On the 3rd of March 2023, a passenger in a private jet died after severe injuries in the cabin, which were originally reported as caused by severe turbulence.

This was quickly picked up by the mainstream media and, for the next few weeks, news and oped articles focused on the ongoing issues of changing weather patterns and increased turbulence.

Incidents of severe turbulence are becoming more common, often leading to injury. This accident stood out as fatalities are very rare. FAA data from 2009 to 2022 shows 163 people suffering serious injuries caused by turbulence, of which the great majority of those injured (129 cases) were crew. Over this time, there are no references to fatalities.

The most recent case, when Singapore Airlines flight SQ321 experienced extreme turbulence in May 2024, led to thirty passengers injured and one death by a passenger who suffered a heart attack. Before that, the most recent turbulence-related passenger flight with a fatality in the NTSB database was United Airlines flight 826 in 1997.

The reason the 2023 flight is not listed is because although tragically a passenger lost her life during the flight, the extreme forces on the aircraft were not caused by turbulence.

The NTSB investigated for over two years before releasing their final report.

The day before the accident, the flight crew flew two passengers from Leesburg Executive Airport in Virginia to Dillant/Hopkins Airport in New Hampshire. The following day, a Friday, the flight crew arrived at the airport to prepare for the flight back to Leesburg. The first officer was in the process of the walk-around, an exterior pre-flight inspection of the aircraft. He was interrupted when an employee of the fixed base operator arrived with ice for the aircraft’s bar. The first officer paused to take the ice and then returned to his pre-flight inspection. He thought he was continuing where he left off; however, he didn’t notice that the red-non-streamer pitot cover was still on the right pitot probe.

Pitot probes do not work when sealed or blocked, so they are routinely covered in order to protect the probes from debris and insects. These days, the covers generally have streamers attached to draw the eye to the fact that the cover is still in place. However, when the operator received the aircraft from the seller, there were no streamers attached to the pitot covers and the operator did not purchase new streamers.

The captain arrived at the aircraft having settled the fuel bill and ramp fees. He also did not notice that one pitot cover was still on the probe and logged that the first officer had finished the walk-around.

The passengers arrived ready for their flight: a couple and their son. The captain got them settled in the cabin and did what he called a normal passenger briefing: “Because they’re regular passengers and pretty much aircraft owners, our normal is about time en route, the weather, and seatbelts on for take-off and landing, as always, and can I get you anything before we leave?”

That day, the captain was the Pilot Flying and the first officer was the Pilot Monitoring.

During the take-off roll, the first officer noticed that although the Challenger 300 was accelerating normally, the right-side primary flight control had stopped showing acceleration above 40 knots. He checked the captain’s primary flight control which showed normal acceleration. He called out that he had no airspeed. Both pilots called out “abort” together. The captain slowed the aircraft from 104 knots and exited the runway onto a taxiway.

Sixteen seconds had elapsed from the start of the take-off run to the successful abort.

The first officer later said that as soon as he saw that he didn’t have airspeed, he remembered being interrupted during this walk-around check. He had a bad feeling that the pitot cover might be the cause of the issue.

First officer: Did … Did I [expletive] take it off? … Did I not [expletive] take it off?

First officer: [expletive]

Captain: I thought you did the walk around.

First officer: I did.

First officer: Ah [expletive].

Captain: Turn ’em off.

First officer: Fff–.

They exited the runway to the taxiway and the captain shut down the left engine.

Passenger: What happened?

First officer: Oh, nothin’. We… we’re getting a… a weird reading with the airspeed on one side. We’re just going to check something.

The first officer got out to take a look. Sure enough, as he crossed to the right side of the aircraft, he saw that the red cover was still on the pitot probe. This had obviously caused the issue during the take-off run. He checked the probe for damage but it looked fine. He returned to the cockpit, no doubt embarrassed at his rookie mistake, but there was no harm done.

