An application of BN to envisage potential accidents in FLNG facility

Md. A. Baksh, Rouzbeh Abbassi, Vikram Garaniya, Faisal Khan

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionpeer-review

2 Citations (Scopus)


The majority of existing fire and explosion consequence models currently used in the offshore LNG process facilities focus on modelling individual accidental hazard. However, such undesired events can occur simultaneously leading to multiple consequences. This study focuses on developing a model-based Bayesian network (BN) approach to envisage the most probable accidental scenario in complex offshore process facilities by considering the evidence of primary causes. The proposed methodology comprises of evolving scenario identification, accident consequence framework development, accident scenario likelihood estimation, and ranking of the scenarios. The approach was tested and demonstrated with a case study for an offshore LNG process facility (Floating LNG). The developed methodology using the BN approach was found to be effective in determining the most credible accidental scenario.
Original languageEnglish
Title of host publicationProceedings of the Twelfth (2016) Pacific-Asia Offshore Mechanics Symposium
Place of PublicationCalifornia
PublisherInternational Society of Offshore and Polar Engineers
Number of pages8
ISBN (Electronic)9781880653982
Publication statusPublished - 2016
Externally publishedYes
EventISOPE Pacific/Asia Offshore Mechanics Symposium 2016 (12th : 2016) - Gold Coast, Australia
Duration: 4 Oct 20167 Oct 2016

Publication series

ISSN (Electronic)1946-004X


ConferenceISOPE Pacific/Asia Offshore Mechanics Symposium 2016 (12th : 2016)
Abbreviated titlePACOMS-2016
CityGold Coast


  • Bayesian network
  • Consequence assessment
  • Fire and explosion modelling
  • Scenario-based modelling
  • Offshore system


Dive into the research topics of 'An application of BN to envisage potential accidents in FLNG facility'. Together they form a unique fingerprint.

Cite this