Plane Crashes

And so on. Actually this sort of thing goes way, way back. When a Northwest Airlines DC-9 crashed on takeoff in Detroit in 1987, the one survivor was a 4-year-old girl. A miracle, naturally — though in the pre-Internet era this invocation was less permeating and therefore more tolerable.

As I wrote the other day, for one to escape alive from an airplane accident is hardly surprising or unprecedented. Statistically, most crashes have survivors. This includes some of history’s blackest disasters. To choose just two: Sixty-one people survived the Tenerife collision in 1977, and four survived the horrific JAL crash outside Tokyo in 1985 — history’s worst and second-worst disasters, respectively. The idea of a lone passenger making it out alive might seem particularly unlikely, as is the inverse, of everybody emerging unscathed, but as we see, it has happened several times. Luck, not some spiritual hand, is often the potent factor.

This same inverse is part of what makes the “miracle” phenomenon so peculiar and annoying. Is it miraculous that nobody was killed, or that everybody was killed save for a single soul? How many fatalities (or lack of them), exactly, separates the miraculous from run-of-the-mill happenstance? How about those survivors at Tenerife and in Japan? Somehow, numbers like “61″ and “4″ don’t inspire such wondrous exaltation.

We’re perhaps not being asked to take the term literally — to believe that some supernatural force or deity has lent its power. But to those who see it that way, consider for a moment the odd benevolence of a deity who saves one at the expense of dozens, or hundreds of others. If those children who were pulled alive from the wreckage of Yemenia and Afriqiyah, et al., were indeed saved by miracles, what of those who perished? Can we chalk up their fate to some reverse miracle? Thus, the pretense of supernatural intervention is not, by itself, the main problem, but rather the way in which it is alleged, whereby one person’s miracle is everybody else’s curse.

This trend isn’t going away, probably. At this point the M-word has become almost as pervasive in American media culture as the H-word (hero). In the US Airways Hudson incident, both of these terms were overplayed to a nauseating extreme.

Maybe this bothers me more than it should, but it seems indicative of, among other things, the devolved state of news reporting and a sort of continued infantilizing of the American consciousness. Aside from the contradictions already cited, it’s exasperating that we are so inclined to credit the supernatural rather than credit, say, the design of the aircraft or skill or the crew. By the same token, much as we’re thankful for this unseen hand when the outcome is something fortunate, we are eager to blame ourselves when it’s not. It’s tedious, hypocritical and offensive, not to mention insulting to the thousands of victims of air crashes who weren’t so, well, lucky.

- – - – - – - – - – - -

Do you have questions for Salon’s aviation expert? Contact Patrick Smith through his website and look for answers in a future column.

I was flying back from Europe that afternoon, and my plane was one of the first to land after JFK reopened to traffic. Smoke was still rising from the crash scene. Only a few miles beyond the gray plume I could see the vacant chunk of sky that had been occupied by the World Trade Center.

Flight 587 was well known among New York City’s Dominican community. Merengue star Kinito Méndez paid a sadly foreboding tribute with his song “El Avion” in 1996. “How joyful it could be to go on flight 587,” he sang, immortalizing the popular daily nonstop.

This was a catastrophe, to be sure. But we’ve since reached an important, somewhat remarkable milestone: The crash of Flight 587 was the last multiple-fatality crash involving a major U.S. airline.

There have been several terrible accidents involving regional planes — all of which have been discussed in this column, from the Air Midwest crash in 2003 to the 2006 Comair crash at Lexington, to the Colgan disaster outside Buffalo in 2009. And in 2005 a young boy in a car was killed when a Southwest Airlines 737 skidded off a snowy runway at Chicago’s Midway airport. Yet amazingly, an entire decade has passed since the last large-scale crash involving a mainline U.S. carrier. Somewhere on the order of 5 billion passengers have flown aboard the country’s biggest airlines in that span, aboard some 35 million flights.

Ten years is a record unsurpassed in virtually the entire history of U.S. commercial aviation.

Absence of a headline tragedy does not indicate an absence of problems, of course, and about the worst thing we could do is rest on our laurels. But despite unprecedented public contempt for the legacy airlines, and despite the fiscal devastation they have endured over the past decade (five bankruptcies and counting), they’ve nevertheless maintained a nearly perfect safety record.

How we got to this milestone is mainly the result of better crew training and, perhaps to a lesser degree, better technology. We’ve engineered away what used to be the most common causes of accidents.

