A Bayesian Network Model for Accidental Oil Outflow in Double Hull Oil Product Tanker Collisions

This paper proposes a Bayesian belief network (BBN) model for the estimation of accidental oil outflow in a ship-ship collision where a product tanker is struck. The intended application area for this model is maritime traffic risk assessment, i.e. in a setting in which the uncertainty regarding the specific vessel characteristics is high. The BBN combines a model for linking relevant variables of the impact scenario to the damage extent with a model for estimating the tank layouts based on limited information regarding the ship, as typically available from data from the Automatic Information System (AIS). The damage extent model, formulated as a logistic regression model and based on a mechanical engineering model for the coupled inner-outer dynamics problem of two colliding ships, is implemented in a discretized version in the BBN. The model for estimating the tank layout is applied for a representative set of product tankers typically operating in the Baltic Sea area. The methodology for constructing the BBN is discussed and results are shown.