REVISED RELIABILITY CENTRED MAINTENANCE STRATEGY PROPOSED FOR AUTONOMOUS VESSELS

The aim of reliability centred maintenance (RCM) methodology is to ensure that a system continues to function as desired. Unlike traditional maintenance strategies, it considers the reliability of ship systems and sub-systems as a whole rather than focusing on individual components.

Achieving reliability can involve preventive maintenance undertaken at scheduled time intervals, condition-based maintenance undertaken when it is observed that critical parameters indicate corrective action is desirable and predictive maintenance that additionally uses historical data and forecasts to assess when action is called for.

In a study published in February 2021, University of Southern Denmark PhD student Stig Eriksen evaluated the seven steps typically used to develop RCM programs, comparing the needs of a typical manned ship with those of an unmanned vessel. Working under the supervision of Dr Ingrid Bouwer Utne from the Norwegian University of Science and Technology and Dr Marie Lützen from the University of Southern Denmark, and with input from a working group involving Kongsberg Maritime, Eriksen then demonstrated the value of his proposed amendments using a case study of a main engine low temperature fresh water cooling system.

RCM methodology

The RCM methodology, originally developed for aviation, typically addresses seven questions about a system that can be summarised as:

1. What are the functions and performance standards?
2. What ways can it fail?
3. What are the causes of the failures?
4. What are the effects of the failures?
5. What are the possible consequences of the failures?
6. What preventative tasks can be undertaken?
7. What should be done if a preventative task is not able to be performed?

While the actual preventative tasks that would be undertaken on an unmanned ship would be the same as those on a manned ship, there is a key difference in timing. On manned ships, a failed equipment unit would be repaired by the onboard crew and the voyage length would therefore not have any impact on the probability of failure of the system in which the equipment unit is installed. For an unmanned vessel, repairs would generally be undertaken once the vessel reached port, or even after it reached a specific port where the appropriate engineers and equipment were available.

This potential delay is critical to understanding the consequences of a failure, and Eriksen proposes that the answers to Question 4 should assess immediate and long-term effects separately, as the assumed “timely” nature of maintenance cannot be guaranteed. Additionally, he proposes that Question 5 should involve risk assessments that are specific to the intended voyage length of the unmanned vessel and that these assessments should distinguish between preventative and corrective maintenance tasks.

Failure modes examined for the cooling system case study included external and internal leakage, and Eriksen notes that while preventative and corrective maintenance reduces risk in manned operations, it is not enough to achieve an acceptable risk level for unmanned operations. On long voyages, there is more time for an otherwise non-critical leak to completely drain the system of oil or water, leading to a loss of function of the system, and in this case a loss of propulsion.

Most corrective maintenance tasks are done manually at sea, something not possible on an unmanned ship, and preventative maintenance alone was not enough to reduce failure risks to an acceptable level: “Maintenance can contribute to maintaining the level of reliability designed and built into the system, but the level of reliability can never be higher than that inherently provided in the system design.”

Design modifications were therefore proposed, but even having redundant heat exchangers could not reduce the risk for unmanned vessels in the case of an internal leak which could cause result in water mixing with the engine lube oil and damaging the main engine bearings. Such a leak could be detected using camera surveillance or remote oil analysis equipment, but the potential for then removing the water is limited on an unmanned vessel. To reduce the risk, a double-wall heat exchanger could be used. Any leak of water or oil would then run into a void space.

While the outcome of risk assessments undertaken as a part of the RCM methodology do depend on the operational profile of the ship, a change to this would not necessarily mean a redesign of systems or the need for greater system redundancy, says Eriksen. “The risk scenario for say ‘loss of propulsion’ would theoretically be different for a ship when it is sailing in dense traffic close to shore than when it is in the middle of the ocean. In a real application, however, a cargo ship would have to able to handle both so you would either average out the operation to a typical operational scenario or maybe do several scenarios.”

Redundancy risks increasing complexity

There is the potential for a Catch-22 situation in the design of systems for unmanned vessels where more redundancy, more remote operation capabilities, more actuators and more sensors are needed to support an RCM regime. Yet the extra equipment adds to the potential for failures. Finding the right balance can involve evaluation of other parameters and also cost. “A big challenge in finding the right balance is a lack of information on the effects of automation and complexity in systems,” says Eriksen. “Uncertainty in general is a big challenge in achieving super reliable systems. There are often interactions between systems or components which have not been considered and unexpected failures happen even in systems with high levels of redundancy.”

He notes the example of the engine failure of the Viking Sky in 2019. The level of lubricating oil in the tanks was relatively low but still within set limits when the vessel encountered heavy seas. The movement of the vessel led to a lack of supply to the pumps which in turn triggered a low-level alarm and caused the three of the vessel’s engines to shut down.

While unmanned operation has the potential to reduce human error as a contributing factor to incidents at sea, Eriksen says it will not disappear, but rather its nature may change. For the specific case study of the cooling system, unmanned operation may not have a significant impact on human errors related to maintenance, since maintenance must still be carried out manually by repair personnel in port.

The maintenance burden on large ships is considerable (in the area of 40% of the total onboard work burden), and unmanned operation will not affect this significantly, says Eriksen. Much of the research on unmanned ships is focused on automating the task of sailing and navigating the ship. These tasks, however, only constitutes a small part of the total work burden (around 10-20% according to a study he has undertaken which is not yet published).

Usage data gaps complicate analysis

RCM has not been widely adopted in shipping, so the move to its use in unmanned vessels will require a shift in current thinking. “The biggest problem in the application of RCM, for manned or unmanned ships, is a lack of failure rate data and the lack of standardization of components. The shipping business is much more fragmented than the airline industry for example. Even the biggest shipping companies don’t have enough of the same components to gather useful failure rate data, and the equipment manufactures often have no access to their equipment once it is installed on a ship.”

While the use of artificial intelligence could expand in applications such as weather routing and energy optimization, Eriksen says he is sceptical about its usefulness to the early development of unmanned ships. “There are many very down to earth practical issues we must solve for unmanned ships to become a reality, and I don’t think we can rely on a computer to figure them out for us.”

Unmanned operation of cargo ships is still in its very infancy, despite what some people say, he says. “There is a clear trend towards more automation, and especially more exchange of data between systems and between ship and shore. Some would say that we are seeing the introduction of autonomous systems in shipping. I would argue that what is now described as autonomy is merely a different shade of automation, which has been a part of shipping for decades.”

Nonetheless, for this study at least, Eriksen concludes that the risks posed by unmanned operation are manageable with design changes to the ship. However, the risks were not found to be lower for unmanned operation than for manned operation in any scenario he examined. “The main difference between manned and unmanned operation regarding reliability is found to be the greatly differing possibilities for corrective maintenance actions. This presents a major challenge to the unmanned operation of commercial cargo ships.”