Yes, we’ve been lucky too. But mostly we’ve been good.

It remains to be seen how the media will note this anniversary, if at all. And when finally this streak comes to an end — and let’s be realistic, it has to end at some point — what then? The worst thing about this inevitable crash will be the loss of life. The second worst thing will be overreaction and a lack of context. Ten years forgotten. Because plane crashes, not a lack of them, are the big news.

And in doing so, either I’m a lousy wordsmith or my audience is a lot more stubborn than is healthy or reasonable. Because as hard as I try, I just can’t put the skeptics out of their misery.

“Well, you are an airline pilot,” emailed one reader, whose doubt was so palpable I could feel it oozing through the screen. “So I suppose we should take your word for it as to how planes are flown.”

No, really. Take my word for it.

Most readers aren’t as cynical (or insufferable), and are willing to trust me — though not always fully.

Which is fair, I guess. Despite my best efforts, I’ve probably left some questions unanswered — and others not answered with the conviction they deserve. That’s the way it will have to remain, I’m afraid. Trying to explain the art and science of flight to a layperson is an extraordinarily difficult thing. And there is only so much that I have the time, the patience and, frankly, the talents to make clear.

And no matter; people like Missy Cummings, with their splashy credentials and even splashier ideas, manage to reach a far broader audience than I do. It’s a losing battle. My only comfort, I suppose, is knowing with 100 percent certainty that none of you will be stepping on board a remotely piloted jetliner at any point in the foreseeable future.

One more shot, though, before letting this go. The challenges — and my responses — below are from a letter by “dizuel,” posted in the Salon forum following my column on Aug. 4. The letter does a good job of encapsulating that lingering sense of incredulity. My responses, cranky and impatient as they are, may or may not be convincing. I’ll leave that up to you. If nothing else, it’s an addendum to my earlier rantings on the topic. 

 ”I think you have established that aircraft are not flown automatically today. That the capability for autoland exists and is functional, but rarely used. But you have not established why aircraft could not be flown without pilots someday. If the capability exists, then why could some future version not handle takeoffs and landings on a regular basis?” 

For one, because this isn’t about takeoffs and landings. For all intents and purposes, takeoffs and landings are the easy part. The hard part is dealing with all of the many contingencies, large and small, that arise during any commercial flight. As I’ve said before, a flight is a very organic thing. For all of its scripted protocols — checklists and SOP — hundreds of subjective decisions are made by the crew, from deviating around a cumulus buildup (how far, how high, how long) to troubleshooting a mechanical issue to handling an on-board medical problem. I’m not talking about Sully-esque emergencies, but the run-of-the-mill situations that arise every single day, on every single flight. I cannot fathom the idea of these things being handled remotely.

I know that it works for military drones and unmanned aerial vehicles. These machines are very sophisticated, obviously. But a drone is not a commercial jet carrying hundreds of people. It has an entirely different mission, and operates in a wholly different environment — with far less at stake should something go wrong. You don’t simply take the drone concept, scale it up, throw in a few redundancies and off you go.

Adapting the UAV model to the commercial realm would require, in addition to gigantic technological challenges, a restructuring of the entire commercial aviation infrastructure, from airports to air traffic control. We’re talking hundreds of billions of dollars, from the planes themselves to the facilities they’d rely on. We still haven’t perfected the idea of remote control cars, trains or ships. The leap to commercial aircraft would be harder and more expensive by order of magnitude.

And for what? You’d still need human beings to operate these planes remotely. Thus I’m not sure what the benefit of this would be in terms of cost.

“Take the A380 engine failure some months ago. The crew did an amazing job saving that airplane. Could that have been done remotely? You could argue it could have … Remote cameras can provide visuals of any part of the aircraft. And, since they are not actually on the plane, the crew may be able to react in a calmer and more calculated manner.”  

They also may not get a full and realistic sense of what, exactly, is happening — not to mention the complexities of dealing with cabin crew and hundreds of passengers. Again, these situations tend to be organic and subject to all sorts of subjective decision-making — much of which would be apparent from afar. If you ask me, about the easiest way of turning a minor problem into a catastrophe would be to move the crew thousands of miles away. 

“It is the ugly side of progress — many valuable skills from the past are being commoditized. It sucks. But it’s life. The reason why piloting went from being a high-paying job to a low-paying job is because it has become easier. More people can do it, and thus the supply has overwhelmed the demand.” 

This is false.

