Digital twins move from design tool to class-assured asset: condition monitoring gains an independent stamp

For years a ship’s digital twin was mainly a design artefact — a virtual model built during engineering and then largely shelved. That is changing as classification societies build the governance and assurance frameworks that let the same models be trusted in service, for condition monitoring and maintenance planning.

A recent marker came in May 2026, when Lloyd’s Register granted an Approval in Principle (AiP) to Spanish shipbuilder Navantia for an AI-based “Digital Twin Ready” framework. Notably, the assessment focused less on the twin itself and more on the discipline around it: development processes, data governance, model validation and lifecycle management. The reference case was an air-cooling system digital twin that uses data-driven algorithms and machine-learning models to analyse sensor data, monitor system behaviour and flag early signs of abnormal performance. As LR’s naval business director Javier de Juana framed it, “As digital twins move from development to operational deployment, robust governance and independent assurance are becoming increasingly critical.” Navantia’s CTO Ana Moya said the assessment “validates the maturity” of the framework, “particularly in areas such as data governance, model validation and lifecycle management.”

DNV has been pushing the same logic on the commercial side, pairing a hull condition-monitoring digital twin with wave, position and sensor data. The combination, DNV argues, makes sensor-based hull-condition predictions “much more meaningful and accurate,” reaching fatigue-critical details — stiffener connections, hatch-coaming terminations — that no sensor directly covers, and helping to optimise instrumentation and survey planning under its alternative survey arrangement rules. DNV also flags the underlying opportunity bluntly: historically, “as little as 1% of the data available on board has been used for shore-side decision support on average.”

The common thread is assurance. A digital twin only earns a place in maintenance and survey decisions once someone independent can vouch for the governance behind it — how the data is validated, how the model is maintained over the asset’s life. That is the bar class societies are now setting, and it is the same bar that separates a demonstration from an operational tool.

Apeks view — The interesting shift is not the 3D model; it is what class societies are actually approving — the governance around it. A twin is only as trustworthy as the validation, data quality and lifecycle discipline behind it, which is exactly why the assurance stamp matters more than the visualisation. For owners, the practical lesson is that a model’s value is bounded by the integrity of the record feeding it: a well-kept condition history is what lets a digital twin — and eventually a surveyor — trust what it sees.