What We Know, and Don't, About the Air India Crash.
Several alarming possibilities have receded. One that's alarming in a different way has emerged. A guide to what investigators have learned, and what news the flying public should pay attention to.
In Mumbai, neighbors gathered at a service for Air India captain Sumeet Sabharwal, who was in command of Air India flight 171 when it crashed and killed more than 260 people last month. Why did the plane go down? The laborious aviation-world investigation process has now focused on Capt. Sabharwal and his First Officer on that flight. (Photo Ashish Vaishnav/SOPA Images/LightRocket via Getty Images.)
I haven’t written about the horrific Air India disaster, because for weeks there was only guesswork to go on. Now things have moved past that stage. The purpose of this post, with awareness of everything else going on in the world, is to summarize revelations to date, and offer a guide to information still to come out.
As I’ve noted whenever this subject comes up, the exceptional safety of modern airlines means that when something does go wrong, the causes usually start out as a mystery.1 That is because nearly every non-mysterious, “normal,” source of risk has already been removed. In the medical world, the adage is: When you hear hoofbeats, think horses, not zebras. Aviation is left mainly with the zebras.
Finding out what kind of zebra can be very slow. It’s often years before the NTSB or its foreign counterparts issue a “Final Report,” tracing accident causes, and recommending procedural or regulatory improvements. But these agencies are expected to issue preliminary reports within 30 days of a serious crash. Sometimes safety issues are apparent even then, and regulators issue emergency orders even as the full investigation grinds on. (The best known recent example was in 2019, when two mass-fatality crashes in Boeing 737MAX airplanes led to a worldwide grounding of that fleet lasting nearly two years.)
Last week Indian aviation authorities released their preliminary findings about the crash, which occurred in Ahmedabad on June 12 and left a total of 260 people dead. As press coverage has noted, the report left many questions unanswered but also contained significant news. You can download a PDF of the 15-page report from the web site of India’s Aircraft Accident Investigation Bureau (AAIB) here.
Until the report came out, the accident story resembled a classic locked-room murder mystery tale. Obviously something had brought the plane down. But the most logical “possibilities” all seemed… impossible.
Although the AAIB preliminary report doesn’t say this directly, its evidence points strongly to a possibility barely discussed in the first days after the crash. That is “suicide by pilot”: for whatever reason, someone in the cockpit intentionally flying the plane into the ground. [Update: This is the standard aviation-world phrase. I imagine it was derived from the phrase “suicide by cop,” for someone who defies law enforcement in a way likely to lead to his own death. The aviation situation is different, and as several readers note in the comments, a more appropriate term would be “murder-suicide” or simply “mass murder.”]
If so, it would be far from the first such case. My friend William Langewiesche, the renowned aviator-author who died last month, wrote a celebrated Atlantic story about the crash of Egypt Air 990 in 1999, which Langewiesche argued must have been intentional. Ten years ago, the young pilot of a Germanwings airliner flew it and the 150 people aboard straight into the terrain of the French Alps. There have been similar cases on airlines in Japan, China, Indonesia, and elsewhere.
This possibility—still unproven, and quickly disputed by Indian (and US) officials and pilot groups—is obviously terrifying on many levels. It suggests a different kind of safety-system challenge than the software issues that grounded the 737MAX fleet. The purpose of the rest of this post is to sketch out how investigators and the aviation community considered other hypotheses before moving toward this one; which information in the Indian preliminary report seemed significant; and how the flying public should think about this aspect of air safety.
What happened to the plane, in simplest terms?
At around 1:30 in the afternoon of June 12, Air India flight 171 taxied for takeoff from Ahmedabad airport in India, bound for London Gatwick. Everything was normal. The winds were light. The skies were clear. The day was hot—96F, which makes it harder for planes to climb—but with the airport’s long runway and the plane’s within-specs total load, that was not an operational risk. The plane was a Boeing 787 “Dreamliner,” one of more than 1,100 such planes in airline service around the world. In the dozen-plus years since the Dreamliner’s introduction, it had never had a fatal crash.2
According to the AAIB preliminary report, the plane began rolling down the runway just four seconds after it was cleared for takeoff. Less than a minute after that, it had reached its “Vr” speed of 155 knots, or about 178 miles per hour. Vr is “rotation speed,” when a flight crew can start pulling a plane’s nose up into the air. Three seconds after that, the plane’s sensors reported that its wheels had left the runway. And three seconds after that, it had accelerated to 180 knots, or 207 mph, and already climbed a few hundred feet above the ground.
