‘Big Data’ needs careful handling but can bring benefits
In an increasingly digital environment, shipping is increasingly reliant on huge amounts of data, but panellists at last month’s ShipInsight conference were divided over its merits. Guided by questions raised by delegates, the panel considered whether large amounts of data could be managed effectively and whether algorithms could be trusted to make decisions based on it.
Capt Kuba Szymanski, secretary general of InterManager, was sceptical whether communications technology was adequate for data-enabled technology to be relied on at sea. He drew a parallel with his land-based communications experience in his London office, where he said there is only one spot where he can get a mobile phone signal.
A similar level of service would be unacceptable at sea, he said, and suggested that satellite communications are not always good enough. He accepted that satellite connections are available everywhere, but in some places they are very expensive so, in general, “when we talk about Big Data and whether we can analyse it, I’m afraid we are not there yet,” he said.
Also on the panel was Stefano Poli, vice-president for business development at Inmarsat Maritime, who had earlier presented a keynote address that reflected that concern, based on a survey of shipowners. Mr Poli reported that 51% of them had said that getting data off a ship in real time is the biggest obstacle to them adopting ‘Internet of Things’ (IoT) technologies.
Yet all of them expect to be using IoT systems by 2023 to monitor fuel consumption, with an average spend of US$2.5M on such technology over the next three years, and he predicted that Inmarsat’s Fleet Data service – which he described as “the IoT platform for shipping” – will provide a better service than current satellite data transfer models.
Bart Schuitema, sales manager at Radio Holland, provided a case study on how effective use of Big Data had improved that company’s service level, although he believes that expression “is an over-hyped term”. It is simply a tool, he said, to be used to address specific business needs. “It does not replace people,” he said.
Radio Holland has long used databases to keep track of equipment fitted on ships to enable a quick response to, for example, a captain who reports a fault on a radar. But now, as more information sources are becoming available – including equipment that reports its own status updates – there is too much data for a service coordinator to plan a service intervention. “This is what Big Data is about,” he said. “It is about having huge amounts of data and different data sources but it’s also about how you apply it. It is helping our people to work in an efficient manner.”
Gijsbert de Jong, marketing and sales director of Bureau Veritas, also underscored the importance of effective maintenance, saying that the value of an asset “lies in its ability to perform its functions”. That depends on two things, he said: reliability and maintainability. Both of these are enhanced by using sensors and connectivity to enable remote monitoring to support strategies such as condition-based maintenance or predictive maintenance, he said.
Those can be supported by referring to a ship’s ‘digital twin’, which is developed during a ship’s design phase. It allows designs to be considered in a three-dimensional form rather than as two-dimensional drawings and can be linked to data-processing tools to make such calculations as strength and stability.
Once the ship is delivered, this digital twin is passed to the ship’s owner who can use it for asset management purposes, he said. On its own, “Big Data is just Big Data, whatever that means,” he said. But technologies such as the digital twin, make that data meaningful, he said.
Whether data analysis can be similarly relied on for safety-related functions onboard was a more contentious topic during the discussion. The Nautical Institute’s training manager, Capt Maneesh Varma, spoke about machine learning, or artificial intelligence, in which a computer will be programmed to find structure in a mass of data.
“That answer can be correct or incorrect,” he said. “That’s where experience comes in.” As an analogy, he referred to a chess computer that learns from previous games it has played. “One time you win. You play the same game again and the machine has learned and will defeat you.” In short, its analysis has generated different answers on different occasions.
One delegate told the panel of a company that is developing a ‘virtual captain’ that will use artificial intelligence to learn and act 10,000 times faster than human. Capt Varma was not impressed. He described this approach as a ‘trial and error method’ in which the software learns from itself. “What’s the use of being faster but incorrect?” he asked.
Capt Szymanski had other concerns about relying on sophisticated data systems on ships. Unlike shore offices, there is no resident IT expert on board most ships to call on if there is a fault. Appointing an IT officer would be “a no-brainer for me”, he said, and welcomed an initiative that he said some “forward-thinking companies” had made by appointing a ‘writer’ to carry out onboard IT and administrative tasks. Otherwise, they fall to the well-paid captain, he said. “Is this really money well spent?”