Air France Flight 447: Not An Open And Shut Case.
I know this is a bit of older news, and there has already been much speculation, but I recently found myself reading a few great articles by the travel blogger Clive Irving, posted on TheDailyBeast.com. Mr. Irving writes about the tragedy of the Air France flight #447, an Airbus A-330-200 that crashed in the southern Atlantic Ocean in 2008. The author has a series of excellent opinions regarding that tragedy, leading up to the publication of the final report. However, his most recent article from last year caught my attention. In earlier posts, Mr. Irving had speculated as to the cause of the crash, based on what he had found prior to the final report. His conclusions in those earlier works seemed logical (albeit nearly all unverifiable), as he looked at the information with a critical eye. In his most recent and final post-accident article on the subject (found here), Irving describes the final report (interpreted/translated by CNN), noting the official conclusions but raising some additional questions that are interesting. His suspicions regarding the politics and timing of the press release are intriguing. But now after reading his latest opinions, I’m even more interested in the technical aspects of the findings. Below are two main questions I have regarding the latest conclusions:
Question 1: Are Large Modern Airliners No Longer Aerodynamic?
Summarizing the accident report, Mr. Irving describes that the main problem was mechanical in nature: iced up pitot tubes relayed bad airspeed data to the flight computer, thus confusing the auto-pilot. (Here’s a CNN graphic on pitot tubes.) However, a ‘leaked’ blame for the crash points to the junior pilots, who were at the controls at the time of the crash. They are blamed for over-correcting because they were not properly trained for that type of emergency (the Captain was resting). I’ll buy that as a possibility, but then the plane is reported to have ended up in a full stall (from the apparent over-correction), and fallen 38,000 ft. with the nose remaining high until impact. Specifically, Irving states “During the fall its nose reared up far more steeply than in a takeoff, and it stayed there.” The data from the flight recorder seems to back this up.
But here’s my question: as far as I know, modern airliners still follow the same basic principles of aerodynamics as a Cessna. (A traditionally designed, stable aircraft always has a tendency to nose down from weather-veining after a stall, given enough altitude.) Sure, anything could fall tail first from a low altitude when there isn’t much time to build air speed, but from 38,000 feet? One possible explanation is that Flight #447 lost its vertical stabilizer, which sheered off (pictured above). That may have neutralized the plane’s natural tendency to turn nose down, but at what point did it sheer off? Wouldn’t it have been after the plane had already begun its rapid descent? If not, and it broke off at low speed, then the aircraft has a serious design flaw; but I doubt that’s the case. Otherwise, and more likely, I would presume that even in the thin air of flight level 380, a plane’s ”rearward’ velocity after a stall could still gain sufficient airflow to force a tail-first aircraft to go back to nose-first before the vertical stabilizer has enough opposing force to break off. Every stall I’ve seen forces a plane to nose-down first. After a stall, a plane may spin, at which point centrifugal force would take over and pieces could break off. But no spin was reported. And really, isn’t it the horizontal stabilizer that plays the biggest role here anyway in a straight stall? But none of that seems to matter here. Of course an exception to this airframe design rule is that some modern military aircraft such as the F-16, are built inherently unstable for high maneuverability and require a computer to keep them flying straight. I’d expect that the A-330, as a people carrier, was most likely designed for stability over maneuverability. So how would a modern A-330, built for flight stability, manage to crash tail first after a stall from such great altitude?
Question 2: Were the pilots that bad or…something else?
So perhaps in response to my first question above, your thought is ‘it could happen’. I would agree that anything is possible, but if the accident report logic is even slightly flawed, then there is room for more investigation. My next question is then: how bad were these junior pilots? The leaked transcript says the stand-ins pulled back and held the yoke in the stall. But I find it implausible that these highly trained commercial pilots’ first instinct would be to pull up in response to a stall horn warning; let alone hold it to no avail for a long time. Now, I found a Youtube video from 1984 showing a fatal crash of a T-19 from inside the cockpit at low relative altitude in Colorado (video here). Apparently the pilot pulled back in a stall and exacerbated the problem to impact. But the T-19 pilot was low and had visible terrain which he was probably trying to avoid, and I don’t know what his training level was. Even if these professionally trained Airbus pilots did the same thing at their cruising altitude of 36,000 feet, they had plenty of height/time to correct, even if their first instinct was wrong.
