Guide · Maintenance strategy
What is predictive maintenance for ships?
Predictive maintenance uses measured condition data to estimate when a machine will reach an unacceptable state, and schedules the work just before it does — instead of waiting for a calendar interval or a failure. This guide explains where it sits alongside planned maintenance, which parameters actually carry the signal, and what it takes to run at sea.
The four maintenance strategies, in order of maturity.
No fleet runs a single strategy. The question is which one each piece of equipment deserves, and criticality is what decides it.
Reactive (run to failure)
The machine is repaired after it breaks. Cheapest to plan, most expensive to live with — failures happen at sea, on the worst day, and often take a certificate or a charter with them.
Planned / preventive
Work falls due on a fixed calendar or running-hours interval, whether the machine needs it or not. This is what a classic PMS delivers and what class expects as a minimum.
Condition-based
Work is triggered by a measured parameter crossing a threshold — vibration, temperature, oil analysis, differential pressure. Effort follows evidence rather than the calendar.
Predictive
The trend in those parameters is projected forward to estimate when the machine will reach an unacceptable condition, so the work is scheduled into a convenient window before it does.
Which parameters actually predict failure.
A predictive programme is only as good as the parameters behind it. These six carry most of the signal on a conventional merchant vessel.
Vibration
Bearing wear, misalignment and imbalance in pumps, purifiers, turbochargers and shafting appear in the vibration signature long before they are audible.
Temperature & pressure deviation
Exhaust gas temperature spread across cylinders, scavenge air pressure, cooler differential and filter pressure drop each drift measurably before failure.
Lubricating oil analysis
Wear metals, water content and viscosity from routine samples give a direct read on internal condition that no external inspection provides.
Performance deviation
Rising specific fuel oil consumption or falling turbocharger efficiency against the vessel's own baseline is a whole-machine condition signal.
Running hours & load profile
Two identical generators with very different load histories do not wear at the same rate; hours alone are a poor proxy without the load behind them.
Insulation & electrical resistance
Megger readings trended over time show motor and cable degradation while there is still time to plan the replacement.
How to start without buying anything.
1. Rank equipment by criticality. List what stops the ship, endangers the crew, or breaches a certificate if it fails. That short list — usually twenty to forty items — is the only place predictive effort pays back.
2. Attach readings to the equipment record. The readings mostly exist already, scattered across engine-room log books and oil analysis emails. Recording them against the machine rather than the date is what turns them into a trend.
3. Set thresholds, then watch the slope. A threshold tells you the machine is in trouble now; the slope tells you when it will be. The second is the one that lets you book the work into a port call.
4. Keep the planned system underneath. Predictive findings should raise real jobs in the planned maintenance system with evidence attached, so class, flag and vetting see one coherent history.
5. Make it work offline. If a reading needs connectivity, it will be skipped on passage and your trend will have holes exactly where the load was highest.
Where Full Ahead Maritime fits.
Full Ahead Maritime holds the equipment register, the condition readings and the job history on one record, and works fully offline at sea. Performance-side signals come from ship performance monitoring, statutory work from the safety management registers, and the whole picture from the ship management platform.
Predictive maintenance questions.
Put condition data on the equipment record.
Start with a 30-day parallel review, and the vessel arrives with an equipment register already in place.
Honest AI · what this feature actually is
Three kinds of “AI”. We name which one you are getting.
Most vendors put one word on all three. The layers marked below are the ones this feature uses — the rest are not in play here.
Published arithmetic over your own records. Same inputs, same score, every time — traceable back to the row that produced it.
Used here
Bounded, auditable corrections to lead times, computed inside your workspace from your own outcomes. No model weights change and nothing is trained on your data.
Used here
Third-party models draft text only. Nothing they produce is applied automatically — a qualified person accepts, edits or rejects it.
Not used here
Proof, not a bullet point
Everything on this page keeps working with the link down.
Cut your connection and try it
Open the CII calculator, turn off Wi-Fi, keep typing. The arithmetic runs in your browser — no request leaves the device. That is the same engine the vessel app uses at sea.
Installable on the vessel PC
Full Ahead Maritime installs as an app and caches your fleet, jobs, certificates and evidence locally, so a superintendent can work a full day with no link at all.
A sync outbox, not lost work
Work done offline is queued and lands ashore in order when the link returns — over VSAT, Starlink or a two-minute port window.