How to Read an NTSB Report.

The latest one, about a fatal accident this month, was unusual. Here's what the investigators were telling us.

Dealing with “Remove Before Flight” ribbons, like this one shown on the wing of an Airbus plane, is part of a standard preflight checklist. A new NTSB report includes warnings like these in the story of a recent corporate-jet tragedy. (Photo Johannes Eisele/AFP via Getty Images.)

Over the past month I’ve written about several commercial aviation close calls.

• A number could have been catastrophic—notably those in Austin, Maui, Boston at Logan Airport, and New York at JFK.

• All appear to have involved human miscalculations, by pilots or controllers, rather than mechanical, electronic, or other system breakdowns, or exceptional weather.

• Fortunately none led to injury or death. Statistically commercial air travel remains exceptionally safe.

• But together they raise the question of whether such incidents are just a run of bad luck or instead signs of dangerous systemwide strain.

Here is one more episode. I mention it for several reasons, apart from its intrinsic tragedy: It led to a death; it illustrates why initial impressions of an accident’s cause can change; and it may be more evidence of a system under strain.

I also offer it as an example of how to read the specific language of an NTSB report—what is said, and what goes unsaid.

---

The initial reports: violent weather claims another victim.

A few weeks ago a late-winter storm brought violent weather to much of the country. On the night of March 1 a Lufthansa flight from Austin to Frankfurt hit very rough air and had to divert to Dulles airport, outside Washington. Seven passengers were taken to hospitals. Airports and aviators in widespread locations reported unusually powerful winds.

On the afternoon of Friday, March 3, with winds still roaring in many places, a small private jet carried a family from Keene NH to Leesburg VA, near Dulles. On board this Challenger 300 jet were five people—two crew members and the family of three, two parents and their teenaged son. The family later said that they had been visiting “schools in New England.”

A few minutes into the flight, when the plane reached an altitude of around 6,000 feet, it suddenly pitched sharply up and down, and like the Lufthansa soared and plunged. Four of the people aboard the Challenger were unhurt. But the mother in the family was badly injured as the plane was pummeled in the skies.

The crew made an emergency diversion to Bradley field, outside Hartford, where an ambulance was waiting by the runway. By the next morning her death had been announced. Soon she was identified as Dana Hyde, a well-known and widely respected figure in policy and non-profit circles in Washington. For instance, during the Obama years she had been confirmed by the Senate as CEO of the Millennium Challenge Corporation, which worked on anti-poverty projects worldwide. In a statement after her death her husband, Jonathan Chambers, said she was “the best person I ever knew.” Chambers is CEO of the rural-internet company Conexon, which also owned the airplane, with tail number N300ER.

The widespread impression was that Dana Hyde was another victim of nature’s violence, perhaps from turbulence intensified by climate change. The NTSB’s first press statement said it was investigating this “turbulence event,” which was also the framing from most news organizations.

But by the following Monday, NTSB investigators said they were “now looking at a reported trim issue that occurred prior to the in-flight upset,” as Michael Laris put it in a Washington Post story.

Last week, the NTSB spelled out what that meant.

---

The terseness of most ‘Preliminary Reports.’

It can take months or years for the NTSB to issue its “final” conclusions about an accident. These can be well over 100 pages long and go into exhaustive detail. Weather conditions, toxicology reports on pilots, metallurgical analysis of engine parts, plus anything else that might support a “probable cause” assessment of what went wrong.

The NTSB’s “preliminary reports,” which come out much more quickly, can be very spare and brief. Sometimes they are just a few paragraphs of the basic who, where, what, when of a crash, with little information on the why. An example is here.

By those standards, the “preliminary” findings about this particular tragedy were almost novelistic in their detail. The report has received extensive press coverage, for instance this in the Post. The report is quoted in full here. From this point on I’ll cite samples from the report’s language, and then suggest what they mean. My purpose is to convey in lay terms what the NTSB has found, rather than to offer hypotheses of my own. I’ll add emphasis in bold at some points for clarity.

---

1) The first attempt at takeoff.

With passengers aboard, the crew made a first attempt at takeoff, which they aborted on the runway just before the plane reached flying speed. The NTSB preliminary report says:

The second-in-command (SIC) reported that during the takeoff roll on runway 2, the airplane accelerated normally, however, he observed that the right primary flight display (PFD) airspeed indicator mis-compared to the left side airspeed indicator and an aborted takeoff was performed.

The pilot-in-command (PIC) slowed the airplane without issue and exited the runway onto a taxiway. The left engine was shutdown, the SIC opened the main cabin door and walked to the front of the airplane where he subsequently observed that the red pitot probe cover remained installed on the right-side pitot probe.

