Dynamic risk analysis of tank cleaning operations using bow-tie-based fuzzy Bayesian network

Abstract

This study aims to enhance maritime safety by focusing on risk analysis in tank cleaning operations, specifically targeting the risk of spillages. The Bow-Tie-based Fuzzy Bayesian Network (BT-FBN) methodology was employed to conduct an equipment-based dynamic risk analysis. A real chemical tanker was selected as the model vessel to ensure realistic calculations, and 'Loss of Containment' (LoC) was identified as the critical event for comprehensive risk analysis. The BT methodology was utilised to identify and systematically map potential hazards. Considering the temporal dynamics of risk, a Bayesian dynamic risk analysis was performed using four-month tank cleaning data from the model vessel, and fuzzy logic was applied to address uncertainties in the data. The findings indicated a steady increase in the risk of 'LoC' from 4.33% to 11.86% over four months, consistent with observed accident rates (8.57%). Furthermore, the likelihood of consequential events, such as 'Delay in Operations & Near Miss' (C-1), showed a similar trend, ranging from 3.69% to 10.09%. Notably, the analysis underscored the critical role of 'Human Error & Management Failure' (IE4) in contributing most significantly to 'LoC', ranging from 2.86% to 8.77%, emphasising the importance of addressing human factors to mitigate environmental pollution.