Probabilistic Risk Assessment of Offshore Production Platform by Bayesian Network Application to HAZOP and Bow-tie Studies
Abstract
Successful risk management in an offshore oil and gas production platform requires accurate and up-to-date probability of occurrence of process safety events. Traditional hazard identification and risk assessment techniques such as HAZOP and Bow-tie analysis are the well-accepted methods in the oil and gas industry. However, these methods cannot effectively cope with dynamic operating environments, which continuously affect the estimated probabilities and risks. Factors such as variations in operating conditions, equipment deterioration, and personnel competency affect the safety barriers’ performances and consequently alter the probability of occurrence of the process safety events. In the past decade, Bayesian network has gained significant attention in the process safety area because of its ability to include new information. It has been integrated with various traditional risk assessment techniques, including HAZOP and Bow-tie studies, extending their capabilities to consider operational variations and revise the probability of occurrence of the process safety events.
This research applies Bayesian network to HAZOP and Bow-tie studies for a loss of primary containment of high pressure hydrocarbon gas from an export gas compressor system. Eleven process safety indicators such as loss of primary containment and maintenance backlog are integrated into the models to reflect changes in safety barriers’ performances. The integration process is realized by aggregating multiple specific indicators into three element indicators, which are mechanical integrity, operational integrity, and personnel integrity.
The updated probabilities from the developed HAZOP-BN and Bow-tie-BN models are considerably different due to dissimilarity in hazard identification approaches and the BN development process. However, both models provide consistent results with respect to the degree of effect from indicators’ input. Therefore, it is prudent for a company to implement the HAZOP-BN or Bow-tie-BN and use process safety indicator data to improve its risk assessment capability. Overtime, probability estimation can be perfected through integration of additional information with the systematic risk assessment method.
Subject
Bayesian networkprobabilistic risk assessment
offshore
oil and gas
HAZOP
Bow-tie
process safety indicator
Collections
Citation
Chaiwat, Pakorn (2016). Probabilistic Risk Assessment of Offshore Production Platform by Bayesian Network Application to HAZOP and Bow-tie Studies. Master's thesis, Texas A & M University. Available electronically from https : / /hdl .handle .net /1969 .1 /158987.