Sure, modern automation means there’s a reduced workload in certain regimes of flight than there used to be. But, if you’re going to tell me that flying a contemporary Boeing or Airbus is “easier,” in terms of required expertise and knowledge, than flying earlier generation planes, you’re simply wrong. On the contrary, I submit that it’s considerably harder once you account for all of the operational aspects of modern flying — not merely the hands-on aspects of driving the plane and running its systems, but familiarity with everything else that a flight entails: planning, navigating, communicating and so on. The emphasis is on a somewhat different skill set, but the volume of requisite knowledge is greater than it used to be.

And the reason piloting became a lower-paying job (in places) is mostly a function of bottom-line economics. Pilots once earned more because the companies they worked for used to be a lot more profitable. The regional airline sector, meanwhile, has essentially tripled in size in the last quarter-century or so, dragging down the average wage substantially. Regional pilots are the lowest-paid, yet in terms of both technical skills and general career challenges, theirs is arguably the most difficult job of all. 

“What you have to provide is a strong argument that flying an airplane is immune to progress and the relentless advance of technology.” 

No, I don’t. Why is elimination of the pilot the inevitable endpoint of this technological evolution? I’ve never understood that. Are modern medical advances striving to eliminate surgeons from operating rooms? What exists in the cockpit today is already a fine example of how progress and technology have improved flying — making it faster, far safer and more reliable than it once was. But it has not made it easy, and it is a long, long way from engineering the on-board pilot out of the picture — something we needn’t be looking for in the first place. I am not against the advance of technology. I am against foolish extrapolations of it.

And with that, I declare this discussion over. At least until the next annoying Discovery Channel segment or Popular Science feature gets the whole thing foaming again. 

- – - – - – - – - - 

GO-AROUNDS 

Re: Air France 447 investigation 

Below are some thoughts from an Airbus A330 captain (he wishes to remain anonymous), tackling some common presumptions regarding what went wrong aboard Air France Flight 447 two years ago. This is the kind of technical commentary I normally stay away from, but it represents a point of view the media has hardly touched on during its analysis of the most recent findings: that perhaps the engineering of the A330 itself played an important role. This is by no means to imply the A330 is an unsafe aircraft. However, focusing solely on pilot error is remiss, and neglects to consider the complexity of the situation — and possible idiosyncrasies of the airplane. And with our just-concluded conversation in mind, somewhere in here, perhaps, we see the dark side of putting too much faith in automation. (I’ve edited the letter for clarity.)

We know the airplane stalled, but the interim reports do not detail how the Air France pilots reached this point in the first place. No way do I believe that the pilots manually commanded an extreme nose-up input, as the report is claiming. To say that a pilot would, for an “unreliable airspeed event,” initiate a 7,000-foot-per-minute climb, with 16 degrees of nose-up input, is crazy. 

But the electronic flight control system (EFCS) of the A330 is capable of generating this magnitude of performance on its own, if, say, the overspeed protection mode was activated by the blocked pitot probes. This is what the A330 simulator displays when this fault is inserted. The auto trim also runs up to about 13 degrees of nose-up trim — the same figure mentioned in the interims — and reduces the nose-down authority for the resulting stall. 

The pilot cannot override the EFCS quickly when it is misbehaving. The checklist procedures for this are time-consuming and confusing. The “unreliable airspeed” checklist is also ineffective if the EFCS takes control of pitch, as the pilot is locked out. 

Meanwhile, operators have since received a “reminder” from Airbus regarding re-engagement of an autopilot after an unreliable airspeed event, because the aircraft can experience a pitch-up command. Maybe that explains how the nose got so high? After all, it is instinctive for pilots to try to re-engage a second or third autopilot if one for whatever reason disengages, as it did on Flight 447. 

Additionally, the plane’s warning systems were presenting a flood of fault messages, aural chimes and alarms; at the same time the pilots were confronted with impending loss of control due to high angle-of-attack, and stall, at night, in bad weather, without knowing which instruments to trust with their lives. 

This is pushing the limits of what a pilot can synthesize and react to.

- – - – - – - – - – - -

Do you have questions for Salon’s aviation expert? Contact Patrick Smith through his website and look for answers in a future column.

Onward to more timely things …

Later in the week I’ll be getting to the Air France 447 findings. In particular, I’ll be looking at the media’s garbled interpretation of them. European investigators released their report on July 29, and I’ve received a blizzard of emails from a confused public. Here’s some advice for the time being: Ignore pretty much everything you see or read about this. What they’re telling you about the report isn’t what the report actually says.