At that point—three seconds into becoming airborne—everything changed. The plane stopped climbing. The now-famous video filmed by 17-year-old Aryan Asari showed the plane’s path shift, from a normal-looking departure to something like a power-off glide.
As the world now knows, after those first seconds the plane continued a controlled-looking glide, right into a hospital not far from the airport. All but one of the 242 people aboard the plane were killed, along with at least 19 on the ground. This made it the highest-fatality airline incident worldwide since the Malaysian Airlines flight from Amsterdam to Kuala Lumpur that Russian missiles brought down over Ukraine 11 years ago, killing all 298 people aboard.
Where did speculation lead, at the start?
I followed the news reports and pilot-discussion boards after the accident. Here were some possibilities mentioned in the first few days:
-Bird strike. Very few things can stop powerful modern jet engines. Running suddenly into a large flock of birds is one of those few. Think of Captain Sully and his bird-strike-induced “Miracle on the Hudson.”
-Contaminated fuel. This problem is rare but has happened. The danger is that with bad fuel the engines might work well enough while the plane is taxiing but then fail when it matters, during a climb.
-A mix-up in the cockpit, by a pilot who happened to grab the “retract flaps” lever, rather than “retract landing gear.” Almost as soon as an airliner’s wheels leave the runway, pilots retract the landing gear—the wheels and accompanying hardware. That is to reduce drag on the plane during its initial climb, and because having wheels-down doesn’t help you once a plane has moved beyond the runway (as it does very quickly).
But unlike the landing gear, the “flaps”—which pilots lower during both takeoff and landing, for reasons explained here—should stay extended until a plane is further up in its climb. If they’re retracted too early, or not correctly set before takeoff, they can make it hard or even impossible for an airplane to climb properly.
Could that have happened here? Could the crew have neglected the right flap setting before takeoff? Or somehow grabbed the wrong lever and changed its configuration, by mistake, at a crucial time?3 For several days aviation forums were full of discussions on this possibility.
-Weight/balance, or gross-weight issues: The two General Electric GEnx engines on this 787 are very powerful. But even on big planes, crews need to be conscious of total takeoff weight, and how that weight is distributed fore-to-aft in the airplane. On such a hot summer day, could this have been a factor? Or could some other maintenance or inspection error have kept the plane from climbing?
-Something fundamentally wrong with the 787 itself: Some problems are discovered only when they make an airplane crash. The 737MAX is the most famous example. Despite its crash-free record so far, might that also prove to be true of the 787? A defect in its electronics? Its automated systems? Its engines? Something else?
Tracing the origin to a 787 design flaw wouldn’t make a difference to the hundreds who perished. But, as with the 737MAX, the ripple effects on worldwide aviation could be immense. They would start with Boeing but would not end there. How long would it take before this workhorse of world travel was considered airworthy again by regulators? Would it ever seem safe again, for the flying public?
-‘Low-altitude capture’ and other outlier possibilities. Airliners spend most of a flight on autopilot. Some rare disasters have been traced to “low-altitude capture”—a plane locking onto autopilot guidance that makes it level off far too early in a climb. Normally a crew would be “hand-flying” these first few minutes of climb, rather than switching to autopilot. But could this have happened? People could not initially rule it out.
-Terrorism or explosion. Almost nothing can be ruled out these days.
Why people began to consider these initial possibilities, impossible.
Here are some of the reasons:
-The straight-ahead nature of the plane’s brief flight. The grainy video of the plane’s departure, by the local teenager Aryan Asari, assumed greater and greater significance in the first week after the crash. The most important thing about it was documenting the plane’s steady, straight-ahead flight path, during its few seconds of normal climb and then during its also-brief glide toward doom.
Why did it matter that the flight was straight ahead? Because the constant, “no yaw” nature of the aircraft’s path, with no deviation right or left, suggested that whatever happened to the plane, most likely happened to both engines at once.
