Bayesian network for risk-informed inspection planning in ships

We present the development of a Bayesian Network (BN) model for risk assessment of ships to facilitate risk-informed inspection planning. The BN consists of four sub-modules: cause, vulnerability, consequence and inspection. These modules are set up in a generic manner by consideration of readily available information (the indicators), such as ship type, ship size, operational conditions. On this basis, we define representative models of the ship and its components, which are parameterized by the indicators and quantities that can be related to the indicators. The generic approach enables a fast and holistic risk assessment of a larger number of ships even if only limited information is available. The BN approach facilitates the straightforward inclusion of new information, e.g. from inspection and monitoring. It has the additional advantage of allowing easier communication of the model to non-experts. Through numerical applications, we demonstrate how the calculated risks can be used to prioritize inspections and how new information can be incorporated to update risk estimates.