Dynamic risk assessment of storage tank using consequence modeling and fuzzy Bayesian network

Abstract

Accidents in process industries cause irreparable economic, human, financial and environmental losses annually. Accident assessment and analysis using modern risk assessment methods is a necessity for preventing these accidents. This study was conducted with the aim of Dynamic risk assessment of tank storage using modern methods and comparing them with traditional method. In this study, bow tie (BT) method was used to analyze the Leakage event and its consequences and model the cause of the outcome, and the Bayesian network method was used to update the probability rate of the consequences. Then, four release scenarios were used. Possible selection and release outcome were modeled using version 5.4 of ALOHA software. Finally, according to the degree of reproducibility of possible consequences and risk number modeling for the four scenarios were estimated. The results of modeling the cause and effect showed that 50 Basic events are effective in chemical leakage and Pool fire is the most probable consequence due to chemical leakage in both BT and Bayesian network (BN) models. Also, the modeling results showed that Leakage 50 mm diameter has the highest Emission rate (80 kg/min) and Leakage of 1 mm have the lowest emission rate. The results of risk assessment showed that the estimated risk number in both models is in the unacceptable range. In this study, an integrated approach including BT, Fuzzy Bayesian networks and consequence modeling was used to estimate the risk in tank storage. The use of these three approaches makes the results of risk assessment more objective than conventional methods. The results of outcome modeling can be used as a guide in adopting accident prevention and emergency preparedness approaches

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 21 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.