Large-scale offshore installation projects operate in a dynamic environment where logistics, weather, asset coordination, insurance policies, and safety interact constantly. Every vessel departure, crane lift movement, equipment availability, and weather window matters. Even the appearance of marine mammals within the vicinity, to protect them from noise, is part of the equation. With daily operational costs reaching multi-million euros a day, even small inefficiencies have a significant financial impact.

Our client, a leading global services provider operating in the dredging, maritime infrastructure and maritime services sectors, is offering offshore monopile placement for windmill parks. Currently, the Dutch provider relies on information found in emails, documents, and reports provided over a long period of time. They now intend to leverage real-time streams of equipment telemetry, vessel movement signals, and environmental data. Turning these streams into actionable intelligence in real time is not just a convenience, it has become a technical necessity to ensure safe, efficient, and predictable execution of their offshore projects. Fully recognising SDG Group, an ALTEN company, for its strong track record in blending IoT capabilities with real-time intelligence, the provider enlisted their support in this digital transformation.

Digital Twin meets the Offshore Environment 

SDG, implemented as a division of ALTEN in the Netherlands, is currently building a semantic digital twin, using both Azure Digital Twins and Microsoft Fabric, that represents the offshore environment at a level where raw operational data becomes contextualised insights. The first project phase focuses on constructing a live digital twin graph of locations, vessels, cranes, monopiles, and installation stages. Next to storing raw telemetry, a digital twin understands what each signal means in the real world and the relationships it has with other digital twins.

Let’s take crane activity as an example. Rather than evaluating signals in isolation, the system identifies characteristic patterns in how multiple variables evolve over time. When a specific combination of thresholds, directional changes, and temporal relationships appears, the platform interprets this as the onset of a particular operational state, and a mirrored set of changes indicates that the state is concluding. An example is a brief rise across several correlated inputs, followed by a consistent shift in their trajectory marks the start of an activity. By correlating these signals, the platform identifies the activity as equipment being returned to storage. This allows project teams to see operational status in real time rather than waiting for manual reporting that happens at the end of a shift or day.

The same approach applies to vessel movement. Automatic Information System (AIS) data, like location, heading, and speed of a vessel, allows the digital twin to determine vessel states automatically. In the same way, when an entity’s behaviour matches a recognisable pattern within a defined context, such as changes in relative position, velocity, or stability, the digital twin infers a transition into a different state class. When the pattern shifts again in a coordinated manner, the system updates the state accordingly. An example is a phase of steady behaviour followed by a clear directional change signals movement into a new operational mode. These are small examples, but they illustrate how semantic modelling transforms telemetry into useful operational knowledge in real time.

Working with geolocations in geofencing applications presents several challenges. While the positions of docks, harbours, and marshalling sites remain fixed, geofence buffers must accommodate variations such as GPS dilution of precision, intermittent signal updates, and alternative vessel manoeuvres. The ability to visualise both geofences and vessel positions significantly simplifies testing-related requirements. Using the recently introduced Microsoft Fabric Maps, we can visualise GeoJSON shapes, like LineStrings and Polygons (see Figure 1). Using layered visualisations, geofence-related requirements can then be translated into effective business rules within the digital twin environment.

Time of Completion & Simulation 

This foundation, based on a digital twin, unlocks layers of intelligence. The next development phase introduces data-driven insights into projected completion time, based on parameters such as the current pace of installation, asset availability and location, and operational conditions. Instead of a static forecast, now the system can continuously evaluate live progress and then compute the expected completion time given the real installation speed observed in the field.

From there, probabilistic forecasting and simulation capabilities will expand this further. For example, if weather models indicate a storm approaching in three days, the platform will estimate the impact on installation tempo, quantify schedule exposure, and surface the most critical bottlenecks such as crane utilisation or vessel sequencing. It can also calculate the new expected result at the time of completion when you add or remove a vessel from the project, making it possible to optimise vessel usage across projects. The result is not a single projected finish date, but a dynamic window with confidence levels and transparent drivers of variance, enabling earlier and more informed intervention.

Finally, a conversational insight engine, powered by large language models, will enable agentic interaction through natural language. A user may ask: “Why is the installation progress slower this week compared to last month?”. The system could respond with a breakdown: e.g., “limited lifting hours due to wind conditions”, “increased transit time because of sea state”, or “unexpected maintenance on a support vessel”. It may even propose mitigation actions supported by historical patterns.

The result is a continuous intelligence loop based on the here-and-now of the digital twin, combined with historical updates over time. Real-time understanding of operations feeds predictive planning, which supports proactive decision making at scale. Even a single day saved protects more than one million euros, illustrating the magnitude of value that this platform can unlock.

We are at the start of this journey. Building the digital twin model today enables simulation and AI-driven insights tomorrow, and, ultimately, a future where offshore execution benefits from data-guided operational excellence.