Marine Incident Investigation: The First 72 Hours

A marine incident investigation is mostly won or lost in the first 72 hours. That is when VDR data can be overwritten, when people's memories are clearest, and when the company decides whether it is looking for a cause or for someone to blame. Doing it properly starts on board, before any investigator arrives.

The duty is not optional. Section 9 of the ISM Code requires the safety management system to include procedures for reporting non-conformities, accidents and hazardous situations to the company, and for investigating and analysing them with the aim of improving safety and pollution prevention. The company must also have procedures for corrective action, including measures to prevent recurrence. What follows is a practical sequence for the first three days.

What should the master do in the first hour after an incident?

Make the ship and people safe first, then protect the evidence. In order:

Make the ship and people safe. Everything else comes after this.
Save the VDR data. As soon as it is safe, use the recorder's data save or backup function as set out in the manufacturer's instructions and your SMS, then arrange a download. If you wait, the recording loop will overwrite the critical period.
Preserve other electronic records: ECDIS logs and screenshots, engine data logger, alarm printouts, CCTV, GMDSS and AIS records.
Leave alone anything that can wait: equipment settings, damaged parts, mooring lines and rigging.
Notify the company and the DPA under the SMS emergency procedure.

Why does saving the VDR data matter so much?

Because once it is overwritten it is gone, and with it the most objective account of what happened. How long you have depends on the recorder.

Recorder	Performance standard	Minimum retention before overwriting
VDR installed on or after 1 July 2014	MSC.333(90)	48 hours on the fixed and float-free capsules; 30 days on the long-term recording medium
Simplified VDR (S-VDR)	MSC.163(78)	12 hours

The investigation into the container ship MSC FLAMINIA shows what happens when the data is not secured. After a fire and explosion in a cargo hold in the Atlantic on 14 July 2012, the crew abandoned ship. Germany's Federal Bureau of Maritime Casualty Investigation (BSU) reported that analysis of the ship's S-VDR recording produced no results for two reasons: the crew had not activated an emergency backup before abandoning, and an auxiliary engine kept running, powering the ship for a further 24 hours, so the data for the accident period was overwritten. The BSU noted that the actual performance of duties on the bridge was unclear because of the missing paper-based and VDR records.

Saving data is rightly not the first priority while people are abandoning a burning ship. The lesson is that the save has to be a quick, practised step on the emergency checklist, so it is done the moment it is safe, by whoever is on the bridge.

What should happen in hours 1 to 24 of a marine incident investigation?

Collect people's accounts and the paper trail before both start to change.

Interview people separately and early. Talk to each person alone, in a calm setting, and ask open questions. Statements written together come out the same.
Photograph everything with time stamps and a scale reference.
Collect the paper trail: logbook pages, checklists, permits, the passage plan, the work and rest records of everyone involved, and the last relevant drill records.
Decide who investigates. A serious incident needs someone independent of the ship and of the line manager.

If a flag State or coastal State starts a marine safety investigation under the IMO Casualty Investigation Code (resolution MSC.255(84), whose Parts I and II became mandatory through SOLAS regulation XI-1/6 on 1 January 2010), the company's own investigation runs alongside it and does not replace it. Cooperate fully, and make sure your own team still reaches its own conclusions under the SMS.

How do you carry out root cause analysis in shipping incidents?

Build a timeline first, then ask at each event why it happened and why the barriers that should have stopped it did not. Where VDR data allows, the timeline should be accurate to the second, with audio, radar, steering and engine data side by side. This is where a VDR incident reconstruction earns its keep: it lets you test each statement against the record.

Good root cause analysis looks at three levels:

What people did: the decision or action just before the event.
Why it made sense to them at the time: fatigue, workload, training, unclear procedures, pressure from the schedule.
What the organisation allowed: manning, maintenance backlog, how the SMS was designed, how earlier near misses were handled.

If the root cause reads "human error" or "failure to follow procedure", the investigation has stopped too early. Ask why the procedure was not followed, and whether it could be followed as written. Our review of a recent container ship breakaway at an Indian port shows how the organisational level often explains what the people on deck could not control.

What should be done between hour 24 and hour 72?

Turn the evidence into a draft finding and a short list of actions, while the people involved are still available to check the facts.

Share the draft timeline with the master and key witnesses and correct factual errors.
Identify the barriers that failed and those that worked.
Check the fleet for the same conditions: the same equipment, procedure, trade or rank.
Issue an early safety alert to the fleet if the hazard could repeat before the final report.

What makes corrective actions stick?

Actions stick when they fix the system and someone checks they worked.

Each action has an owner, a deadline and a way to check that it worked.
Actions fix the system (procedure, training, design), not only the person. If the procedure itself was the problem, a simpler, usable SMS may be part of the fix.
Lessons go fleet-wide within weeks, in plain language, not as a circular nobody reads.
The same incident type is tracked across the fleet for 12 months afterwards.

If you need an independent view after an incident, our incident investigation team is there to help.

Frequently asked questions

What does the ISM Code require after an incident?

Section 9 of the ISM Code requires the safety management system to include procedures for reporting non-conformities, accidents and hazardous situations to the company, and for investigating and analysing them to improve safety and pollution prevention. The company must also have procedures for corrective action, including measures intended to prevent recurrence.

How quickly should VDR data be saved after an incident?

As soon as it is safe. A VDR built to MSC.333(90) keeps at least 48 hours on its capsules, but an S-VDR may keep only 12 hours before overwriting. In the MSC FLAMINIA case, data was lost because no backup was made and the recorder kept running after the crew abandoned ship.

Is a company investigation the same as a flag State investigation?

No. A marine safety investigation under the IMO Casualty Investigation Code is carried out by the flag State or another substantially interested State, and it is separate from other investigations. The company still has to investigate under its own SMS, as required by section 9 of the ISM Code, and act on what it finds.

What is a root cause in a marine incident?

A root cause is the organisational or system factor that, if fixed, stops the same type of event happening again. It usually sits in manning, scheduling, training, procedure design or the handling of earlier near misses. "Human error" alone is not a root cause; it is where the questions should start.

Who should lead a marine incident investigation?

For anything serious, someone independent of the ship and of the line manager responsible for it, with sea experience relevant to the event. The investigator needs authority to see all records and interview everyone involved, and the company must accept findings that point at its own systems, not only at the crew.

Sources

Back to Blog