There are two types of failures that occur on naval battleships. One is normal failure. This is not fatal as it can still carry out its mission if it does occur. On the other hand, critical failure is fatal because it must enter the home port immediately upon occurrence. Unlike normal failure, critical failure occurs with an extremely small probability and has the characteristics of a rare event. The critical failure data of battleships is too small to estimate the distribution. For this reason, in previous studies, only the constant expected value of critical failure was estimated. The variance of failure phenomena is usually very large. That is, a failure cannot be represented by a single constant expected value. Therefore, in this study, the variance of major failures by age of battleships was estimated over the entire lifespan of battleships. Investigate the relationship between critical failure and normal failure, and use this to estimate the population distribution of critical failure. The results of this study warn of the need to improve the Navy's maintenance capabilities. The novelty of this study is that it presented a method for estimating the population of rare events.

Digital twin (DT) has been around for many years, but there is no widely accepted standardized tool or method. In this study, system dynamics was proposed as a tool that can be integrated into multi-scale, multi-physics, and multi-disciplinary, which are continuously becoming issues in the DT field. Various heterogeneous data from multiple protocols or platforms could be integrated into one model. Through the five-step model building process, it was possible to integrate the theories and various models studied in the past. In this study, the operation and maintenance system of ROK Naval ships is implemented as a proposed method. Various physics, scales, and disciplines such as failures of ships, maintenance ability of repair shops and schedule pressure of mechanics were reflected. It was possible to observe non-intuitive correlations and potential problems caused by the latent effect of the high-fidelity DT model. The proposed method is also capable of updating through continuous data calibration or real-time interworking with external statistical analysis tools.

Predicting equipment failure is important because it could improve availability and cut down the operating budget. Previous literature has attempted to model failure rate with bathtub-formed function, Weibull distribution, Bayesian network, or AHP. But these models perform well with a sufficient amount of data and could not incorporate the two salient characteristics; unbalanced category and sharing structure. Hierarchical model has the advantage of partial pooling. The proposed model is based on Bayesian hierarchical B-spline. Time series of the failure rate of 98 Republic of Korea Naval ships have been modeled as hierarchical model, where each layer corresponds to ship engine, Engine type, and Engine archetype. As a result of the analysis, the suggested model predicted the failure rate of an entire lifetime accurately in multiple situational conditions, including the amount of prior knowledge of the engine.