Returning to minkyhaha’s original question, the relevant part of the article is this:
We’re not told exactly what information the families of the victims were given, or how it was worded — only the inference they’ve drawn from it. It’s entirely understandable that they might prefer to believe that their relatives’ last moments weren’t filled with unbearable terror, and I honestly wouldn’t blame whoever was briefing them for letting them think that.
Quoth Whack-a-Mole:
But is that 40 degrees relative to terrestrial vertical, or relative to the airplane’s locally apparent vertical? I mean, that 707 Tex Johnson was flying made a 360 degree roll (relative to terrestrial), and while I won’t say that nobody on the plane could have noticed, it probably wouldn’t have been panic-inducing.
Actually, if you were sitting in the 707 when Tex Johnson rolled it, and you couldn’t look out the window, you might not notice the maneuver, particularly if there was something else occupying your attention. That, however, was a carefully controlled and deliberate maneuver.
What happened on Flight 447, however, was not deliberate, not controlled, and not gentle.
In this context, 40 degrees was almost certainly “40 degrees relative to the ground”.
That Guardian article I quoted says the pilots were speaking in French. One of the last things they shouted was, “je ne comprends rien”, I don’t understand a thing.
The A330 pilot says:
The professor and member of the Royal Aeronautical Society said:
Now I don’t know, maybe they’re exaggerating or fabricating how calm it would have been. The article itself does describe the plane as “rocking”. My completely uneducated guess would be that it’s somewhere in the middle - that the passengers would have certainly noticed something was wrong and would have been terrified, but that it would have felt like severe turbulence rather than something out of Lost. Which, I think, is what you seem to be suggesting as well.
Trying to catch up the various ideas since my last post …
Attitude indicator. All airliners have 3. 2 primary & a backup. In modern electronic airplanes like the A330 they’re pictures on LCD screens rather than mechanical gizmos. The current term of art for the LCD-screen style is “PFD” meaning Primary Flight Display. In addition to the raw “Which way is up?” horizon display PFDs will typically also have airspeed, altitude, rate of climb/descent, navigation error values, & a couple other parameters displyed all in one nice easy to interpret schematic.
So assuming no failures of those instruments, the pilots knew whether they were pointed nose up, down, or rolled one wing down or the other. The BEA interim report includes a bunch of values at various times for these parameters. The fact the data recorder was recording reasonable values gives support to the idea that the pilots were probably being shown valid data. If there had been a loss of attitude data displayed to the pilots we’d expect a) they’d say something about it, and b) the BEA interim report would have mentioned them saying something about it.
Language: English is the “official” language of aviation worldwide. But … In most non-English speaking countries most local aviation radio comunication is conducted in the local language.
When local ATC deals with any foriegners, English is the common lingua franca. E.g. When Air France is departing Brazil English is used. So both the pilots & ATC need to speak “Aviation English.” There is a lot of formally internationally standardized terminology to ensure folks understand each other correctly.
This system is far from perfect. It can often be hard to understand a pilot or ATC person whose native language isn’t English. Native English speakers can forget to use textbook perfect phraseology, speak slowly, & enunciate carefully when dealing with non-native speakers. A lot of valuable understanding of the air traffic situation around you can be gleaned from hearing & understanding all the radio chatter directed at planes other than your own. Much of that is lost if multiple languages are spoken. Finally, there have been systemic failures where either the pilots or controllers only knew “Phrasebook English” and had no way to effectively communicate a non-standard message.
I don’t have personal first hand knowledge of this, but everything I’ve ever read indicates most cockpit conversation, including formal things like checklists, is done in the pilot’s local language.
It gets very interesting if the *crew *is not all common language speakers. From the 60s on, much of East Asia big jet flying was done by expat Americans, Brits, Aussies, etc. mixed in with local Japanese, Chinese, or Arab crews. I know a few guys who did that. English was the official cockit language, but … Between the culture gap & the language gap, you can imagine the potential for challenges once things become non-routine.
Having said that, I often fly to the major airports of Mexico. As such I’ve learned to speak carefully, listen carefully, and to understand at least the gist of most of what I hear in Spanish directed at somebody else. I’d fail utterly at trying to speak aviation Spanish though. I bet most of my *compadres *are similarly situated.
