Digital Twins for Predictive Maintenance in Maritime Engine-Room Systems

Within the current context and with the changes that have taken place in recent years, the maritime sector is facing a set of transformations that may lead to the adoption of complex cyber-physical systems. All these developments can help us achieve high reliability in the operation of the vessel’s critical elements, both in the power generation plant and in the safety of navigation systems, as well as in energy efficiency and environmental sustainability (2). Until now, traditional maintenance models (corrective, predictive and static predictive) have been our main resource. These models, however, face several limitations (5). They require personnel with very high levels of experience on the systems that are being operated, that personnel are scarce, and generational turnover can create bottlenecks on finding well trained personnel. Additional challenges include increasing technological complexity, and the growing demand for continuous availability on the operation of the ships with minimal downtime. The aim of this work is to explore, from both a technical and conceptual perspective and through practical tests, the application of digital twins as more advanced tools for maintenance management. The study seeks to examine and assess how a virtual replica, bidirectionally synchronized with the physical system, embedded in its calculations and fed by sufficiently rich information flows, can provide a comprehensive view of the vessel and, in turn, send accurate data back to the installation about its operational state and needs, adapting the plant’s maintenance plans accordingly. The objective is therefore to find the requirements and demonstrate the capabilities of this technology, to examine its applicability, and to study the improvements and regulations currently being developed in this field, including the regulatory framework governing the secure management of such data. The methodology follows an exploratory and verification-oriented approach, combining a review of the technical literature and up-to-date information on the state of the art in this domain with the development of functional digital architectures on self-made representations of engine rooms and models taken. These architectures integrate data acquisition, simulation of outcomes, dynamic modelling, and predictive analysis based on machine learning, together with the use of realistic physical systems. Preliminary analyses suggest that implementing a modular architecture makes it possible to supervise the installation at different levels of complexity. Moreover, these models are already showing the ability to detect minor deviations in machinery performance. This should make it possible to identify the root causes of failures that will occur in the installation if no countermeasures are taken, well before they become irreversible, to estimate the remaining time before such failures, and to support decisions that prevent unplanned shutdowns. However, depending on the purpose, the necessary data and calculations may be too time-consuming or too expensive, so it is advisable to classify systems according to their capabilities and accuracy to facilitate their correct application. The results obtained and the data analysed provide indications of the viability of this approach, being capable of detecting RUL with MAE reduction of 77,56% (6), although they also show that it significantly increases the complexity of the installation. It is, however, a technology rooted in solid foundations, such as the physical twins historically used by NASA. The findings likewise reveal major challenges, particularly in managing the large volumes of data generated on board and in addressing potential vulnerabilities to cyberattacks on these data. It has been observed that the effectiveness of these tools is closely tied to the correct modelling of the twin and to the availability of high-quality data that are properly labelled and segmented and there are no strong evidences yet that they can reduce the cost of entire fleets. The technology is promising, but its implementation at plant level entails a degree of complexity that calls for progressive development and rigorous validation across multiple operational scenarios.