The SIC removed the cover, did not see any damage to the probe, and returned to the cockpit.

During the first takeoff attempt [the plane] reached a maximum airspeed of 104 kts displayed on the left PFD airspeed indicator and 2 knots on the right PFD airspeed indicator before the abort was initiated….

What does this mean? As a plane begins its takeoff roll, pilots watch all the engine monitors, to make sure they’re producing power as expected. And they watch the airspeed indicators, to be sure the plane is accelerating as it should—and that the airspeed indicators are working. In our little propeller airplane, “engine alive” and “airspeed alive” are two of the normal calls you make in those first few seconds.

The Challenger pilots quickly noticed that something was wrong with their airspeed indicators. One showed that the plane was going 104 knots, just about fast enough to take off. The other showed 2 knots—essentially, standing still.

How could this be? Apparently neither pilot had noticed that the right-side “pitot” tube still had its red protective cover on, presumably with a red “Remove Before Flight” ribbon. Here is what pitot tubes and their covers look like, on the same kind of plane involved in this crash.

Courtesy of Bruce’s Custom Covers and Aircraft Covers. Inc., a photo of a Challenger 300 jet with pitot covers. The covers involved in the recent tragedy may have been different models. But normally any such cover is red and has a “Remove Before Flight” ribbon.

As explained here, pitot tubes, pronounced like Frito or Cheeto, are the way pilots know how fast a plane is going. They have tiny inlets pointing directly forward into the wind. If the tube’s opening is plugged up or iced over, or still housed inside its red cover, the pilots can’t know the plane’s airspeed.

Why does this matter? Because accurate readings of airspeed are fundamental to nearly everything else in aviation. Are you going fast enough for takeoff? Slow enough for landing? Fast enough to avoid a stall while banked for a turn? And so on. The Air France crash over the Atlantic began with a pitot tube problem.

Everyone who has dealt with airplanes has missed things on pre-flight checklists, and even on the final “to-be-sure” walk-around look at the airplane. You’re more likely to miss when you’re rushed, or under stress, or distracted. For whatever reason, two pilots who were taking the company’s CEO and his family home, in the company-owned jet, properly removed the red pitot cover on one side of the plane but failed to do so on the other.

---

2) A warning before the second takeoff.

The report says:

Shortly after the left engine was started [after the aborted takeoff], the crew reported that an Engine Indicating and Crew Alerting System (EICAS) advisory message of ‘RUDDER LIMITER FAULT’ annunciated.

The PIC reported that he attempted two ground avionics “stall tests” to clear the message, as he had received this advisory message in past ground operations, however, the tests did not clear the annunciation. The flight was continued given that the message was an advisory, and not a caution or warning.

I’ve never flown this kind of plane and don’t know its systems first-hand. But in any modern avionics system, some of the info you see on startup can be routine, much like a “service due soon” light on a car’s dashboard. Some are life-and-death serious.

A difference between cars and airplanes is that judging what’s serious isn’t just a personal call. Every airplane has a formal “minimum equipment” list and other go/no-go guides to systems that must be working if you want to take off. We’ve all been on airline flights where the captain announces that something has shown up in the pre-flight checks, and the plane can’t go anywhere until a mechanic comes aboard to resolve it. That’s not just because the captain has a “bad feeling.” It’s the no-go list.

For some airplanes, a “RUDDER LIMIT FAULT” warning, like the one the crew described to the NTSB, would mean no-go. I don’t know about the Challenger. The NTSB will tell us.

---

3) An omission on the second takeoff.

The report says:

The flight crew further reported that during the second takeoff, the acceleration was normal, however, the SIC noticed that the V-speeds were not set. The SIC called V1 and rotate at 116 knots from memory and the PIC entered the climb without issue.

What does this mean? “V” speeds (from French vitesse) are the building blocks of flight procedure. There are dozens of them, as you can read here. Vy, for the fastest rate-of-climb. Vx, for the sharpest angle-of-climb. Va, the “maneuvering speed” when entering turbulence. Lots more.

And V1, which you can oversimplify as “the speed of no return.” When a plane passes V1 on its takeoff roll, it’s too late to call things off. There’s not enough runway to slow down. The only option is to go faster, and go up. For this flight V1 was 116 knots.

For professional flight crews, usually one pilot would call out “V1” to the other when the plane reached that speed. That means, “we’re committed to take off.” And a usual part of the pre-takeoff checklist would be to set a visual indicator, known as a “bug,” on the airspeed gauge. It’s one more cue for the flight crew that they have crossed this crucial threshold.

