Probabilistic meta-models evaluating accidental oil spill size from tankers

Abstract

In this paper, we introduce two probabilistic meta-models evaluating accidental oil spill volume for two types of tanker: VLCC and Aframax and two types of accident: collision and grounding. The metamodels utilize two engineering models, one for ship damage estimation due to an accident, another for evaluation of an oil spill size given the damage extent. Both models are coupled and executed for a large number of accidental scenarios, thus two extensive datasets are developed, one for collision and one for grounding accident. Subsequently, the datasets are used to develop two casual, probabilistic meta-models by applying the Bayesian Networks. The obtained results are compared with the real-life cases, and good agreement is found.