As to a “perfectly good” airplane “falling out of the sky”. Not really. Or at least that interpretation hides more than it reveals. I’ve already talked about the fact that a 100% mechanically healthy airplane can be flown *into * several distinct situations it can’t be flown back out of. The one and only cure for these risks is to avoid flying into those situations. Focussing solely on the machine’s mechanical state while ignoring its dynamic state is silly. It takes both to fly.
These guys flew into one of those traps. We know the initiating trigger was loss of all airspeed indications. From the BEA interim report we have some parameters for what happened next. Which gives a lot of partly-informed folks with soap boxes (including me) a basis to say a few things about the big picture of *what *happened. But not the details of *what *nor much of anything about why.
What (IMO) we can say now is that once all the airspeed (& apparently AOA) data became defective, then some as-yet unknown combination of bad luck, bad instant decision making, bad hands-on skill, weather induced turbulence, administrative error on the ground, airplane computer “helping” by actively making things worse, other mechanical issues, etc. caused them to drive, in the space of 15-30 seconds, past the point of no return.
Cabin experience. I have never personally stalled or gotten close to stalling a big jet. But I did talk, not 3 months ago, to one of our pilots who had been through a big jet near-stall while riding as a passenger. This was not on an Airbus, but the general picture will probably apply to an A330.
For whatever combination of reasons, the pilots got slow enough at altitude that the aircraft *almost *stalled. They were in smooth air, not in clouds and it was daytime.
From everything appearing completely normal, there was suddenly agressive short sharp banging turbulence. And mild wing-rock, 10-15 degrees each way. Wing rock doesn’t feel like just a mild S-turn. More like a juddering slap off to the left. Then to the right. etc. For the first few seconds there was gasping from the crowd. About the third time a wing dropped, a few people screamed. At each successive rock the screaming got louder & involved more people.
The aircraft was in a recoverable situation when the problem started and the pilots did the recovery correctly. After about 30 seconds of the above experience and (WAG) 8000 feet of altitude loss (including ears popping a bit), the ride was over.
There was no doubt in my co-worker’s mind that a sizeable fraction of the pax were in fear for their lives. This was a domestic US flight. There was lots of screaming of Jesus! & Oh God! etc.
He said his own reaction was a mixture of “I know what’s going on”, “I know how badly this can end”, “I know these guys ought to be able to recover” and “But if they goofed up enough to get into this situation, how much can I beleive they’ll get back out?” He was real concerned at first (ironically before most of the pax had gotten excited) but as he felt the right pilot reactions happening he became optimistic it would come out OK, even as the rocking got worse.
They changed an awful lot of seat cushions in that airplane. Many people had soiled themselves.
Was the A330 accident exactly like this? Beats me. But I bet it was pretty similar; just 7 or 8 times longer and at night in the weather. Whether the night & weather made it scarier or better since there was nothing to see outside to measure the motion against is an open question.
But I’d bet a paycheck nobody was sound asleep or reading their magazine by the time the ocean arrived. And I only get paid once a month.
Thanks LSLGuy. Clears a lot up.
I am curious about this “coffin corner” thing mentioned in one of the linked articles above. It suggested that planes at cruising altitude are flying in a very small window of “everything is ok”. Apparently the loss of a few knots of speed is enough to muck things up.
They also said (and I believe them) that planes fly in that window as a matter of course and it is quite usual.
To me though it sounds like a recipe for disaster. Since disasters do not occur often with planes I wonder if they are overstating how narrow the parameters are for flight in this case or if the parameters are gone out of it is trivial to recover or what the deal is.
Just seems if you are flying on the bleeding edge of your flight envelope sooner or later someone will tip-toe across it and with millions of flights a year it’d happen with some regularity. Since planes are not dropping all over what is the deal with this?
That is a good thing but don’t they make you practice stalls in a simulator?
My second father was a private pilot (single engine) and my mom took lessons and they said it was part of training to actually stall their single engine plane in flight and recover.
I realize that a commercial jet is an entirely different beast than light aircraft and I completely understand them not wanting to actually have pilots stall a real commercial jet to train them. That is what simulators are for.
In short, I though stalling and stall recovery are a major focus of pilot training at all levels.
ETA: This is not to disregard your point that you can fly a plane into an unrecoverable situation. Just asking as a general point.
Coffin corner in general - If all is working as planned, there’s plenty of leeway for normal operations. Kinda like the speed at which you normally drive around a cloverleaf on/offramp leaves enough room for typical surprises like a wet spot or loose gravel. But encounter a puddle with gravel in it that happens to have a big oil spot in it from a prior accident and you’ve got tires due for replacement and suddenly you’re bashing the guard rail.
