← Back to use cases
Heerema pilots machine learning to plan Sleipnir’s Atlantic transit
Added 2026-10-14 · Reviewed 2026-10-15 · Heerema Marine Contractors
Read the source ↓What the company did
Heerema reports piloting Amphitrite’s machine-learning routing tool on Sleipnir’s late-2025 Rotterdam–New York transit. Five years of voyage records supported a vessel-specific model of speed and power consumption, combined with weather, waves and currents to advise routing and departure decisions.
How it went
Heerema describes the crew using the recommendations during a demanding North Atlantic crossing and reports gains against the originally planned route. These are company estimates from one transit, without independent verification or evidence of repeatable fleet-wide savings.
Worth knowing
- The captain says crews remained in control, with the model adding vessel-specific environmental context to their judgment.
- Practical takeaway: Compare an advised route with the original passage plan, arrival obligation and power assumptions. Preserve the forecast available at the decision time when reviewing the outcome.
Sources & provenance
Data-Driven Voyage Optimization Delivers Measurable Impact On Sleipnir Transit ↗
Heerema Marine Contractors · Published 2026-07-07 · Reviewed 2026-10-15
“develop a dedicated Machine Learning model of Sleipnir”
Supporting evidence and review
Human review required
- Review forecasts and modeled performance against observed vessel behavior.
- Keep operational limits and the master’s route authority explicit.
Evidence & limitations
- Company-reported pilot on a late 2025 transit, published 2026.
- A single transit does not establish fleet-wide or repeatable savings.
- Reported time and emissions savings are omitted from this entry.