A data-driven risk assessment framework for ammonia pipelines: Insights from 232 accident cases (1970–2023)

作者：Lin Teng, Peidong Ling, Chunbo Zhao, Zengqiang Zhang, Bin Liu, J. Li, Pengbo Yin, Zhenchao Li, Lilong Jiang · 发表于：Journal of Pipeline Science and Engineering · 年份：2025 · DOI：10.1016/j.jpse.2025.100388 · 被引用次数：5 · 研究领域：Risk and Safety Analysis、Water Treatment and Disinfection

Ammonia pipelines are gaining renewed attention as key enablers of hydrogen energy systems, offering a practical solution to hydrogen transport and storage. However, current safety standards remain insufficient, as they often rely on risk models developed for natural gas or petroleum pipelines. This paper presents a novel, data-driven risk assessment methodology tailored specifically developed to establish scientifically justified safety distances for ammonia pipelines. A comprehensive dataset of 232 ammonia pipeline accidents spanning 1970 to 2023 was systematically collected and analyzed, revealing that material and welding defects account for 36.21% of failures. By applying Bayesian updating to failure data, individual and societal risk assessment models tailored explicitly for ammonia pipelines were developed. Risk assessment simulations indicate that Pipeline-1 (250 mm / 25 bar) reaches an individual risk level of 10 -6 at 300 m, while Pipeline-3(150 mm / 20 bar) shows similar risk grade at 190 m. In comparison, Pipeline-2(200 mm / 25 bar) and Pipeline-4(100 mm / 15 bar) reach 10 -7 at 325 m and 145 m, respectively. The proposed methodology delivers a scientifically grounded basis for determining safety distances, and offers actionable insights for improving ammonia pipeline design, permitting, and safety regulation.