How to address model uncertainty in the escalation of domino effects?

Journal article (2018)

Authors

N. Khakzad Rostami Safety and Security Science - TPM
Paul Amyotte Dalhousie University
Valerio Cozzani University of Bologna
G.L.L.M.E. Reniers Safety and Security Science - TPM
Hans Pasman Texas A&M University
Research Group
Safety and Security Science (TPM) (TU Delft)
Graph theory Model uncertainty Domino effect Oil terminal Dynamic Bayesian network
More Info
expand_more

Published Date

2018

Language

English

Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Faculty

Technology, Policy and Management

Department

Values Technology and Innovation

Research Group

Safety and Security Science

Abstract

Modeling potential domino scenarios in process plants includes the prediction of the most probable sequence of events and the calculation of respective probabilities, so-called escalation probabilities, so that appropriate prevention and mitigation safety measures can be devised. Domino effect modeling, however, is very challenging mainly due to uncertainties involved in estimation of escalation probabilities (parameter uncertainty) and prediction of the sequence of events during a domino effect (model uncertainty). In the present study, a methodology based on dynamic Bayesian network is developed for identification of the most likely sequence of events in domino scenarios while accounting for model uncertainty. Verifying the accuracy of the methodology based on a comparison with previous studies, the methodology is applied to model single-primary-event and multiple-primary-event domino scenarios in process plants.