“We no longer have any valid indications” was the ominous statement from the non-flying pilot soon after the captain of AF447 finally emerged from his resting quarters into the cockpit of the Air France flight AF 447 operated with an Airbus 330, on a scheduled run from Rio to Paris. (underline added)
A routine flight it was supposed to be, but for an unpredictable event – yet to be confirmed – that was going to unleash a chain reaction leading to a deeply disturbing civil aviation accident in recent times.
Such a sudden statement, in the first paragraph above, can be interpreted as an overall and quick assessment of the doomed airliner’s situation which, after reaching an unplanned altitude of 38,000 feet, went into an unrecoverable stall and final plunge in the ocean below, at an abnormal descent rate of about 10,000 feet per minute.
AF447 was trapped, according to various sources, in an aerodynamic condition, perhaps a stall, that the flight crew was unable to correct, and that resulted in a rate of descent equivalent to nearly 120 miles per hour vertical speed, much greater than the Airbus 330’s glide ratio. Something was terribly wrong, but what was causing it? To this day, over two years after the crash, we still don’t know for sure. Many scenarios are being circulated in aviation circles, none conclusive so far. It is now up to France’s Bureau d’enquêtes et d’analyses (BEA) to identify and report on the most plausible triggering event or events that led to the crash of AF447, as well as on ways to prevent such a deadly occurrence.
Why did this accident happen in the first place on a routine long-distance flight? How could this happen on a modern airliner supposedly equipped with the best computerized flight control system available at the time, not to mention the captain and the first officer’s cumulative qualifications and experience?
There seems to have been a total disconnect here between man and machine, perhaps comparable to the one imagined in the acclaimed Space Odyssey 2001 science-fiction novel written by Arthur C. Clarke decades ago. A well trained crew, on the one hand, and an ‘intelligent’ space craft, on the other, working at cross-purposes.
Most readers of this blog would know how simple a Pitot Tube is (also called an airspeed sensor), however sophisticated the airliner it is attached to. Comparing forward dynamic air pressure against static pressure is its basic function.
In the case of AF447, there were three such identical Pitot tubes providing essential information to the fly-by-wire computers and auto-pilot operating the airliner under the watchful eyes of the flight crew. Icing is the suspected culprit for knocking off all three airspeed probes in a matter of minutes.
There were three Pitot tubes feeding information to the three on-board computers. There were three flight crew members in the cockpit when trouble began, each as clued-out as the other, as AF447 was about to start a three minute fatal plunge in the Atlantic Ocean below.
‘Hal’, in the Space Odyssey 2001 story, was the one and single journey control computer, designed on heuristic principles allowing it to follow dutifully preprogrammed mission instructions while learning to adapt to various unplanned situations as the space mission unfolded. In short, Hal was able to correct, without human input, what it perceived to be conflicting instructions. Unfortunately, as the novel shows, Hal’s logic did not always match that of humans. Its logic was preeminent when a man/machine conflict of logic arose, and Hal’s logic prevailed with disastrous consequences, not out of hubris, but simply because it genuinely believed to be acting in the best interest of the space mission.
This arrangement stands technically but not necessarily conceptually, in contrast to AF447’s three fully programmed, mutually cross-checking computers relying on input data from… three identical Pitot tubes, among other essential sensors.
AF447’s computers were not so bold as Hal. When the Airbus’ fly-by-wire system computers could no longer handle the situation, they handed control of the airliner over to the pilots by disconnecting the autopilot and displaying illogical (read: “no longer valid”) data on flight instruments in the cockpit. Quite a double whammy for the flight crew, at a time when the airliner was in a tight area of its flight envelope!
In recent months, a French daily ran an article stating that, had the flight crew followed proper procedures, the Airbus A-330 operated on AF447 would have been flyable without the autopilot. Whether that was the case, is for the BEA to confirm or deny. This might be one of the pivotal issues in the current technical investigation. With time, we’ll learn from the BEA whether the flight crew was adequately trained, if trained at all, to deal with such unlikely situations.
Even if the BEA were to find that the Airbus A-330 operated on AF447 was manually recoverable from an upset at 35,000 feet, the conceptual question will remain as to whether automation turned the table on Airbus designers and on travellers the aircraft was supposed to carry safely to destination.
Could it be that 30 years ago or so, airline pilots caught in a situation similar to the AF447 flight crew would have said “We no longer have any reliable indications?” After all, in those days, there were no such thing as a ‘computer laws’ in the cockpit of airliners. Today’s airline pilots fly with a different frame of mind in many ways because of automation, by far for the better, one would think.
The three Pitot tubes were as good as one another while subject to identical vagaries due to freezing conditions that prevailed in the area where AF447 went down with its precious load of unsuspecting passengers and crew, not to mention their clueless relatives, friends and associates ready to meet them at the Paris airport or at other final destinations.
Months after the AF447 crash, Boeing proudly advertised that its (slow-coming) Dreamliner, the B-787, was NOT flown by computers. Before and after the sad crash of AF447, competition between Airbus Industries (EADS) and Boeing was fierce and still is, witness the 49th International Paris Airshow. It begs the same old question: which way is the man vs. machine interface evolving on trend-setting modern airliners? Is improving the man/machine interface a question of better flight control system design or better pilot training, or both?
Could it be that the three hapless flight crew members on duty on AF447, when the ‘music died’, decades after the publication of Space Odyssey 2001, were totally clued out as to what situation they were in? Situational awareness was of no help? Known emergency procedures were of no help? Cockpit Crew Management procedures were of no help either, even with three pilots in the cockpit instead of the usual roster of two? The void as to what flight conditions hit AF447 is huge for the time being.
What are the designers of the Airbus A-330 thinking to themselves at present? Flight crew error or design error, or a mixture of both? It takes a lot of humility to canvass all three possibilities from the comfort of ground-based offices.
The remaining Space Odyssey 2001 astronauts unknowingly monitored by the eye of Hal, the lip-reading computer, finally managed to shut it down, one logic module at a time. There was no rush in doing so, just method and personal resolve by the remaining astronauts. In contrast, the AF447 flight crew did not have much time to take appropriate action and recover from the high altitude jet upset.
AF447, with its three flight crew in the cockpit, with three minutes to go before the fatal splash-down, from an altitude of about 30,000 feet, did not have the luxury of time. They could not figure out, so it seems, what action to take in their predicament, one that was more difficult to deal with, having to contend with three runaway computers because of three allegedly iced-up Pitot tubes. As AF447 neared the surface of the Atlantic ocean in warmer air, the Pitot tubes started to respond again. By then, it was most likely too late to regain control of the aircraft.
Somewhere in France in the near future, a sort of triumvirate of investigative, industrial and judicial interests will eventually issue findings, positions and accusations over the unjustifiable loss of too many passengers and crew in the world’s most advanced civil aviation systems. At this point, one can only trust they will actually behave as a triumvirate for the betterment of civil aviation safety.
For a more sobber and technical view of man/machine interface issues in the world of advanced airliners, please check this three-year old news item regarding a Qantas Airlines incident.