In the interim, let’s have a look instead at something that happened over the weekend down in Georgetown, Guyana.

The other morning, a Caribbean Airlines Boeing 737-800 overran the runway at Cheddi Jagan International Airport, sliding to a stop and cracking in half. There were no fatalities, but the jet came perilously close to sliding down a steep embankment.

I’ve flown into Cheddi Jagan. Located in the town of Timehri, it’s an immaculate little airport surrounded by lush rain forest, about an hour’s drive from the capital.

There is no radar at Cheddi Jagan. Controllers rely on position reports from pilots — altitude, distance and bearing from the airport’s VOR station — as a means of sequencing traffic. The main runway, numbered 06/24, is a relatively short strip of just over 7,400 feet. There are steep drop-offs at both ends, which can make the runway appear like a sort of inland tropical aircraft carrier, hovering over a sea of vegetation. The airport is prone to morning fog, and lacks an instrument landing system (ILS) — what we call a “precision approach.” Pilots rely instead on a simpler VOR or GPS procedure with parameters that aren’t as tight. These lack the strict vertical guidance of an ILS, and your decision point is slightly higher, and slightly farther out, than it would be with an ILS.

All of that sounds kind of scary, and these are the aspects that the media, both internationally and in Guyana, have been harping on. Which is at least partly unfair.

Taken together, these facets indeed make the airport challenging and less forgiving than others, but they do not make it unsafe by any stretch. It is unclear if any of these characteristics played an important role in the accident. Operating without radar, Georgetown’s controllers have an excellent reputation for their handling of traffic. As for that non-precision, so long as you’re in the slot, so to speak, touching down from one of these procedures should use up no more runway than landing from an ILS. And no approach, ILS or otherwise, should be continued if things become unstable.

With respect to the short runway, It’s logical to assume that had it been longer, the jet would not have gone barreling off the end. And sure, on some statistical level, a shorter runway is less safe than a longer one. But that is not the same as unsafe, and not every runway can be 15,000 feet long. Some will always be shorter than others, and at 7,400 feet, Timehri’s 06/24 is roughly the same length as those found at LaGuardia, Washington-Reagan, Chicago-Midway and various other high-density airports, from which thousands of flights operate daily without incident. Whether a given strip is long enough to land on is not a matter of guesswork; performance is guaranteed based on weight and surface conditions. At least on paper there is always room enough to come safely to a stop, and that distance includes a buffer. 

I’m reminded of those infernal “world’s most harrowing airports” lists that surface from time to time, under slightly different headlines, bouncing around the Web and needlessly scaring the crap out of travelers. Do not pay attention to these. They are always exaggerated, misleading and otherwise full of nonsense. If an airport were even remotely dangerous, no airline would be operating there.

In other words, there is nothing about the Georgetown airport that said this accident had to happen.

However, there was limited margin for error. Reportedly the flight touched down in heavy rain, in darkness. The possibilities that jump out at me would include a microburst — a form of wind shear associated with thunderstorms — or an unstable, inadvertently long landing on a wet runway in rainy, perhaps windy conditions. Did the crew fly too close to violent weather? Did they come in too high or too fast? Was there a tailwind? Was there a malfunction in the plane’s anti-skid or spoiler systems? Or was it a combination of things? With incidents like this, usually there is more than a single, simple cause.

Runway overrun incidents are on the rise worldwide. Looking through the Ask the Pilot archives, I’ve covered at least half a dozen, including the 2005 Air France near-disaster in Toronto,  and the Southwest accident in Chicago later that same year.  On one hand, this is alarming. On the other hand, global air traffic has doubled over the past 25 years, vastly increasing the average daily number of takeoffs and landings. Taking in all types of mishaps, the global accident rate — fatalities per passenger-miles flown — has fallen sharply. If nothing else, we have clear focus of where the problem is. And better training, more than infrastructure upgrades or new technology, is the key to fixing it. In the majority of cases, human error is the culprit.

Caribbean Airlines, by the way, is the reincarnation of the old BWIA (British West Indies Airways), based out of Port of Spain, Trinidad. BWIA had been in business for more than 60 years before closing its doors in 2006. Caribbean Airlines started up shortly thereafter, employing many of the old BWIA staff (at substantially reduced wages). Earlier this year Caribbean acquired Air Jamaica, which had been around since 1968. Caribbean Airlines 737s now wear an Air Jamaica sticker on the forward fuselage.

