Physics-Based Models Aren't Cutting It: It's Time for AI

Aleksandar-Saša Milaković

Physics-based models were a marvel of their time, revolutionizing marine shipping by making emissions estimates far more predictable — and helped to make voyages more profitable. But physics-based models haved reached their limits. Today vessels require more predictability and efficiency than ever before, and we need tools that are smarter, better, and faster to achieve the emissions goals of the global fleet.

Today, even the best physics-based models only achieve 90% accuracy — and it’s not unusual for physics-based models to be off by as much as 30%. As new CII regulations come into force this year, even a difference of a few percentage points can mean millions of dollars of lost revenue. With that kind of money at stake, physics-based models just aren’t up to the job anymore.

The Problem

Behind every physics-based model are humans calling the shots. They try to plug in all the pertinent parameters and conditions that might influence a ship’s performance. But there are hundreds of variables that shift with every new voyage, making it virtually impossible to keep track of every factor.

There’s no accurate way to mathematically represent a multi-front wind hitting the ship while currents shift below, or the toll such punishment takes on the engine pushing through it. Wave heights mean nothing without knowing the duration between them, or the force of the wind that propels them. What about other environmental load considerations? Ocean currents, fuel types, load distribution, and many other factors matter for successful voyages. What if the physics model omits some criteria?

The Solution: AI

Physics-based models, with their human inputs, simply can’t account for the multitude of variables that influence a given sea voyage. But artificial intelligence can. AI like Bearing’s, powered by titanic amounts of data, routinely delivers 50% better accuracy than physics-based models do on their best day.

Unlike physics models, AI’s job never ends. Throughout a ship’s voyage, it collects AIS data. In fact, AI’s enormous capacity for learning means it constantly processes global shipping fleet data and marine weather conditions. Together, these give Bearing’s AI its comprehensive understanding of seafaring outcomes worldwide, ever improving its ability to accurately model fuel consumption and emissions.

That big-picture perspective is helpful, but AI goes further by parsing unique data on every ship in a fleet. With the collected AIS reports, vessel specifics, and the option of easily inputting noon reports once a ship docks, the AI better understands how vessels weather the conditions at sea, and how such conditions affect their fuel consumption and emissions.

That’s where AI really leaves physics models behind, because now fleet owners can follow any ship’s anticipated CII ratings throughout the year. If there’s room for improvement, and there always is, AI offers suggestions for achieving a better grade with the least impact to your bottom line. Fleets using physics-based models are slowing ships down for emissions reductions, but AI turns to surprisingly effective alternatives. From modifying routes and using alternative fuels to even adding an extra hull-scraping to the vessel’s annual maintenance, AI finds small solutions for big results in vessels needing to improve emissions.

Being a data-based solution means you can be running Bearing’s CII module fleet-wide in just days. With 95% accuracy, can you afford obsolete models that don’t help you improve performance after every voyage? Learn more about how we may be able to help you go green without slowing down your fleet.

More Articles