At night, over the sea, approaching the hover, with poor references and a sneaky failure, things can get weird. Quickly 👀
On 28 February 2024, a Sikorsky S-92A was descending towards a hover over the North Sea.
The helicopter was doing exactly what the crew expected it to do.
Until… it wasn’t.
The nose began to rise as the aircraft slowed (which was normal).
But instead of stopping at around 10 to 12 degrees nose-up, the pitch attitude kept increasing.
15 degrees, 20 degrees, then towards 30 degrees.
By the time the crew recognised what was happening and attempted to recover, the helicopter was approaching a condition from which, at that height, recovery would become impossible.
Seconds later, they impacted the sea.
It was an interaction between a technical failure, automation, monitoring, expectations, procedures, training and a really small potential window in which the crew could have broken the chain.
So why didn’t that happen? Let’s take a look 👀
Ready to learn from real aviation decisions, mistakes and challenges? Get our latest Notes from the Cockpit sent straight to your inbox, twice a month ⤵️
💥 Accident Overview
LN-OIJ was a Bristow Norway S-92A operating a Search and Rescue training flight from Bergen Airport, Flesland, on behalf of Equinor.
There were six crew members onboard.
The mission involved dropping a training beacon, conducting hoist training with a vessel, and then returning to locate and retrieve the beacon.
You can see the aircraft flight path here:
Conditions for that final part of the flight were challenging:
It was night, overcast, with no moon or starlight and very few external visual references.
Winds near the surface were around 35 knots from the South-East, and a significant wave height of 3 to 4 meters.
Not an easy situation to be in as a helicopter pilot!
To retrieve the beacon, the crew used the S-92 SAR automation mode called Mark on Top, or “MOT”.
MOT can take a marked position, turn the helicopter into wind, decelerate and transition towards a hover.
Bristow’s procedure was to adjust the resulting hover height to 150 ft.
The commander was PF and the co-pilot PM.
As the helicopter approached the final position, everything initially appeared normal.
At 18:39:19 the commander said:
“And the helicopter will stop shortly.”
The co-pilot subsequently called:
“Approach coupled.”
Then:
“Radalt coupled” and “It should level out.”
Sikorsky indicated that around 10 to 12° nose-up was normal during the deceleration, although Bristow stated that attitudes up to 18° were not uncommon during the final deceleration of a MOT.
The problem is… the pitch didn’t stop there.
The commander recognised that something was wrong.
He pushed the cyclic forward and called for a go-around.
But the aircraft had already reached approximately 30 degrees nose-up, entered rearward flight and entered vortex ring state.
Resulting in the helicopter impacting the sea.
Five occupants ultimately survived. One crew member, the SAR nurse, died.
So… What in the world happened here?
🔍 Investigation Findings
The investigators interviewed the crew, there’s a bit to unpack here.
The report states:
“The commander has explained to the NSIA that he saw the pitch attitude increasing, and that he felt that something was off. He then briefly glanced at the standby attitude indicator to verify that his primary attitude indicator was functioning correctly. “
And:
“He then realised that the helicopter attitude was wrong and tried to initiate a go-around by pushing the cyclic forward and pulling collective.”
The MOT normally finishes with a transition to a position hold mode, but this never happened.
The report does state:
“According to FDR data a pitch degrade was likely annunciated on the PFD. However, this was never identified by the flight crew.”
MOT was still engaged when the commander took over.
The helicopter reached approximately 30° nose-up and was already accelerating rearwards.
By the final reliable FDR data (at just 15 ft above the sea) it was travelling backwards at approximately 40 kt.
The report concludes:
“Based on all available evidence the NSIA concludes that it is likely that the failure of the pitch trim actuator was present before the crash.”
And:
“The failure of the pitch trim actuator influenced the pitch motion of the helicopter and failed to arrest the expected pitch-up manoeuvre.”
Regarding the recovery from the flight crew, the team mentions;
“This was not recognised by the crew in the three to six second window the NSIA believes the situation could be identified as abnormal and theoretically corrected.”
But they further state that this wasn’t necessarily a fault of the crew, as it went from expected to unexpected behaviour at a crucial time;
“The NSIA has no indication that the crew of LN-OIJ differed significantly from other pilots that had received the same training during the SAR South mobilisation.”
This illustrates further that it wasn’t “mishandling” by the individual crew, and more of a human factors element:
Detecting a transition from normal to abnormal operation at a crucial point in time.
Like with many accidents; this could have been any other pilot.
💡 What can we Learn from This?
Quite a few learning points here, let’s start with the biggest one.
1️⃣ Monitor the aircraft, not just the automation.
For us, one of the biggest lessons here is that monitoring automation isn’t the same thing as checking what automation is or isn’t engaged.
Understanding expected aircraft behaviour, and comparing it to observed aircraft (or automation) behaviour, is one of the main pillars of active monitoring.
Of course this is way easier said than done, especially while flying low, poor conditions, at night, over the sea, with multiple other influencing factors around you.
Even with perfect monitoring, a failure state like this would be difficult to detect in the moment.
2️⃣ Expected behaviour can hide abnormal behaviour.
The helicopter here didn’t suddenly do something obviously wrong.
It started doing something expected, pitching up to decelerate, and then simply continued too far.
That makes deviation much harder to detect.
3️⃣ Monitoring responsibilities need to be explicit.
“Both pilots are monitoring” can be weaker than clearly defining who is responsible for attitude, flight path and automation during critical phases.
The accident highlights how task-sharing ambiguity can create gaps at exactly the wrong moment.
4️⃣ Train failures in their realistic context.
The commander said the accident did not feel like the trim-hardover events practised in the simulator.
Training needs to reproduce not just the failure, but the subtle way it can present while automation is already commanding a legitimate manoeuvre.
5️⃣ Ask why the system depended on a few seconds of human intervention.
The NSIA estimated there was only a three-to-six-second window to recognise and correct the abnormal pitch.
The deeper safety question therefore should be about how aircraft design, alerting, procedures and training can make the next failure easier to detect and manage.
A 3 to 6 second window won’t end well, as this accident demonstrates.
💭 Conclusion
The strange part of the LN-OIJ accident is that there isn’t an obvious “villain”.
Two highly experienced S-92 pilots encountered a likely pitch trim actuator failure, low over the sea, at night, with few visual references.
They had perhaps three to six seconds to recognise that an expected pitch-up had become abnormal.
The NSIA found no indication that this crew differed significantly from others who had received the same training.
So perhaps the better question isn’t:
“Why didn’t they catch it?”
It’s:
“Why did the system require them to catch it within seconds?”
That shifts the conversation towards automation design, alerting, procedures, training and task sharing.
Automation can tell us what it intends to do.
But we still have to monitor what the aircraft is actually doing.
You can find the final NSIA report here.
Ready to learn from real aviation decisions, mistakes and challenges? Get our latest Notes from the Cockpit sent straight to your inbox, twice a month ⤵️
0 Comments