Risk assessment of fire accidents in chemical and hydrocarbon processing industry

Yang, Ruochen (2019) Risk assessment of fire accidents in chemical and hydrocarbon processing industry. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (9MB)

Abstract

Fire disasters are among the most dangerous accidents in the chemical and hydrocarbon processing industry. Fires have been the source of major accidents such as the Piper Alpha disaster (1976), the BP Texas City disaster (2005), the Buncefield oil depot fire (2005), Puerto Rico’s fire accident (2009), and the Jaipur fire accident (2009). The catastrophic impact of fire accidents necessitates a detailed understanding of the mechanisms of their occurrence and evolution in a complex engineering system. Detailed understanding will help develop fire prevention and control strategies. This thesis aims to provide a detailed understanding of fire risk in the hydrocarbon production and processing industry. In order to realize this objective, the work presented in the thesis includes three parts: i) Developing a procedure to study potential fire accident scenarios in an offshore facility with different ignition source locations. This procedure helps to design safety measures. The effectiveness of safety measures is verified using a computational fluid dynamics (CFD) code. This work emphasizes that an FLNG layout must be considered with the utmost care since it is the most effective measure in limiting a potential LNG release and subsequent dispersion effect, and directly influences the fire dynamics and thus limits the potential damage. ii) An integrated probabilistic model for fire accident analysis considering the time-dependent nature of the fire is developed. The developed model captures the dynamics of fire evolution using three distinct techniques Bayesian networks, Petri Nets, and a CFD model. The Bayesian network captures the logical dependence of fire causation factors. The Petri Net captures the time-dependent evolution of a fire scenario. The CFD model captures the dimension and impact of the fire accident scenario. The results in this work show that a time-dependent probability analysis model is necessary for fire accidents. iii) Whether fire alone can cause a domino effect is demystified in the last work. A solid-flame model is used in a CFD framework to calculate the escalation vector for a domino effect; escalation probability is assessed using a probit model. The results demonstrate that a pool fire alone sometimes may not cause a domino effect in the current industry. It is other factors, such as explosion and hydrocarbon leakage, work together with a pool fire to escalate into a domino event, for example, the results shown in the case study of the Jaipur fire accident.

Actions (login required)

View Item