Root-Cause Analysis in Marine Engineering: A Practical Approach

Most machinery faults on board get resolved the same way: something fails, it gets repaired or replaced, and the vessel moves on. Most of the time that's entirely correct — there's no need to investigate further when a straightforward repair holds. The problem is the minority of cases where it doesn't hold, and the same fault, or a close variant of it, comes back weeks or months later.

Root-cause analysis is simply the discipline of not stopping at the first plausible explanation.

Why symptom-led repairs sometimes fail to hold

A failed component is a symptom. It's rarely the whole story. A bearing that fails prematurely might be a bearing quality issue — or it might be a symptom of misalignment, contamination, incorrect lubrication, or a vibration source entirely unrelated to the bearing itself. Replace the bearing without addressing whichever of those is actually responsible, and the replacement bearing is simply on the same trajectory as the one it replaced.

This is the pattern behind most "recurring fault" callouts: a competent repair was carried out, but to the symptom rather than the cause.

What a root-cause investigation actually involves

Establishing what's actually known, versus assumed

It's worth being explicit early on about what's confirmed by evidence and what's being assumed from experience or convenience. Maintenance logs, operating parameters at the time of failure, and the failure history of the specific component all count as evidence. "It's probably the same as last time" does not.

Looking for contributing factors, not a single cause

Failures are often the result of more than one factor combining, rather than one clean root cause. A system might tolerate slight misalignment fine on its own, and tolerate a slightly degraded lubricant fine on its own, but fail when both are present together. Treating either factor alone as "the" cause and fixing only that one can leave the underlying vulnerability in place.

Testing the explanation, not just proposing one

A plausible explanation isn't the same as a confirmed one. Where practical, a genuine root-cause investigation looks for a way to test the hypothesis — through inspection, measurement, or observing whether the fault behaviour changes when a suspected factor is corrected — rather than closing the investigation as soon as a reasonable-sounding cause has been proposed.

Considering the maintenance and operating context

The same physical fault can have a different root cause depending on how the vessel is operated and maintained. A failure that would be a simple wear-and-tear replacement on one operating profile might indicate a genuine design or maintenance-interval mismatch on another. This is why root-cause work often needs to look beyond the failed component itself, into how the system is actually being run day to day.

What this looks like in practice

A structured investigation typically moves through: understanding the operational context and history, gathering physical and documented evidence, identifying candidate causes and contributing factors, testing those candidates against the evidence, and only then proposing a corrective action — one that addresses the cause, not just the symptom that prompted the investigation.

General principle versus a specific vessel's fault

The approach above is a general framework, not a diagnosis. Every vessel's machinery, maintenance history and operating pattern are different, and root-cause work has to be grounded in that specific context rather than applied generically. What it does mean is that a fault which has already resisted one or two repair attempts is usually a sign that the underlying cause hasn't yet been correctly identified — not that the equipment is simply unreliable.

This is the basis of Torven Marine's troubleshooting & investigations work — independent, evidence-led investigation rather than another round of symptom-led repair. If you have a fault that's recurred more than once, get in touch and tell us what's been tried so far.