Dynamic safety risk modeling of process systems using bayesian network

Abstract

Process complex systems in particular oil and gas plants due to dealing with hazardous materials at severe process conditions are much prone to catastrophic accidents. In this context, safety risk analysis is a crucial tool to develop effective strategies to prevent accident and provide mitigative measures. Dynamic risk analysis (DRA) is one of the most practical approaches for risk analysis that helps provide safer operations of complex process systems. The present work is aimed at demonstrating the application of an integrated DRA approach to comprehensive quantitative modeling and analysis of the both aspects of risk, that is, probability and consequence assessments. In this approach, first, the worst case scenario is identified and then a robust tool is developed for dynamic accident scenario modeling and risk assessment by means of Bayesian Network. This approach is applied to risk analysis of a flammable liquid storage system at a gas refinery. The work provides valuable information on the identification and comprehensive analysis of worst case accident scenarios, their main consequences, critical basic events, and minimal cut sets which lead to accident scenarios and also for dynamic updating of probabilities and risk. The obtained results are more appropriate and rigorous to developing preventive and mitigative strategies for potential accident scenarios and thus increase the safety level in the complex process systems. © 2017 American Institute of Chemical Engineers Process Saf Prog 36: 399–407, 2017

Quantitative risk assessment of a natural gas pipeline in an underground utility tunnel

“…To establish a comprehensive risk assessment framework for gas pipeline accidents in a utility tunnel, the present article used a BT diagram to identify potential hazards and possible accident scenarios and applied a BN for dynamic quantitative risk analysis of gas pipeline accidents in a utility tunnel. Compared with conventional risk analysis methods, BN has proven to be effective for capturing and integrating qualitative and quantitative information from various sources and can facilitate accident scenario modeling with multistate variables [13,21,22], in particular for dynamic risk analysis [23][24][25][26]. The BN for the gas pipeline accident in a utility tunnel is transferred from the BT diagram through a developed mapping diagram.…”

Footprint of knowledge acquisition improvement in failure diagnosis analysis

“…BN analysis may be qualitative, quantitative, or both, depending on the scope of the analysis as well as FTA, and it is popular in statistics, machine learning, artificial intelligence, and risk and reliability analyses. Additionally, BN has also been widely applied in numerous risk and reliability studies, improving the safety performance of a system, updating failure probability, mapping static or dynamic FTs into corresponding BNs, and in recent work combining BN and petri‐nets aimed to analyze dynamic safety system …”

“…BN analysis may be qualitative, quantitative, or both, depending on the scope of the analysis as well as FTA, and it is popular in statistics, machine learning, artificial intelligence, and risk and reliability analyses. Additionally, BN has also been widely applied in numerous risk and reliability studies, [47][48][49][50] improving the safety performance of a system, 51 updating failure probability, [52][53][54] mapping static or dynamic FTs into corresponding BNs, [55][56][57] and in recent work combining BN and petri-nets aimed to analyze dynamic safety system. 58 The main purpose of the current study is providing novel framework in order to improve knowledge acquisition for analyzing fault diagnosis in a FT and comparing the obtained results with listing of available approaches.…”

Knowledge acquisition development in failure diagnosis analysis as an interactive approach

“…BN analysis may be qualitative, quantitative, or both, depending on the scope of the analysis as well as FTA and it is popular in statistics, machine learning, artificial intelligence, and risk and reliability analyses [42]. Additionally, BN has also been widely applied in numerous risk and reliability studies [43][44][45][46], improving the safety performance of a system [47][48][49][50], updating failure probability [51][52][53], mapping static or dynamic FTs into corresponding BNs [54][55][56][57][58], and in recent work combining BN and petri nets aimed to analyze dynamic safety system [59].…”