Integrating QRA and SRA Methods Within a Bayesian Framework When Calculating Risk in Marine Operations: Two Examples

Wenche K. Rettedal, Terje Aven, Ove T. Gudmestad

Journal of Offshore Mechanics and Arctic Engineering · 2000 · 14 citations · 4 references

Concepts

TL;DR

Integrating quantitative risk analysis with structural reliability analysis in marine operations is best approached within a Bayesian framework, though the precise Bayesian interpretation varies across methods. The paper presents two marine operation examples that compare a classical Bayesian approach, which estimates objective risk, with a fully Bayesian approach, which treats risk as uncertainty about future observations. Both examples employ a fault‑tree model that integrates structural reliability analysis to assess an initial accidental event under two Bayesian frameworks. The study concludes that adopting the fully Bayesian approach is the most suitable framework for integrating SRA and QRA.

Abstract

This paper concerns itself with the integration of QRA (quantitative risk analysis) and SRA (structural reliability analysis) methods. For simplicity, we will use the term SRA instead of SRA methods in the paper. The Bayesian (subjective) approach seems to be the most appropriate framework for such integrated analyses. It may, however, not be clear to all what the Bayesian approach really means. There exists alternative Bayesian approaches, and the integration of SRA and QRA is very much dependent on what the basis is. The purpose of this paper is to present two marine operation examples, implementing two different Bayesian approaches: the “classical Bayesian approach” and the “fully Bayesian approach.” Following the classical Bayesian approach, we estimate a true, objective risk, whereas in the fully Bayesian approach, risk is a way of expressing uncertainty about future observable quantities. In both examples, one initial accidental event is investigated by using a fault tree and by integrating SRA into this fault tree. We conclude that the most suitable framework for integrating SRA and QRA is to adopt the “fully Bayesian approach.” [S0892-7219(00)00703-2]

References

4