LAUSR

Abstract Ships require regular inspection and maintenance for corrosion defects that appear due to the failure of protective coatings in corrosive environments. However, maintenance planning is inhibited by the fact that coating failures (i.e. arrival of corrosion defects) are random and physical models for coating degradation and subsequent corrosion progression require data that is usually unavailable. In this paper, the maintenance history of frigates in the Royal Australian Navy is used to estimate the parameters of a Non-Homogeneous Poisson Process that describes the statistical properties of the coating failures in specific compartments of a vessel. A grouping heuristic is developed for fleet-wide data aggregation and parameter estimation. Finally, the predictions from these models are used to optimise the inspection and maintenance plan to simultaneously minimise the defect backlog and the unavailability of the ship due to repairs (since detailed cost data is unavailable). A case study is presented using data from a real fleet of eight ships and results show that 1) the grouping can slightly improve the generalization of the compartment models and 2) the trade-off between the defect backlog and the ship unavailability is clearly displayed as the set of Pareto-optimal solutions of the bi-objective optimisation.