While the captain was waiting, he noticed an advisory on the Crew Alerting System: RUDDER LIMITER FAULT. He’d seen this error a few times before, when the Challenger 300 was “cold soaked”: exposed to cold temperatures for an extended period, for example, parked in freezing conditions or in an extended high-altitude flight. On those occasions, he’d been able to clear the message with an avionic stall test (STALL/RUD LIM test). He figured the fault was actually in the rudder test, not in the rudder itself, as it cleared once the aircraft had warmed up.

He ran the avionic stall test twice but the advisory was still there.

Once the first officer was back in the cockpit, they discussed the rudder limiter fault message. There was an additional issue: the flight director, which is part of the aircraft’s guidance system, was in pitch mode. In pitch mode, it commands pitch movements (up and down) but not lateral guidance. This could be connected to the rudder limiter fault, which is part of the aircraft’s lateral control systems.

Captain: Why is it in pitch?

First officer: I’m not sure. [mutters] Nav… flight director, rudder limiter fault.

Captain: Yeah, but that’s not it.

First officer: I’ll call them.

Captain: Who you calling?

First officer: You want to take off with a rudder limiter fault?

Captain: It’s advisory only.

The message was in blue, which meant it was an advisory message, rather than in red or yellow along with a specific action. The captain knew this advisory message came up in the cold without any flight issues. His response was clear: he didn’t think it was worth holding up the flight for.

The captain said later that he didn’t see any need to look it up again, as on previous occasions he’d consulted the Minimum Equipment List and there was no MEL item for it. His recollection was that according to the Quick Reference Handbook, the advisory could stay on for up to ten landings. The first officer agreed: it was just an advisory and it didn’t specify that they should go looking into the GO/NO GO guide in the Quick Reference Handbook.

In all actuality, the GO/NO GO guide shows that the Rudder Limiter Fault is a clear NO GO item. When asked later, the Director of Operations said that there was a 24/7 maintenance control service that the crew could and should have called. If they had, he said,  they would have been instructed to shut down both engines, depower the aircraft, and then start everything back up in order to clear any computer-driven faults. If that reset didn’t clear the message, then the control service would have checked the issue against the Minimum Equipment List.

But the flight crew didn’t call. It was just an advisory, after all.

First officer: Well, that might be causing what … the reason it’s not going out of pitch [mode].

Captain: Well, it’s … I don’t know why it won’t go into takeoff.

The TO/GA button activates the flight director’s take-off mode.

First officer: Yeah, I know. I think it’ll go into altitude though.

Captain: Yeah.

First officer: Let’s go.

Captain: ‘Kay.

They taxied back to the runway and began the take-off roll again. The captain pressed the take-off/go-around (TO/GA) button one more time. The flight director command bars, which provide visual guidance to the Pilot Flying for the climb, did not appear. He continued.

The first officer called out that he had airspeed; taking off the pitot cover had resolved that issue.

However, at the same time, he noticed that the airspeed indicator wasn’t showing the V-speed bugs: as Pilot Monitoring, he needed to monitor the V-speeds to give guidance to the Pilot Flying. He had entered the data to define these before the flight and he was sure the V-speeds had appeared on the previous take-off run. As it happened, he still remembered the speeds from that aborted take off. He called out eighty knots followed by V1, the decision speed, at 116 knots. Then he called out “rotate” as they reached take-off speed.

The Challenger 300 was airborne.

Captain: Gear up.

A master caution sounded.

First officer: Trim fail. Four hundred feet.

Captain: Flaps up.

First officer: And I’ll give [air traffic control] a call.

First officer: Boston Airshare Three Hundred [is] fifteen hundred feet for six thousand off of Keene.

Once they were established in the climb, the captain engaged the autopilot and the turn onto their course to Leesburg. As they climbed to 6,000 feet above mean sea level, the Crew Alerting System showed repeated caution messages. The flight data recorder does not record specific caution messages. The two pilots remembered seeing MACH TRIM FAIL, AP STAB TRIM FAIL (Autopilot Stabilizer Trim Failure) and AP HOLDING NOSE DOWN, although they couldn’t remember the order or if there were additional messages to these.

