The best-documented use of software to cut costs at sea is speed planning, not an automated captain. Under the International Maritime Organization's Just In Time (JIT) Arrival concept, a ship adjusts its speed so that it reaches the port's pilot boarding place when a berth, a fairway, and pilots and tugs are actually free, and modeled results range from about 4 percent to 23 percent less fuel depending on how early the port shares its timing. The IMO's 2020 guide does not mention artificial intelligence or machine learning, because its obstacles are contracts and data quality.
Key takeaways
- JIT Arrival is not slow steaming: the guide says total voyage duration stays the same, with less time spent waiting at anchor.
- Published fuel savings come from simulations and tracking-data analysis, and they range from 4 percent to 23 percent.
- Ships spend up to 9 percent of their time waiting at anchorages, according to an analysis of 2018 global tracking data.
- The main barriers named by the guide are charter contracts, demurrage, and weak or late port-call data.
- Machine learning could fit in the forecasting of delays, an area the guide does not evaluate.
What just-in-time arrival means
The Just In Time Arrival Guide was published in 2020 by the GloMEEP Project Coordination Unit of the International Maritime Organization, together with members of the Global Industry Alliance to Support Low Carbon Shipping. It describes a familiar pattern, "hurry up and wait." A ship sails at full speed toward a requested arrival time, then finds that the berth is occupied and anchors for hours, days, or even weeks.
The guide gives a worked example. A ship leaves port with a requested arrival at the pilot boarding place on day 14. Three days into the voyage, port delays move that time to day 17, but the ship is never told, so it keeps its speed and then anchors for three days. Under JIT Arrival the revised time is passed to the ship, and the master slows down to meet it. The guide stresses that this is not an average speed limit: the voyage may take more days at sea, but waiting is minimized and fuel use per mile falls.
What the studies measured
Every figure below is a modeled or analytical result, not a measurement of a fleet that adopted the practice. The savings differ because the studies assume different notice periods and different parts of the voyage.
| Source | What was examined | Reported result |
|---|---|---|
| Port of Rotterdam and TNO analysis, cited in the IMO guide | Container ships calling at Rotterdam in 2018, if each had known its arrival slot 12 hours ahead | 4 percent lower shipping emissions in the last 12 hours, or 134,000 tonnes of CO2 |
| Desktop exercise in the IMO guide | One container ship between Bremerhaven and Rotterdam, 247 nautical miles (about 457 kilometers), with early arrival-time updates | 23 percent less fuel than the base case |
| MarineTraffic and Energy and Environmental Research Associates, 2022 article by IMO GreenVoyage2050 | Global container voyages in 2019, speed optimized over the whole voyage | 14.16 percent mean fuel saving per voyage |
| Same study, shorter notice | Optimization in the last 24 hours, and in the last 12 hours | 5.90 percent and 4.23 percent |
The pattern is consistent: the earlier a ship learns when a berth will be free, the more speed it can shed. The 23 percent result comes from a single route in a desktop simulation, so it should not be read as a typical outcome. The guide itself adds that at higher assumed speeds the theoretical saving grows, but so does the safety concern of a ship holding position near busy approaches.
How much time ships wait
MarineTraffic analyzed global tracking (AIS) data for 2018 covering ships of 5,000 gross tonnage and above in dry bulk, wet bulk, container, and gas trades. It found that, depending on ship type, ships spend up to 9 percent of their time waiting at an anchorage, and the guide notes that ships may spend 5 to 10 percent of their time waiting to get into port. The analysis excluded ships that anchored for other reasons, such as waiting for new orders. The guide also points out that anchored ships still run auxiliary engines and boilers, so waiting produces emissions in the port area, and it does not aim to remove anchorage time entirely.
Where software and forecasting come in
Computing the best speed for a known arrival time is a routine optimization problem. The harder part is knowing the arrival time. A berth slot depends on the ship ahead, on cargo readiness, and on equipment. The guide notes, for example, that the accuracy of a terminal's expected completion time for a container ship can be thrown off by stuck containers or a breakdown in loading machinery. Forecasting such delays is a task for statistical or machine learning models, but the guide does not test or recommend any, and no vendor claim about this was verified for this article. What the guide does establish is that even a perfect forecast is useless unless it is shared quickly and accurately with the ship.
Why it is not yet universal
The guide lists several barriers. Many ports allocate berths on a first come, first served basis. In voyage-charter trades, a ship that arrives late can earn its owner demurrage, a payment for lost time, and that payment may exceed the fuel saved, which removes the incentive to slow down. Terminal planning can be commercially sensitive, so operators may be unwilling to share it, and timestamps are often inaccurate or infrequent.
The guide proposes starting with container shipping. It states that containerships make up about 15 percent of the international fleet but about 35 percent of shipping emissions, sail on fixed schedules, and face fewer contractual obstacles when on time charter or owned. Bulk and tanker segments follow.
Checklist for reading a savings claim
- Is the figure modeled, or measured on real voyages?
- How many hours of advance notice does it assume?
- Does it cover the whole voyage or only the last hours?
- Is it an average across many voyages or one route?
- Does the contract between owner and charterer allow the ship to slow down?
FAQ
Is just-in-time arrival the same as slow steaming?
No. The IMO guide separates the two: JIT Arrival matches speed to berth availability so that the overall voyage duration is not affected, while slow steaming lowers speed regardless of port conditions.
Does the IMO guide recommend artificial intelligence?
No. A text search of the guide finds no mention of artificial intelligence or machine learning. It focuses on contracts, data ownership, data sharing, and the accuracy and frequency of timestamps.
How large are the fuel savings?
The published studies range from 4.23 percent to 23 percent, depending on the notice period and the route. All are modeled or based on tracking data analysis rather than on a fleet-wide trial.
Which ships would benefit most?
The guide argues that container ships are the best first target because of their share of emissions, higher speeds, scheduled rotations, and existing contracts with terminals.
Related reading
- Efficiency and safety technologies at sea
- AI in maritime law
- Emerging technologies and industry benefits
- Which industries keep human control
More in AI Tools and Industry.
Last verified: October 2026
Sources
- Just In Time Arrival Guide: Barriers and Potential Solutions, GEF-UNDP-IMO GloMEEP Project and members of the Global Industry Alliance to Support Low Carbon Shipping, International Maritime Organization, 2020
- Lowering containership emissions through Just In Time arrivals, IMO GreenVoyage2050, June 7, 2022