Artificial intelligence (AI) and autonomous shipping (unmanned vessels) are set to transform the mercantile fleets of the world. In addition to the greening of the industry and the smartening of the logistics, there will soon have to be a wholesale revamp of design with respect to understanding of the changes that will be wrought in the maintenance and repair of cargo vessels. Crews are essential in the design of current vessels. Autonomy is all well and good, but will need a full redesign of the present norms. This article outlines the steps that need to be addressed and proposes some systems of work to modernise maritime maintenance.
An introduction to maritime maintenance
Maritime maintenance comes in many flavours, but can be categorised into five core domains:
- Breakdown which is “fix it if it breaks, leaks or squeaks”.
- Periodic/Preventative which is “we do the laundry every Tuesday whether it needs it or not”.
- Risk-based which is basically preventative, but somebody looks at the washing to see if it is dirty.
- Condition-based which is measuring all the variables and dealing with situations when factors such as heat, vibration or noise drift outside acceptable limits.
- Prescriptive maintenance which is condition-based, plus AI spotting trends and pre-guessing repair and replacement needs. All of these have their pros and cons depending on the system and industry in which they are deployed.
AI is perfect for optimising what needs to be done, but not so hot on the gruntwork. This can include changing out a burst hose, cleaning a clogged fuel filter, tightening nuts and bolts, clearing ventilation pipes, and dealing with vermin to name but a few. These are all tasks that an autonomous ship will still need to do if the current propulsion for mercantile ships continues.
Sadly, maintenance and repair have a lowly status in the hierarchy of shipboard activities, much to the chagrin of many of the engineering officers. It is easy to imagine how physical maintenance would be low on the list of priorities for autonomous ship designers. While they will, doubtless, design systems so that there can be no single point of failure in the actual machinery, there are many other misfortunes that can stop the smartest ship dead in the water.
We can turn to data, AI and advanced machine learning methods. We can apply advanced mathematical methods and utilise computing capacity specifically developed for the marine industry. However, while this may help to minimise unplanned downtime in equipment, machine learning has difficulty with black swan events.
Maintenance in the face of uncertainty
Dealing with rare and unpredictable outlier events (black swans) is difficult for any maintenance and repair system. It is also easy to fall victim to two of the psychological traps that beset this subject, namely the frequency illusion and using the past to predict the future. The writer has experienced the gruntwork mentioned in the introduction, sometimes more than once; the cleaning of clogged fuel filters being the most onerous. For those who think modern fuels wouldn’t clog, it is a fact of life that everything clogs. LNG, methane, ammonia and any fuel all have clogging mechanisms that occur at the most inappropriate moments and all the testing and quality assurance will only mitigate the problems rather than eliminate them. Let’s face it, stars are just clogs in the vacuum of space.
The frequency illusion refers to an unusual occurrence that then becomes commonplace. Or when remembering events, it is easier to recall the unusual and attribute greater significance to its value. However, rare events are never as rare as we think. An example given in a previous article was the satellite discovery that there are, at any moment, roughly ten “hundred-year waves” rolling around the oceans. So having fallen consciously into both these traps, the question arises as to how autonomous shipping will deal with them.
Autonomous maritime maintenance stratagems
Autonomous shipping, at first glance, would seem to be confined to large fleet owners meaning budgeting and scalability would be less of a problem. The reason being that, in all likelihood, designs would be standardised to a degree that allows teams in each port to undertake standard maintenance without extensive retraining. This would avoid major challenges from the number of assets, equipment and the range of possible faults, failures and malfunctions in different vessel configurations. Standardised design would also allow a holistic approach to improving vessel efficiency and reliability. Scalability of a wide range of vessel types that make up a typical fleet presents problems in itself as motive power requirements for a Very Large Crude Carrier (VLCC) versus for river transport versus coastal trade are vastly different.
What AI does not do as effectively as a marine engineering team is diagnose and determine the actions needed to remedy a risky situation. Even if there is a team in every port capable of standard maintenance for the range of vessels likely to berth, there needs to be a maintenance repair hit team ready to fly to vessels drifting at sea. These need to be highly-skilled international engineers, able to transit through any country at short notice and who spend most of their time idle. This is a recipe for freelance sub-contractors and not employees of the fleet owners.
Vessels capable of autonomy already exist, mostly in the oilfield, military and research institutions. They are characterised by a high degree of redundancy and mostly fully electric driven propulsion, manoeuvring or station-keeping. Unfortunately, high degrees of redundancy is not high on the shopping list of mercantile fleet owners. The general rule for larger ships is one huge engine and a generator for ancillary loads. Electric propulsion is not good for large vessels. The largest electric motor (36.5 MW) is only two-thirds the size of the engine driving the largest boxship (56.8 MW after de-rating) and, for reliability, a good rule of thumb is never to buy the largest nor the smallest in any range of products.
Electric drives and systems tend to be far more reliable and need less maintenance than either hydraulic, pneumatic or mechanical systems. 5-10 megaWatt motors are available and widely used in the offshore drilling industry with solid-state speed control so autonomous ships up to a certain size would be relatively reliable, and fairly easy to maintain. The systems would be somewhat more expensive than current marine diesel, and detailed cost accounting would be needed to assess the relative returns. The returns, in terms of safety and a decrease in insured damages, might easily tip the scale in favour of autonomy.
There can be little doubt that autonomous vessels for inshore, lakes and inland waterways are sensible where they do not with leisure water users. Many cities are equipped with an extensive waterway infrastructure that is often barely occupied by commercial users and therefore holds significant potential for the relocation of transport from the road onto the waterways. The benefits of this could be extensive, most pertinently diminishing road congestion and air pollution.