At the same time, the Air Traffic controller cleared the flight to continue to climb to FL230 (23,000 feet).

The first officer asked the captain if he wanted a lower altitude but the captain didn’t. “No, get the checklist.”

The first officer attempted to re-input the V-speeds into the flight management system. “I think it’s a configuration issue from the beginning,” he said.

A “cavalry charge” alert sounded in the cockpit, an alert that the autopilot had disconnected.

The first officer asked whether the autopilot had failed or if the captain had disconnected it.

“I did that,” said the captain.

According to the Flight Data Recorder, the autopilot was engaged twice more during the climb. Each time, the Crew Alerting System displayed multiple caution messages and then the autopilot was disconnected. Each time the autopilot disconnected, there was a manual adjustment of the horizontal stabiliser trim.

First officer: I’d just leave the autopilot off.

Captain: All right. Get the checklist going.

First, the first officer explained about the V-speed issues again and asked the captain how to program the Flight Management System. They discussed the system and the V-speeds for about four minutes, at which point it seemed to work.

First officer: OK, there we go. They took those.

Captain: All right. Run the checklist.

The first officer pulled up the quick reference card. There was only one trim fail checklist on the card:  the PRI STAB TRIM FAIL (Primary Stabiliser Trim Failure) checklist. He thought it seemed like it could be the root cause of their problem.

First officer: Okay, primary stab trim fail. Autopilot holding nose down. I’ve got that… the first one here…

He showed the captain the checklist and the captain agreed. The first action was to move the stabiliser trim switch from primary to OFF. The first officer called out “Stab trim off” and then moved the switch to off.

The autopilot immediately disconnected with an alert sound. The nose-down force from the autopilot stopped and the elevator moved to neutral, causing the aircraft to rapidly pitch up.

Captain: Woah.

First officer: [expletive]

The captain quickly pushed forward on the control column to apply nose-down force. The aircraft pitched down violently and the captain must have reduced the force in response, as the nose then pitched back up.

Captain: [expletive]

“I did not expect it to pitch as rapidly as it did in either direction,” he said later. He had expected that once they turned the stabiliser trim switch off, the autopilot would disconnect, as it did, but wasn’t prepared for the sudden pitch up. The first officer said that he thought the autopilot was already off.

Understanding what happened here is easier with the FDR data. The autopilot had been holding 5.3° nose-down elevator to compensate for the nose-up trim condition. When the crew moved the trim switch to off, the elevators suddenly moved to neutral. The aircraft quickly pitched up from 3° to 11° and the normal acceleration hit 4g. The startled captain pushed forward on the controls with 90 pounds of force. This caused g-forces of -2.3g, pushing everyone to the ceiling. The nose went up again to 20° nose-up, again causing g-forces of over 4g. For context, the aircraft design limit was +2.6g.

“Turn it on, turn it on,” shouted the captain over the robotic voice calling STALL STALL STALL. The first officer moved the stabiliser trim switch back to primary. The captain used the manual pitch trim and the aircraft flew normally again. They had regained control.

“We shouldn’t have had the autopilot on,” said the first officer.

“Yeah,” said the captain.

A voice: “Is everyone all right?”

Captain: Ask them if they’re OK.

The husband said later that it happened without warning and it felt like the aircraft was breaking apart. After about twenty seconds of this, he turned and saw his son lying on the ground next to his seat. Then he realised his wife wasn’t in her seat. She was lying in the aisle near the lavatory, unconscious and bleeding.

He called out to the pilots that his wife had been seriously injured. The first officer moved to the back of the aircraft to help. He returned a moment later to tell the captain that there was a medical emergency: the passenger had a serious cut at the top of her head. They needed to land.

The captain left the autopilot off and flew manually. They contacted air traffic control to say they needed to descend and land, asking for an airport with medical facilities nearby. They agreed to divert to Bradley International Airport in Windsor Locks, Connecticut.

