The global shipping sector is approaching its next big transition. From 2020, shipping companies operating outside emissions control areas (ECAs) will be forced to meet tighter regulations in the shape of the well-documented global 0.50% sulphur cap.
On 1 January, the 2020 cap—sometimes referred to as IMO 2020—will bring the current sulphur limit outside ECAs down from 3.5%. For most shipping companies, the changes will require varying combinations of enhanced vessels, streamlined routes, new ports, unfamiliar equipment, and higher fuel costs.
Given that the changes come into effect in less than 12 months, that’s a lot to think about.
To survive this period of upheaval and change in a sector already grappling with heightened security risks, recruitment challenges and rising operational costs, machine learning will be a vital tool in the arsenal of any shipping company, regardless of size, location, or mode of operation.
Embracing machine learning can no longer be regarded as a ‘nice-to-have’: in the post-2020 world, failing to make the most of machine learning could spell disaster.
Getting to grips with the basics
With machine learning, it all starts with data. Big data. With tens of thousands of vessels traversing the world’s oceans under endlessly varying conditions and operational patterns, shipping operations have always had the potential for generating large amounts of data, but it’s capturing and making use of this information that has historically been the big challenge.
Data is simply a collection of numbers until it is analyzed and interpreted. Until this happens, it simply cannot be put to use. When information is interpreted and acted upon, that raw data can finally become something with the power to change and enhance business outcomes.
Machine learning is an advanced form of artificial intelligence in which systems use powerful algorithms and logic to cut through the noise. By learning from experience, and by being able to identify patterns and discover trends, decisions can be made with minimal, or no, human involvement. Imagine a system with the capacity to reschedule a cargo pick-up directly with a client, automatically respond appropriately to a customer based on their historical behavior, or inform you that a vessel is at risk of a regulatory breach and a fine you could do without.
Machine learning models are highly adaptive. They are continuously revised and refined, becoming more accurate as new data enriches the dataset. Solutions based on machine learning enable businesses to solve complex problems quickly, and much more effectively, than has been possible traditionally.
Benefits large and small
Machine learning algorithms can enhance business scalability by analyzing global trade patterns and helping operators to optimize capacity. They can predict increases or decreases in demand from different customer types and forecast spikes in local, regional or national activity.
Machine learning can also help ships to realize significant fuel savings by selecting routes that avoid unfavorable sea and meteorological conditions, ultimately leading to great levels of dependability, and therefore higher profits.
Some of the business benefits of machine learning are small and incremental, while others are far-reaching. If an operation has numerous inefficiencies that go unchecked and unresolved, it will become ever more difficult to succeed; using technology to find areas that require improvement, and then coming up with the requisite solutions, is vital.
In short, it’s all about improving efficiency to survive during and beyond rapid and deep change. By helping human experts to make better-informed decisions, accurate predictive automated modelling frees up the creative human for alternative tasks, which may include responding to emergencies or other unforeseen events. Of course, properly-deployed machine learning can also minimize those.
Learning on the job
Machine learning can also help revolutionize deck ops. By detecting the signs leading up to a preventable event, the ‘machine brain’ can notice trends and patterns, allowing decision-making that will minimize operational disruption.
Take that valve that seems to fail every two or three years, for instance. Is there a common reason for the failure? How often does the cause occur? Could it be tracked, and its significance calculated? Could the system automatically flag up that it’s time to repair or replace before another failure occurs? Or is it time to take the plunge and take a completely new approach that embraces more reliable systems?
Machine learning gives you the ability to build a complete fleet-wide picture of maintenance needs, ensuring you can plan effectively and efficiently, which will cut maintenance costs, reduce vessel downtime, and avoid nasty surprises.
Machine learning can also help in the backroom, with issues such as fraud detection and other financial irregularities, malicious online activity, risk management, and even investment planning.
Quite simply, if you can capture and interpret the right data, and then understand how best to make use of your findings, there are no limits to what machine learning can help you to achieve.
Lessons from other industries
Shipping bosses only need cast a glance at other sectors to understand the benefits of machine learning. Just as loyalty card data helps a retailer model customer purchasing patterns, shipping must embrace machine learning to monitor and analyze trade patterns, predict customer behavior and produce prescriptive insights that can help to increase customer base or boost customer retention through more tailored services.
The pharmaceutical and medical sector is increasingly employing data analysis to better meet healthcare needs such as diagnosis of diseases and outbreak prediction. Using similar tactics in shipping could enable companies to reduce risk through improved planning in virtually all areas of ship operations, such as supply, maintenance, crewing requirements, vessel network planning, container pairing, route optimization, and emissions or waste reporting.
In an era where companies’ bottom lines are increasingly threatened, optimizing vessel usage and routing is essential. Sailing an unladen vessel to reposition it for a pickup is frustrating, expensive, and environmentally undesirable. With the correct application of machine learning, such scenarios can be all but eliminated.
Machine learning can help balance import and export activities at ports, transforming the way cargo deficits are handled. An upward shift in capacity utilization could mean not only improved cost efficiency and consistently enhanced rates offered to customers, but could also free up vessels for other voyages to increase income.
Time to embrace the learning curve
Machine learning technology can set the foundations for future developments and technological advances in shipping, and will play a large role in changing and underpinning the whole sector.
For an industry steeped in years of tradition and culture, change can be hard to accept. But ship operators who adapt and embrace machine learning will be the ones who beat their competitors, who future-proof themselves, and who survive beyond this tumultuous period.