Page 38 - 2024 - Q3 - Minerva in Focus

MINERVA IN FOCUS Marine Technology

Data preparation and harmonisation Advanced analysis for efficiency
For a data-driven TVA approach to be effective, the data itself must and safety
be carefully prepared and harmonised. Shipbuilding is a complex pro- Predictive models created through data-driven TVA provide shipbuild-
cess that involves collecting data from a variety of sources, including ers with advanced analytical tools that can greatly improve both the
engine sensors, environmental trackers, and operational performance efficiency and safety of new vessels. For example, Key Performance
reports. This data must be filtered, synchronised, and harmonised to Indicators (KPIs) like Specific Fuel Oil Consumption (SFOC) can be
ensure accuracy. For instance, time data must be aligned across all derived from vibration data, giving shipbuilders insights into how
sources to avoid discrepancies that could lead to faulty conclusions. efficiently the engine is operating under various conditions.
One critical aspect of data preparation is ensuring that the model By incorporating these KPIs into the ship design process, engineers
accounts for the various operating conditions a ship will face. can make informed decisions about everything from engine place-
These can range from hull fouling, which increases drag, to fluctu- ment to hull structure and materials. This not only leads to more
ating cargo loads, which affect engine performance. By combining efficient ships but also helps reduce the environmental impact of
historical data with real-time sensor readings, shipbuilders can marine transportation, an increasingly important consideration in
create designs that are both efficient and adaptable to the many today’s maritime industry.
challenges a vessel will face at sea.
The future of data-driven
Building and validating predictive shipbuilding
models As the maritime industry continues to evolve, data-driven TVA will
After preparing the data, engineers use it to build predictive mod- become an integral part of the shipbuilding process. The combination
els that simulate how the ship’s propulsion system will behave of predictive analytics and real-time monitoring offers shipbuilders
under different conditions. These models are designed to estimate the ability to create vessels that are not only built to last but are also
key variables, such as the engine’s torsional vibrations and fuel designed to continuously improve over time. Data analytics will play
efficiency. Machine learning frameworks, such as decision trees, a central role at every stage of shipbuilding, from the initial design
are often used because they can handle complex, nonlinear re- to the ongoing maintenance and operation of the ship.
lationships between variables. In summary, the use of data-driven Torsional Vibration Analysis
Once the models are built, they must be validated to ensure they in shipbuilding marks a significant shift in how modern vessels
accurately predict real-world performance. One method is to are designed and maintained. By leveraging the power of Big
aggregate data from multiple ships of the same class to test how Data and machine learning, shipbuilders can create ships that are
well the models generalise across different vessels. This process more efficient, reliable, and environmentally sustainable. As more
helps ensure that the model is robust and adaptable, allowing companies in the marine industry adopt these technologies, the
shipbuilders to confidently use it in the design and construction future of shipbuilding looks set to be one of continuous innovation
of new ships. and improvement.

Follow us

on LinkedIn

38 ISSUE 29 / Q3 2024