Dynamic predictive reliability assessment of ship systems

Recent research shows that maritime industry has adopted innovative and sophisticated inspection and maintenance practices. A flexible framework, applicable on complex machinery, is introduced towards ship maintenance. A holistic inspection and maintenance notion is implemented, introducing different strategies, methodologies, and tools, suitably selected, for each required ship system. The proposed framework enables predictive reliability assessment of ship machinery, while scheduling maintenance actions by enhancing safety and systems' availability. This paper presents the Probabilistic Machinery Reliability Assessment (PMRA) strategy, which achieves predictive reliability assessment and evaluation of different complex ship systems. The assessment takes place on system, subsystem and component level, while allowing data fusion of different data types from various input sources. In this respect, the combination of data mining method (k-means), manufacturers' alarm levels, dynamic state modelling (Markov Chains), probabilistic predictive reliability assessment (Dynamic Bayesian Belief Networks) and qualitative decision making (Failure Modes and Effects Analysis) is suggested. PMRA has been clearly demonstrated in a case study on selected ship machinery. The results identify the most unreliability systems, subsystems and components, while advising practical maintenance activities. The proposed PMRA strategy is also tested in a flexible sensitivity analysis scheme.