Bayesian networks for probabilistic modelling of still water bending moment for side-damaged tankers

Abstract

In this paper, a probabilistic model based on a Bayesian Network is presented for the still water bending moment acting on a tanker struck in a collision. The damage is assumed to occur along one side of the ship and is idealised by means of a rectangular hole leading to flooding of the breached compartments. The stochastic characterisation of the position and extension of the hole along the ship is derived according to two different probabilistic models found in literature. The results are not only compared in the light of the different features of the input probabilistic models but are also interpreted in terms of a ranking of the various situations according to their probability of occurrence, the magnitude of the generated load effect and the spatial correspondence between the maximum bending moment and the damaged area. The final goal is to provide information for the selection of accident scenarios and design values for this variable.

Keywords:

• accident scenarios
• collision
• probabilistic damage assessment
• Bayesian Networks