Structural risk analysis model of damaged membrane LNG carriers after grounding based on Bayesian belief networks

Abstract

Risk assessment is regarded as an effective tool for the design of ship structures, based on the safety level approach (SLA). However, there are gaps between the theoretical realisation of the approach and its practical application, such as the lack of specific risk analysis tools for different structural failure modes. Ship structures have various failure modes under different hazards (e.g., ultimate limit state, accidental limit state, and fatigue limit state). In an accidental limit state, structural failure is generally a casual process from local damage to overall hull collapse, and is affected by structural uncertainty factors and accidental random impacts. In this paper, a structural reliability analysis (SRA) model based on a Bayesian belief network (BBN) is proposed for the hull girder collapse risk after accidents. In this model, a BBN is used to represent the random states of variable risk events after accidents, as well as the dependencies between events; and a SRA is used to evaluate the failure probability of hull girders for each possible accident condition.The hull girder collapse risk of a membrane liquefied natural gas (LNG) carrier after grounding is analysed using the BBN-SRA model. Compared with the conventional methods, the risk level obtained is more reliable, given that different possible accident conditions are considered using the new model. The extreme accidental condition of the LNG leak, and subsequent cryogenic impacts on the structural strength are also considered using the nonlinear finite element analysis (FEA) method. In addition, the inverse inference and information updates of the BBN are confirmed to be useful as supports for decision making in risk-based designs and emergency rescue processes.