The ability to gather masses of high-frequency data is nothing new in maritime. Enabled by global connectivity and internet of things technology, the stream of numbers covering fuel consumption, performance, weather and hundreds of other variables has never been greater
The question is how to make sense of all of that information; a challenge well beyond manual analysis that requires sophisticated analytics tools capable of comprehending information at scale and across timelines.
Which is, of course, where artificial intelligence (AI) comes in. More specifically, machine learning (ML) systems and their algorithms that are capable of churning through disparate data points throughout the history of operations to bring to light insights among the noise: key relationships between variables that can be used to predict future outcomes.
There are plenty of ways this kind of capability can be put to work. Predictive vessel management and maintenance is one, spotting the signs in the build-up to a preventable event and stepping in before it disrupts operations. Another is the ability to optimise voyages in real-time, a technique that promises to substantially improve operators’ returns.
Understandably, much of the hype around digitalisation and the advent of AI in maritime is focused on boosting those bottom lines. More often than not this is viewed in terms of fuel consumption. And rightly so, particularly given the 2020 sulphur cap and the US$15Bn fuel cost hike expected to hit the global shipping container industry.
But sustainability is now an issue in and of itself. The latest report from the UN’s Intergovernmental Panel on Climate Change laid out a bleak prognosis for the future of the planet, including the fact that we may have just 12 years to take meaningful action on climate change. The report outlined the need to cut global emissions by a whopping 45%.
The maritime industry has a moral responsibility – and in some cases, a legal requirement – to improve on sustainability. And although it is not going to be easy, the promise of AI and ML is real.
The initial challenges, however, are pretty fundamental, as co-founder of data-driven shipping sustainability platform We4Sea, Dan Veen explains.
“There are many companies with different and even conflicting interests who control part of the supply chain: shipowners, managers, charterers, agents, suppliers and equipment manufacturers,” he said.
“They all play a role in getting, distributing or analysing data. Transparency in the supply chain, enabled by AI and big data, is not [necessarily] in the interest of all of those parties.”
It is also about the technical limitations of those taking decisions relating to sustainability. “The knowledge on IT-related issues is limited,” he said. “Many managers are former captains – very experienced in those kinds of matters – but without training in the fields of AI and ML.”
Those technical deficiencies form something of a snowball effect when paired with issues of methodology and mentality. “A lot of shipping companies are managing the company like a ship: just like a captain, you are trained to solve all issues yourself.”
The solution is easy to see but not necessarily one that shipowners are taking. “With the speed of development in IT, there is no other choice than to partner with external companies that can help you to learn and develop faster,” said Mr Veen.
For example, he points to the huge majority of vessels still using noon reports as their primary source of fuel efficiency data.
“That means that all data is relying on a manual, error-prone report sent once every 24 hours. Knowing that data on vessel speed, weather data, heading and draft is available every 2-3 minutes, it’s surprising that many shipping companies do not use the abundant flow of data... to monitor vessels in near real-time, or even to predict things that may happen.”
“The data is there on our doorstep, we just have to learn how to use it,” he concluded.
Enabled by global connectivity and internet of things technology, the stream of numbers covering fuel consumption, performance, weather and hundreds of other variables has never been greater
ML-driven data science
Maritime tech company Nautilus Labs is another offering ML-driven data science to ship operators, despite admitting that the potential of the technology has not yet been fully realised.
Nautilus Labs chief executive Matt Heider outlined reasons for the gap between the hype and the current state of affairs. The first of which is that “many maritime businesses still trust manually collected and reported data more than high-frequency data,” he said.
“This leads to a dependency on this type of data – even though it is limited and error-prone. That reliance on old methods is hard to change.”
There is also the fact that, for machine learning in maritime to reap the rewards it has promised, AI needs access not only to huge swathes of accurate, high-frequency vessel data, but also other data sets about the environment the vessel operates in.
“Tying those two points together, if a business does not have or trust the elemental dataset, then it is impossible to create actionable insights with ML,” he said.
But there is hope that AI can drive improved sustainability. “We see ML and the artificial intelligence it will create in the future as the bedrock of a slew of innovations that will bend the emissions curve of the industry and bring it to a more sustainable place,” continued Mr Heider.