The NTSB is saying that, for whatever reason, the crew didn’t take the standard step of setting an indicator for V1 on the pilot’s display. Instead one of the pilots called out V1 “from memory.” Pilots need to do countless things by memory. The reason for aviation’s checklist discipline is to guard against distraction and oversight.

(Both pilots were highly experienced overall, with many thousands of hours of flight time. In this model of airplane, one had 88 hours of experience, and the other had 78 hours. This is not a lot.)

---

4) The real trouble begins.

The report says:

The flight crew reported that around 6,000 ft, they observed multiple EICAS caution messages. The crew recalled EICAS messages of ‘AP STAB TRIM FAIL’ ‘MACH TRIM FAIL’ and ‘AP HOLDING NOSE DOWN’. Neither crewmember could recall exactly what order the EICAS messages were presented. They also reported that additional EICAS messages may have been annunciated.

These messages all related to the plane's "trim" — the settings on the plane's control surfaces, especially those that point the plane's nose further up or down as it flies. The most important message is “AP HOLDING NOSE DOWN.”

Why did this matter? Details are below, but essentially it meant that the plane’s autopilot (“AP”) was applying unusual force to keep the plane’s nose from pitching up too steeply.¹

The report says what happened then:

The PIC asked the SIC to refer to the quick reference handbook. The SIC, via an electronic flight bag (iPad), located the quick reference card and the ‘PRI STAB TRIM FAIL’ checklist. The SIC visually showed the PIC the checklist, and they both agreed to execute the checklist.

The first action on the checklist was to move the stabilizer trim switch (‘STAB TRIM’), located on the center console, from ‘PRI’ (Primary) to ‘OFF.’ The SIC read the checklist item aloud and he subsequently moved the switch to off.

As soon as the switch position was moved, the airplane abruptly pitched up.

Turning off the trim switch also disconnected the autopilot. The instant the autopilot’s force was removed, the plane was “out of trim” and its nose dramatically shot up. Imagine one side in a game of tug-of-war suddenly letting go. The other side would fall backward. Something like that occurred. What had been balanced forces—in this case, nose-up force by the trim, nose-down from the autopilot—were instantly out of balance, and mainly on one side.

The report says:

The airplane immediately pitched up to about 11° and reached a vertical acceleration of about +3.8g. The airplane subsequently entered a negative vertical acceleration to about -2.3g.

The airplane pitched up again to about 20° and a vertical acceleration of +4.2g was recorded.

These violent up and down forces are more than virtually any passengers have ever experienced aboard an airliner. Here is the way one online rendering² of ADS-B data depicted the sudden changes in vertical speed:

What to notice here is the spatter of blue dots, within the marked purple rectangle. The dots represent every-few-seconds changes in the plane’s vertical speed. It had been climbing at a steady rate of between 2,000 and 3,000 feet per minute. Instantly it shot up above 6,000 feet per minute, or 100 feet per second. Then plunged back down. That rectangle covers about two and a half minutes in real time.

These extreme forces might have damaged the plane itself, since they exceed the “load factors” most planes are designed to withstand. Certainly they did lethal damage within the plane.

The preliminary report does not specify where the passengers were within the cabin, or whether they were wearing seat belts.

---

What we know, and don’t.

Final reports take a long time; we don’t know what else the NTSB investigators will find or conclude.

We can’t know whether the early missteps—overlooking the pitot-tube cover, failing to set the V1 indicator—had any significance. Or whether the pre-takeoff “RUDDER LIMITER” warning was related in any way to the later problems. Or what else the flight crew could have done when things went wrong in the sky. Or whether this is a horrific case-of-one, versus a symptom of broader systemic issues.

We know that it was a tragedy, about which the NTSB has been more detailed than normal at this stage.

Deepest sympathies to Dana Hyde’s family and her many admirers and friends.

1. In short: pointing a plane’s nose more steeply up, or more steeply down, will affect both its airspeed and its rate of climb or descent. For any given power setting, there’s a certain nose angle that will give the right combination of speed and climb (or descent, or level flight). Pilots determine the nose angle by pushing forward or pulling backward on the controls. This affects the plane’s horizontal stabilizer—its “tail”. As the tail goes down, the nose goes up, and vice versa.

In this case, something in the “trim” settings of the tail was pushing down too hard, pulling the nose up too much. In order to fight this force, and keep the plane at its desired speed and climb rate, the autopilot was working extra hard to override the “trim.” That is what “AP Holding Nose Down” meant.

But when the autopilot was turned off, the plane was suddenly “out of trim,” and the nose immediately shot up.

2. The site where I originally found this image no longer seems to be live. But this ADS-B data and graphing are consistent with other reports.