Something went wrong at Georgetown, clearly. And had the 737 gone slightly farther, it would have skidded down that embankment and almost certainly caught fire. But it’s worth noting that BWIA/Caribbean/Air Jamaica have never recorded even a single fatality between them, going back over six decades. Not even Qantas can say that.

Still a perfect safety record. Of sorts.

- – - – - – - – - – - -

Do you have questions for Salon’s aviation expert? Contact Patrick Smith through his website and look for answers in a future column.

Which it did not.

Much is being made, too, about the fact that Flight 447′s captain was not on the flight deck when things began unwinding. “Captain not in the cockpit,” sang one headline. An especially ridiculous NPR headline (later amended), read, “Report: Air France Pilot Resting as Plane Plunged.” Another yelped incredulously at how the “copilot was at the controls.”

Good grief, a copilot at the controls!

As I’ve explained previously in this column, transoceanic flights carry augmented crews, typically consisting of a captain and two first officers — copilots in the vernacular. This threesome rotates through a series of breaks, and for a portion of the flight the captain will be resting while the two first officers are manning the cockpit. This is industry standard and is not the least bit unsafe.

Copilots are not apprentices. They perform just as many takeoffs and landings as captains do, and are qualified to operate the airplane in all regimes of flight, including emergencies. The captain has ultimate responsibility for the flight, and a larger paycheck to go with it, but owing to the peculiarities of the seniority system, it is not uncommon for a copilot to be older and more experienced than the captain sitting next to him.

Not only that, but copilots assigned to relief duty in the captain’s seat must also be fully “type rated,” as it’s called, in the applicable aircraft. This is an extra step to ensure thorough familiarity with the plane and its systems. In other words, the copilots of AF 447 may have earned a somewhat smaller salary, but they carried the exact credentials of the captain. For the press to be harping on this aspect of the accident is unfair and distracting — if predictable.

Otherwise, what we’ve learned from the initial findings isn’t terribly different from what some of us were expecting. The jet encountered terrible weather, suffered a series of systems failures, stalled and crashed.

It’s the “stalled and crashed” part that reporters are seizing on, pointing to a series of erratic — and incorrect — control inputs made by the pilot.

A plane stalls when, loosely put, its wings are no longer able to generate sufficient lift. This can happen for a number of reasons: flying too slowly; flying too fast; banking or pitching too steeply; or some combination thereof. All pilots experience stalls in their primary training, usually in light aircraft where the phenomenon tends to be tame. Stalling in a jetliner, on the other hand, is an extremely serious thing, and a fully developed stall may not be recoverable. For this reason, pilots always fly with a substantial buffer, at speeds and angles that are well clear of a stall onset. Nevertheless, if things go wrong, there is plenty of advance warning of an impending stall — not only aerodynamic seat-of-the-pants cues, but cockpit alarms as well. Should a crew ever find itself in such a position, the recovery technique is piloting 101 stuff, the gist of it being that you lower the nose, not raise it, while leveling the wings and adding power.

How the pilots of Flight 447 found themselves on the verge of a stall in the first place, and why they reacted the way they did — apparently pulling the nose up rather than pitching it forward, creating a series of worsening oscillations that ultimately resulted in a full, non-recoverable stall — is not yet clear. It may never be understood fully.

And we’re reminded of the Colgan Air disaster outside Buffalo, N.Y., in February 2009. In that instance too, for reasons unknown, the crew reacted counterintuitively to an impending stall, resulting in loss of control and a crash.

That was a different set of circumstances, though. In the case of Air France, the pilots’ response to the stall was only one in a chain of events.

It is easy to look at the early report, with its voice recorder transcript and summary of control inputs, and come away with a seemingly clear picture of the crew making a bad situation worse, to the point of catastrophe. They pulled up when they should have pushed down. But was it really that simple? Perhaps they did not react exactly as they should have, but bear in mind the greater context: They were dealing with serious equipment failures — including loss of primary airspeed data — degraded flight controls, in terrible weather, at high altitude, in darkness. I’m surprised they stalled, but not necessarily shocked.

There are no easy answers. People want and expect something succinct — a “cause.” Was it failure of the data sensors? Was it violent turbulence? Was it pilot error? I suspect it was a combination of those things — an unlikely but deadly cascade.

- – - – - – - – - – - -

Do you have questions for Salon’s aviation expert? Contact Patrick Smith through his website and look for answers in a future column.