氢
管道运输
巴黎法
管道(软件)
工程类
结构工程
使用寿命
诚信管理
压力(语言学)
断裂力学
服务(商务)
天然气
韧性
法律工程学
天然气储存
断裂韧性
核工程
可靠性工程
强度因子
基线(sea)
机械工程
结构完整性
材料科学
压力容器
故障评估
混乱
范围(计算机科学)
作者
Luyao Xu,Shane R. Finneran
出处
期刊:
日期:2026-02-09
卷期号:: 1-19
摘要
Abstract Repurposing existing natural gas pipelines for the transport of hydrogen or hydrogen-natural gas blends offers a cost-effective and scalable pathway to support decarbonization goals. However, pre-existing crack-like defects in these pipelines pose significant integrity challenges in hydrogen service due to hydrogen embrittlement, which degrades fracture toughness and accelerates fatigue crack growth. This paper evaluates fatigue-driven crack propagation in hydrogen pipelines using multiple fatigue crack growth rate (FCGR) models, including API 579-Air, ASME B31.12, ASME Code Case 220 (CC220), and the DNV Lab Testing model. API 579-Air serves as a baseline under inert conditions. ASME B31.12 provides a conservative semi-empirical model, while ASME CC220 introduces factors for hydrogen partial pressure and stress intensity ratio, enabling more realistic assessment. The DNV Lab Testing model is developed based on experimental data obtained from specific hydrogen environments, offering a more representative view of crack growth behavior under actual service conditions. Fatigue life predictions are conducted under varied operating scenarios, accounting for differences in hydrogen partial pressure, cyclic loading, and flaw size. Results reveal notable variability in fatigue life calculations across models, indicating the impact of model selection on integrity evaluations. This comparative study presents a structured methodology to guide integrity engineers in choosing appropriate FCGR models for hydrogen pipeline applications. The findings contribute to more robust hydrogen pipeline integrity management and fitness-for-service assessment by improving fatigue life prediction accuracy, supporting the safe and reliable conversion of existing NG infrastructure for hydrogen service.
科研通智能强力驱动
Strongly Powered by AbleSci AI