Compute Maritime says its GenDSOM project—funded by the UK Government—has produced a generative-AI ship design for an offshore wind crew transfer vessel that, in detailed modeling, saves 101,671 liters of fuel and 258.7 tons of CO2 per vessel each year.
The consortium approach combines NeuralShipper AI hull-form work with a diesel-electric hybrid propulsion system developed with Siemens Energy, then runs it through additive-manufacturing workflows intended to keep design constraints inside the AI loop.
GenDSOM builds the CTV around NeuralShipper hull-form optimization
GenDSOM is aimed at bringing generative AI and additive manufacturing into ship design, with consortium partners Siemens Digital Industries Software, Rapid Fusion, HP, BYD Naval Architects, and the University of Southampton.
The designed vessel is a 32.5-meter twin-hull CTV intended to carry 24 offshore wind technicians and four crew. Its hull form was developed using Compute Maritime’s Neural Shipper AI, and it was paired with a diesel-electric hybrid propulsion system developed with Siemens Energy.
Compute Maritime’s performance modeling claims the AI-driven hull reduces the power required at the vessel’s 25-knot service speed by 6.3%, with reductions of up to 11.6% at higher speeds.
Modeled day-energy performance and full speed delivery
Compute Maritime’s modeling compares the NeuralShipper-optimized vessel against a conventional baseline operating the same offshore wind duty cycle. The combined result is claimed to be an 11.1% reduction in annual fuel consumption and an 8.9% reduction in CO2 emissions against a like-for-like conventional diesel vessel.
On a modeled full day of operations, the baseline vessel ends the day with a 34 kWh energy deficit, is said to draw the batteries beyond their safe discharge limit, and is said to fall short of the 25-knot service speed. The Neural Shipper-optimized vessel is said to finish the day with a 106 kWh surplus within the battery operating envelope and still deliver full service speed.
Adding constraints into the AI loop via Rapid Fusion Apollo LFAM
GenDSOM also pushes manufacturing into the design workflow. Compute Maritime and Rapid Fusion developed an additive-manufacturing toolset intended to integrate constraints such as build volume, support structures, and material behavior into the NeuralShipper design loop, aiming to make components printable from the outset rather than reworked for manufacture afterward.
Using this approach, the consortium designed and produced a hydrofoil using Rapid Fusion’s Apollo robotic large-format additive manufacturing (LFAM) system. The project description says Apollo uses high-deposition pellet printing to build large composite and thermoplastic components.
The hydrofoil is characterized as a working demonstration of an unbroken pathway from AI-generated geometry to a finished, production-ready marine component.
Charging-ready propulsion and a roadmap toward ~95% CO2 reduction
The vessel is designed as a future-proof, upgradeable platform, and the project description says specifying an offshore fast-charging inlet during the build stage prepares the vessel to draw power from charging infrastructure installed at sea by wind farm operators.
As the electricity grid and offshore charging infrastructure decarbonize, Compute Maritime’s roadmap claims the vessel can progressively shift from diesel to battery power without changing its core propulsion system. It also says the roadmap points toward emissions reductions of around 95% over the vessel’s 25-year service life.
GenDSOM is described as responding to the UK’s 2025 Maritime Decarbonisation Strategy, which targets emissions reductions of 30% by 2030 and 80% by 2040 for domestic maritime. The funding is said to come from the UK Government through the UK Shipping Office for Reducing Emissions (UK SHORE) program in the Department for Transport, with Innovate UK (part of UK Research and Innovation) as the main delivery partner for UK SHORE interventions.