The art of digital maintenance

The drive for cost savings and efficiency improvements is an important goal to operators in a marine industry that is still reeling from a downturn in the market. Since the advent of the digital era, many suppliers have expounded that shipowners need to invest in digital systems, data analytics and machine learning if they are serious about being key players in the marine industry in the future.

There is a range of software solutions coming onto the market that enables more collaboration between departments and moves the industry toward autonomous ships. But digitization and the use of machine learning and artificial intelligence (AI) is not something that will happen overnight. Instead it is a step-by-step process, as the industry learns and adapts to working with the new technology. Krishna Uppuluri, vice president of digital products at GE’s Marine Solutions, believes that there are certain aspects to machine learning and AI that need to be improved before it can be fully implemented in the maritime industry. “The marine industry has been lagging in digitization,” says Uppuluri. “But AI will have to earn its stripes as well – and this will mean a step-by-step introduction to the marine industry – first through non-critical systems, then to redundant systems, and finally to the critical ones.”

Algorithmic predictions
Companies such as Rolls-Royce have also been working on AI as part of developing autonomous ships. The company believes that through machine learning and better data-driven optimization, AI will not just save costs in maintenance, but also the time spent maintaining vessels. Machine learning is a set of algorithms, tools and techniques that mimic human learning behavior to solve problems. Rolls-Royce is using machine learning algorithms to analyze data from currently operational marine equipment and is training software models that can recognize unknown patterns in the data and make a prognosis about how that equipment is performing. Kevin Daffey, director of engineering and technology, commercial marine, Rolls-Royce, says, “If the data we analyze is ‘big’, then the model can recognize more complex patterns and make more accurate predictions about the state of the marine equipment than any human could.” Potentially this means maintenance in the future could be carried out in a more timely and cost-effective way and could further improve the reliability of equipment.

“Machine learning can perform predictive analytics far faster and more accurately than any human can. The potential for marine maintenance is to move completely away from timebased scheduled maintenance, to maintenance that is based on equipment use and true plant condition,” says Daffey. However the question arises as to how this AI-enabled approach to maintenance will fit in with manufacturers’ requirements in the future – for example, with timed checks that are stated for their products, which if not followed can invalidate warranties.

Daffey believes that this area of concern is still up for debate, because most engineers, when faced with a choice over how to maintain equipment, prefer to default back to manufacturers’ requirements and operating parameters. Daffey says, “We need to encourage the industry to change its thinking toward systems. It may be the case that we may not move all systems over to condition-based monitoring in the future.”

New tools
Engineers can use 3D digital models with operational data to improve designs “Enabling streaming data from ships opens up the opportunity for a lot of analytics on the data. This can be for individual ships as well as for learning across a whole fleet. Mogens Mathiesen, co-founder and commercial lead, Arundo Analytics move all systems over to condition-based monitoring in the future.” Better tools for better results Kongsberg launched its AI solution, Kognifai, to the market last year. The software brings all the Kongsberg solutions onto a single platform, where data can be shared and used by the company to enhance product development for the end user.

Hege Skryseth, president of Kongsberg Digital, says, “Data and the knowledge we can gain are central to the ongoing digitization of the maritime industry. “There are many opportunities in marine maintenance. The end goal is to provide better tools to the decision makers responsible for maintenance, helping them do their job in the best possible way.” The enabling of predictive maintenance through data-driven systems is expected to add further value to the maintenance process. “It’s safe to assume that AI could help early failure detection in all types of equipment and machinery on board a vessel.

As AI used for predictive maintenance is fairly new to the maritime industry, we expect the first use cases will be centered on critical equipment such as engines and generators,” says Skryseth. Rolls-Royce also has R&D projects looking at predictive maintenance and some of its customers have seen gains with early deployments of the technology. The company has found that the use of predictive maintenance depends on how the ship is operated, as well as how appropriate the equipment is for maintenance and care.

For one customer, Rolls-Royce performed an energy audit of a platform support vessel. “We discovered from an analysis of data – which included the power load on the engines, vessel fuel consumption, propulsion shaft RPM and ship speed – that for nearly 13% of the operational time, the vessel could be run more efficiently on one engine rather than two,” says Daffey. “The customer made changes to its operating practices, which resulted in a large fuel saving and reduced the wear rate of the engines.” Daffey says that AI could be used for predictive maintenance for condition monitoring, predictive monitoring, optimization of operations, simulation of operations and autonomous operations.