The Dreamliner has two engines, one under each wing. If only one of the engines had been affected, the plane could have kept climbing. (Though it would then have circled back for an emergency landing, rather than continuing the flight.) Modern engines are that powerful, and airline certification standards are that rigorous. But it would at least momentarily have veered to the right or left, because of the asymmetric thrust.
Because the plane’s path was so steady, and as other details emerged, attention shifted away from some initial scenarios, and toward those that could have affected both engines at once. This meant:
-Bird strike: unlikely. Could birds have completely taken out both engines at the same instant? Conceivable, but not likely. Also, close examination of the videos showed no sign of bird flocks. Nor were there reports of them from other flights, or remains of them in the crash debris. (After a bird strike, such remains are obvious on a plane.)
-Fuel contamination: also unlikely. If the airport had a batch of bad fuel, presumably it would have affected some other airplanes too. But no other flights reported fuel problems. Could one plane have chanced to get the only truckloads of bad fuel? Conceivable. But again not likely.
-Flap-and-landing-gear confusion: unlikely. As mentioned, at first this seemed one of the least-unlikely scenarios. It was even mentioned in some mainstream news reports. It was strengthened by screen shots from the takeoff video that seemed to show “clean wings” on the 787 as it was departing—that is, flaps not properly extended for takeoff.
But later examination of the plane’s wreckage found clear evidence that the flaps had been in the correct position. They just hadn’t shown up in the camera angle for the brief video. And 787 pilots and experts pointed out that even if a crew member had grabbed the wrong switch, the plane wouldn’t simply have stopped flying, as it did.
-Explosion: no. Zero evidence in the videos or at the crash scene to suggest this.
-Other maintenance or operations issues: unlikely. No reason emerged to suspect this. The same with “low altitude capture” and some other remote possibilities.
—Larger design or safety flaws for the 787: possible but not evident. The investigation goes on. But in contrast to the 737MAX situation, so far no publicly-known evidence, or past pattern of problems, seemed to suggest fundamental flaws. And as we’ll get to in a moment, the Indian authorities had a very significant statement on this point.
And what about this ‘RAT’ deployment?.
This point is inside-baseball, but worth mentioning since it’s cited so often in discussions of the crash.
RAT is shorthand for Ram Air Turbine. The RAT is essentially a tiny windmill that automatically and immediately descends beneath the plane when other power has been lost. Its little propeller turns in the airflow wind, and generates emergency power for the plane’s electric and hydraulic systems. It’s roughly like a hospital’s emergency generator, which immediately kicks in so a surgical team doesn’t lose power in the middle of an operation. Its very existence is another sign of aviation engineers and regulators asking themselves: “Well, if Plan A and Plan B and Plan C all fail, what about Plan D?” On this scale, RAT would be something like Plan Q.
Videos of the brief flight, and recovered cockpit data plus debris analysis, all showed the RAT automatically deploying three or four seconds after liftoff. This mattered as one more piece of evidence that something caused the airplane’s normal power, from its two engines, to shut off at exactly that same time.
But what could that something have been?
What changed with the initial Indian investigative report.
Last week’s “Preliminary Report” from the AAIB (again, India’s NTSB), was explicitly preliminary. Evidence is still to come, and conclusions might change. But in two ways it significantly changed assessments of the possibilities.
One was a dog that didn’t bark—an element the report did not contain. That was any suggestion that the company making the 787 (Boeing) or its engines (GE), or the scores of airlines flying that aircraft (Air India and many others), needed to change their safety practices because of this crash. Again for comparison: At this stage in the 737 MAX investigations, emergency warnings were already going out to other airlines around the world. At this stage in the NTSB’s look into the recent fatal collision over the Potomac, authorities had already changed helicopter routes in the area.
It’s hard to prove a negative, especially in a preliminary report. But if AAIB and other investigators, including a US team from the NTSB, had reason to think the airplane itself was the problem, they likely would have said so. Jon Ostrower, of the authoritative publication The Air Current, gives details on why AAIB’s silence on this point matters.4
The other, more explicit news in the AAIB report was the cockpit data, both digital and aural, recovered from flight recorders.
Keep reading with a 7-day free trial
Subscribe to Breaking the News to keep reading this post and get 7 days of free access to the full post archives.