Another hypothesis based on the report that the pilots received ‘conflicting air speed readings’ was that perhaps the faulty airspeed indicator showed an over-speed condition, and the engines may have been throttled back either automatically or by hand. Of course, if they had been in a stall this would have further exacerbated it, as a power on stall is easier to recover from than a low power stall. But you would still expect the nose to have fallen eventually as the correction is the same in both cases. However, the final report indicates that the pilot(s) pulled back and held it.
So how bad were these junior pilots? Both had hundreds of hours of flight time and even the newest pilot was anything but a novice. Even bad training can’t be that bad if you’re trusted with an A-330 and peoples lives. So then perhaps the real question is how trustworthy is that data? We know there was probably a pitot problem, and there probably was a stall. But in either a high speed stall or low speed stall, the resulting action probably should have been the same, and there was plenty of time to correct for either. Now, I’m not an expert in high altitude flying or accidents, but what else could have happened? Perhaps they lost all instruments and in black out conditions they were completely disoriented (although the recorder didn’t show that). Again, how could they not at least attempt pushing down? Something else must have happened.
Additional theories I have are that the engines, at 100% throttle and then 50% throttle (as reported), could have somehow maintained the orientation of the aircraft despite the aerodynamics during the fall. If so, that sounds like a design flaw as well, but that wasn’t a probable conclusion from what I read. Or perhaps a 10,000 feet per minute free fall, which was also reported, makes recovery impossible? But again wouldn’t the top of the stall arc prior to the 10,000 feet per minute drop have been slow, building to that point, and allowing for recovery first?
I’m Afraid Dave
I have one final theory. The pilots had ‘help’.
Perhaps the autopilot, under severe and uncommon (for an Airbus) failure conditions, might not have fully disengaged. I think it’s quite plausible for the autopilot to have first mishandled the error by performing the wrong correction, and then also not disengaged correctly, preventing the pilots from overriding it. Current fly-by-wire systems have no physical linkages, so it’s not a matter of physical strength to overcome an autopilot; perhaps the controls were simply useless. One interesting speculative paragraph from Mr. Iving’s pre-data recorder post of May 27th, 2011 may support this theory. He said the following:
“It’s a fundamental part of the Airbus philosophy about automation that pilots should not be able to assert individual control outside of a defined envelope that guarantees the stability of the airplane. As long as the autopilot and engine auto-thrust are engaged, the computers inhibit pilots from superseding the commands of the system. Boeing, in contrast, within its own version of a fully automated flight deck, has a system that allows pilots to intervene at any moment when they feel it necessary.”
Where this information comes from isn’t known, but if true, then it gives the autopilot theory some merit. Personally, I think it’s also quite possible that the autopilot may have played a role in maintaining the odd, unnatural, nose high free-fall of the aircraft in some way. I can’t really imagine how that’s possible, but was it ruled out? Since I wasn’t able to read the French report or the original report in English, perhaps I missed a key point. But neither Mr. Irving, or CNN went into detail on this aspect, so it either wasn’t in the report, or wasn’t fully investigated. I would love to know the answers to some of these questions, as I find the current explanation too simplistic.
- Although Flight #447 was flying through severe weather, I do not believe it played a significant role in the pilots’ complete loss of control of the airplane leading to the plunge. The weather may have triggered the pitot tube problems, and maybe even the stall, but certainly could not have been the cause of the odd circumstances of the crash, from what I’ve read.
(Update April 30, 2012) A great, detailed article was published in the Telegraph here discussing the potential problems with the Airbus design and philosophy on auto pilots.