Within seventeen minutes, they had landed safely and taxied to the ramp where an ambulance was waiting. The paramedics carried the woman out of the aircraft and to the hospital but it was to no avail. She died in hospital later that day.

An inspection of the aircraft showed the damage: items scattered all over the cabin and shelving broken inside of cabinets. Oxygen masks dangled over seats. Over the aisle, where they’d found the unconscious woman, part of the curved wood panelling and ceiling was cracked and dented.

There was no evidence of any seat belts breaking free.

There was no turbulence. But what had happened?

This all started with the covered pitot probe on the first take-off run.  The first officer saw that he had no airspeed on his side and they correctly pulled off the runway to deal with the issue. When the first officer removed the pitot cover, the flight crew thought they had solved it.

However, the aircraft’s systems had also detected the fault: mismatched airspeed data lasting over five seconds. The systems had no way of knowing that the fault had been rectified. Thus, it logged the issue that the airspeed data was not to be trusted (CONFIRMED MACH VALID was set to FALSE).

In order to clear this, the flight crew needed to reset the trim control system: either by pulling out its dedicated circuit breaker (HSTECU) or by powering the aircraft down and back up again.

The crew obviously did not realise this, nor did they associate the RUDDER LIMITER FAULT, which the captain had seen before, with the aborted take-off roll with the pitot tube cover still in place. As a result, they returned to the runway with the fault condition still in place, which is what led to the repeated warning messages.

Accurate airspeed data is critical for safe trim operation, especially at high speeds. Stuck in a state of “we can’t trust the Mach/speed data”, the Horizontal Stabiliser Trim Electronic Control Unit (HSTECU) entered a protective degraded mode with limited trim functionality as a safety response. This mode disabled the autopilot’s ability to adjust the trim and reduced the rate of manual trim movement. These adjustments would prevent potentially dangerous trim adjustments based on bad data.

The warning messages were directly related:

• AP STAB TRIM FAIL: The autopilot cannot trim the stabiliser
• AP HOLDING NOSE DOWN: The autopilot is holding elevator pressure that it shouldn’t need.

The aircraft had been out of trim the entire flight and the autopilot was compensating.

A complete power reset would have cleared the faults and allowed them to continue safely. Having failed to do this on the ground, the crew’s only hope was the correct diagnosis of the fault messages.

When the first officer chose the wrong checklist and moved the stabiliser trim switch to OFF, it disconnected the autopilot, which very suddenly released the 5.3° nose-down pressure that the autopilot had been holding. This caused the sudden pitch-up, which the captain countered with an immediate pitch-down. The passengers who were not seated or not wearing their seatbelts were thrown up into the ceiling with no warning.

If they had followed the checklist for AP STAB TRIM FAIL and AP HOLDING NOSE DOWN checklists in the Quick Reference Handbook, then they would been warned against abrupt changes in control forces when the autopilot disconnected. Those checklists specified that the flight controls must be held firmly, with an explicit warning to minimise changes to the airspeed and configuration. But because the captain wasn’t ready, he wasn’t able to hold the controls against the sudden pitch up and, startled, he reacted too strongly in trying to get the aircraft level again.

When the NTSB released their final report, two years later, the media did not cover the new story: there was no turbulence. The violent forces that killed a passenger were entirely pilot-induced oscillations caused by a chain of human errors.

The media latched onto the story of severe turbulence because it fit a familiar narrative. It’s something passengers understand and fear. But the actual sequence of events tells a different story: one of seemingly minor mistakes compounding into catastrophe. A used aircraft delivered without streamers on the pitot covers. A pitot cover left on. A repeated fault that the captain had become used to, but this time with a very different cause. Confirmation bias leading to a decision to fly with a “NO GO” advisory. The wrong checklist selected at the wrong time.

So while the initial reports blamed turbulence, the truth is both simpler and more complex. This wasn’t one dramatic failure but a series of small decisions that added up to tragedy. The turbulence everyone feared never existed. The real story was far more human.