“All across the value chain, there are small ways that different components of the market can be made more efficient with advanced software – and those small efficiencies aggregate to massive savings over time. For our clients in shipping, ML-based fleet optimisation – once fully embraced and developed – has the ability to reduce total fuel usage by up to 30%.”
In the case of We4Sea, using the data on the doorstep involves developing digital twins of ships to see where performance can be improved through benchmark comparisons, data analysis and simulations.
Nautilus Labs combine proprietary algorithms with KPI’s, weather data, route plans and more to construct a voyage optimisation calculator that clients can leverage on a daily basis.
Supply chain visibility
But sustainability is not all about voyage optimisation. AI and ML can also help to eliminate unnecessary voyages in the first instance. San Francisco-based ClearMetal provides predictive supply chain visibility software for shipping companies and other logistics carriers.
“The reality is that AI and ML are only as good as the data you feed them,” ClearMetal’s director of marketing Tyler Holmes explained. “The supply chain is one of the most complicated things for shippers around the world to get a complete picture of. The end-to-end supply chain is massive, spread across an enormous amount of data formats, systems, direct partners, and third parties. Add to this challenge that often the data shippers have access to is late, incomplete, inaccurate or never arrives at all.”
The silver lining of that scale and complexity is that even the smallest of changes can have a big impact. “AI and ML are the only way for those changes in efficiency to be made at scale because there is just no way for a human to crunch the amount of data involved,” he said.
“It is imperative to leverage these technologies to make changes in an iterative fashion because even small changes that a human might not pick up on translate to enormous gains at scale.”
Mr Holmes’ advice is applicable to maritime leaders whether they are looking to optimise voyages or harness the power of improved supply chain visibility. “What we recommend to our customers is ‘start where you can make an impact’,” he said.
“As fast as technology evolves, those traditional long-term projects run the risk of being obsolete by the time they are complete. The opportunity cost of not taking an iterative approach to optimisation is scary when we are talking about trillions of dollars in the global economy.”
Which brings us back full circle to the cold hard facts of maritime sustainability. Despite the pressing need to take action on climate change, decisions will ultimately be driven by dollars.
“Left unchecked, the maritime industry could account for a fifth of global greenhouse gas emissions by 2050,” said Mr Heider.
But AI solutions could provide “a meaningful step towards not only improving sustainability efforts but also improving voyage economics, and ultimately the two have to be linked,” he said.
“The industry has no compelling reason to act if it is not driven by economic motivations, and we see ML and AI as the technological bridge that connects improved efficiency with both better financial and environmental outcomes.”
Fortunately for the planet, the numbers bear that out. Mr Veen described a successful project in which the company’s ML platform was used to analyse two sister ships of a Dutch shipowner burning 30 tonnes of fuel per day at sea.
By comparing data collected with the company’s digital twin technology, it became clear that both vessels were not sailing according to their design specifications. “Most noticeable was the relatively low speed of sailing, he explained. “That had an impact on the average engine load, which was often below 40% of its maximum rating.”
“As the shipowner had no torque meter data, this remained unnoticed for 15 years. So we simulated a different scenario where the engine was de-rated to be closer to the actual use of the engine and showed a possible saving of 5-6% on fuel consumption.”
Nautilus Labs also shared some bold predictions about how AI and ML could make small differences on a huge scale in maritime’s battle to become more sustainable.
“These numbers really add up,” said Mr Heider. “Even if we conservatively estimate an average reduction of 0.5M tonnes of fuel waste per day due to improved monitoring, transparency and accountability, that equates to many millions of tonnes of reduced consumption per year globally across the world’s fleet.”
And that could be just the beginning. “The follow-on impact for the environment is massive, and that is just by eliminating the most basic forms of fuel waste. As we continue to get into more advanced forms of the decision support for both the crew and the shoreside teams, the fuel savings increase dramatically, as does the benefit to the environment and our clients’ bottom line,” he said.
No doubt Nautilus Labs, We4Sea, ClearMetal and others in the maritime efficiency space will hope impending climate disaster will be the catalyst for maritime leaders to embrace AI, ML and data science in a bid to reduce fuel consumption and emissions. But you don’t need an algorithm to conclude that it will not quite be that straightforward.