CHEMICAL ENGINEERING TRANSACTIONS
VOL. 77, 2019
A publication of
The Italian Association
of Chemical Engineering
Online at www.cetjournal.it
Guest Editors: Genserik Reniers, Bruno Fabiano
Copyright © 2019, AIDIC Servizi S.r.l.
I SBN 978-88-95608-74-7; I SSN 2283-9216
A Dynamic Approach to Fault Tree Analysis based on
Bayesian Beliefs Networks
Tomaso Vairo
a
*, Maria Francesca Milazzo
b
, Paolo Bragatto
c
, Bruno Fabiano
d
a
ARPAL – Grandi Rischi, via Bombrini 8, 16149 Genoa, Italy
b
Dip.Inge. – Department of Engineering, University of Messina, Contrada di Dio, 98166 Messina, Italy
c
INAIL – Technological Innovation Department, Research Centre, via Fontana Candida, 1, 00078 Monteporzio, Italy
d
DICCA – Civil, Chemical and Environmental Engineering Dept., Genoa University, via Opera Pia 15, 16145 Genoa, Italy
tomaso.vairo@arpal.gov.il
According to the Seveso Directives, the risk assessment is crucial for an effective control of major accident
hazard. Nevertheless, the complexity of many Seveso sites, due to human, technical and organizational
factors makes recognized common practices limited because of their intrinsic static nature. In this paper, a
dynamic approach for risk assessment is proposed, which allows evaluating moment by moment the state of
the system under analysis by Bayesian belief networks. A petrochemical coastal storage was selected as
applicative case-study to verify the capability of the dynamic approach. Network training is performed by
entering historical reliability data, near-miss and accidents data series collected on-site by periodical
inspection plans on critical elements, as well as from the evidences of SMS reports. Upon proper refinement
and further validation with reliable field data, the predictive approach may be used as a management decision-
making tool.
1. Introduction
As many other contexts, risk analysis has been strongly affected by the big data era, that means by the ability
to provide continuous acquisition, effective process and meaningful communication of such information. This
amount of data is particularly interesting in managing risk and asset integrity of process plants. Research in
this context is faced with the dilemma that, while there have been significant developments in understanding
how accidents occur, there has been no comparable development in understanding how to adequately assess
and reduce risks (Bouloiz et al., 2013), considering both process and personnel side of safety (Fabiano et al.,
1995). In process safety and risk management area, a need for an integrated and holistic system approach
has been explicitly described and advanced research trends include knowledge-based methods combined
with process models, such as Petri nets, signed digraphs, and dynamic simulation (Jain et al., 2018). The
application of BN in the field of risk and reliability was explored by many researchers, e.g. Abbasi et al., 2016.
A system is safe if it is impervious and resilient to perturbations, thus the identification and assessment of
relevant hazards is an essential prerequisite for system safety. Nevertheless, traditional methods for risk
assessment do not take into account interactions between system components and do not adequately
address human and organizational factors, thus being not appropriate for complex systems (Leveson, 2004).
Some efforts have been made to include human and organizational factors (Milazzo et al., 2010), while few
works attempt to integrate organizational and human factors in a dynamic approach. As examples based on
Bayesian theory, Kalantarnia et al. (2009), proposed a method for dynamic safety management and Meel &
Seider (2006) estimated the dynamic probabilities of accident sequences having different severity levels by
using statistical analyses of near-miss and incidents. In this work, a dynamic approach for risk assessment,
based on the evaluation of the state of the system under analysis, is outlined to be applied for those cases
when a static assessment method is not trustable. The Bayesian networks are constructed from Fault Trees
Analyses (FTA) and failure rates represents a priori probabilities. The modelling provides a set of independent
nodes (root elements of FTA, i.e. critical items) and intermediate events for the top event. The network training
is performed by using historical reliability data, near-miss and accidents data collected on-site by periodical
DOI: 10.3303/CET1977139
Paper Received: 14 November 2018; Revised: 17 May 2019; Accepted: 12 July 2019
Please cite this article as: Vairo T., Milazzo M., Bragatto P., Fabiano B., 2019, A Dynamic Approach to Fault Tree Analysis based on Bayesian
Beliefs Networks, Chemical Engineering Transactions, 77, 829-834 DOI:10.3303/CET1977139
829