Probabilistic fire risk analysis of process pipelines

Abstract

Pool fires are considered catastrophic events in hydrocarbon liquid process pipelines. There is much research performed on fire risk analysis of liquid pipelines; however, probabilistic-quantitative methods that are based on the leak size are yet to be further developed. This paper develops a framework for quantitative fire risk analysis in liquid process pipelines based on probabilistic analysis and computational fluid dynamics. The significance of leaked volume and its dependency on the leak size, here the pipeline failure frequency is distributed into the frequencies of different leak sizes using historical data and the Bayesian network. The frequency of fire occurrence is then determined by combining the ignition probability of each leakage rate, the flammability specification of released materials, and the frequency of each leak size. In the vulnerability assessment, the thermal dose received by individuals is modified to account for the possibility of escape. The developed framework is next employed in a gasoline pipeline with a 1 km length. Results show that the pool fire scenario resulting from the rupture of the pipeline, with a frequency of 1.18 × 10−7 (km−1. y−1) and an area of exposure of 5498.4 m2, is the maximum-credible scenario. Two separate escape routes with different distances to the fire center are defined. As well, two escape speeds of 4 and 6 m/s are considered. The results show that the initial radiation and the escape speed can significantly affect the sustained damage to individuals. This study is a step forward in dynamic-quantitative individual risk analysis in pool fires of hydrocarbon liquid pipelines.