formula was chosen, aimed at accurately formulating the strain-based limit state equation. To enhance the efficiency of traditional Monte Carlo (MC) random sampling methods, an efficient sampling strategy based on Hamiltonian Monte Carlo-Subset Simulation (HMC-SS) was introduced, effectively optimizing the reliability analysis process. Additionally, the Sobol global sensitivity analysis method was employed to comprehensively evaluate the influence of each parameter of girth welds on structural failures, which allows for quantifying the sensitivity of the main influencing factors.

Results The results obtained using the HMC-SS method showed relative errors as low as 0.90％, while the calculation efficiency of the proposed model was improved by 99.70％. Under large strain conditions, the strength matching coefficient was identified as the most sensitive factor affecting failures in the girth weld area of high-grade steel pipelines, proving to be significantly more sensitive than other parameters. The total effect index, denoted as S_T, was 1.09, while the first-order influence index as S₁ was 0.951.

Conclusion The study findings underscore the importance of achieving equal strength or a high strength matching between welding materials and base metals during material processing. Based on the resulting second-order influence indices, the reliability-based parameter design effectively accounts for interactions among various parameters, thereby ensuring the safety of pipeline design from a global perspective.