If the airplane is heavier than you think, due to a clerical error in loading (or just assuming all 300 of your American passengers each only weigh 200 lbs including their carry-ons), you may be closer to the max safe altitude than you think. In other words, the corner is tighter.
If it is warmer than expected at altitude, that also places you deeper into a crack. In domestic operations that isn’t a big deal; you just ask for & immediately receive a lower altitude, then climb up later when you are lighter or the temperature declines to the expected value. You’ll burn a bit more fuel, but almost certainly not enough to matter for getting to the destination. Conversely, over the ocean ATC is very inflexible and it can be very hard to get another altitude, and once you do get a lower one, you may be stuck there for the rest of the crossing. Which may lead to enough increased fuel burn over several hours to mean you’ll be cutting it close at the destination. That conundrum leads to climbing as high as you can as soon as the charts say you can do so safely. Which means sometimes operating right at the book limits. There are still safety margins at the book limits, but by definition you’ve got less than you might typically have.
Finally turbulence can make a huge difference in maximum safe altitude for a given weight & temperature. The books typically have 2 limits, smooth to light turb & moderate or greater turb. If somebody planned for light & encountered moderate they could suddenly have been 5000 or more feet too high.
Also, there isn’t any “turbulence meter” to know the actual engineering definition of when “light” becomes “moderate”. We use our finely calibrated butts. Nor is there any way to really know the conditions ahead of the airplane. There are rules of thumb, forecasts of varying quality & freshness, and sometimes reports from other airplanes’ calibrated butts ahead. But particularly over the ocean this is mostly still a matter of being aggressive to avoid gross errors, but otherwise go give it a try & see what we encounter and retreat if necessary.
Coffin corner & Air France - We know from the BEA interim report that the temp at altitude was warmer than forecast to the extent that the crew was concerned that they were operating near the published limits. We also know it was getting more turbulent.
In and of themselves, each of these things is innocuous. Now add that maybe (I have no evidence) the airplane was a couple tons heavier than the paperwork says. Or the engines on this one were just at the lower limit of thrust output. Or …
Add all that together & what looked like a 25% safety factor is really a 1% safety factor. Which meant the’d be fine as long as the pilots & computers payed attention & didn’t get even 10% slow or do any unwarranted inadvertent climbing.
But when the airspeed quit working, in the initial confusion that’s exactly what happened.
As you said in an earlier post, accidents happen when a bunch of random factors all line up just wrong.
Stalls: Yep, we practice them in the sim. And pretty much all pilots learned to fly in small airplanes and did stalls and in some cases spins for real.
Stall training in real airliners went out in the 60s after sims got good enough to do the job and after a couple of 707s were lost during training sorties with several complete crews on board. Turns out engines tended to break off the wing when everything was flapping around out there during a real stall.
Historically (say '70s to 2000) there was a lot of training enphasis on stall training & recovery in the low altitude scenario. Obviously there isn’t much room to recover if you stall while manuevering for approach. It’s a busy time of flight & the canonical scenario is distraction, plus some other errorelse leads to not getting enough flaps out early enough & getting slow. Or losing an engine after takeoff & trying to climb at the normal steepness which won’t work with 1/2 power.
At any rate, the industry has come to the conclusion that high altitude stall recovery A) needs a very different recovery technique than low and B) needs to be trained explicitly & spearately from low altitude recovery. This accident wasn’t the initiator for this official attitude shift, but it was suspected of being one more brick in the wall when it happened, and the interim reports tend to confirm that suspicion.
In one type I flew, it was discovered that for some conditions of weight & loading at altitude, the first indication of stall would be the actual aerodynamic event. All the various warning gizmos wouldn’t go off until *after *the stall had occurred. Down low and under other conditions, the warnings would work as advertised, alerting the crew well before things progressed to an actual stall. That discovery led to some interesting revisions in books & procedures, butnot yet a change in the machine itself.
I’ve enjoyed our little back-and-forth. I’ve gotta go change into my work clothes & go fly jets for a few days. I’ll be back to the Dope on probably Tue morning or maybe it’ll have to wait to Wed mid-day. So don’t interpret me suddenly going dark as anything more than me being unavailable. See y’all.
thanks very much for your informative posts - clear skies!
Yes, I’m disputing expert opinion here. I’m aware that you don’t know me from Adam and why should you take me at my word, right? I’ll label it my opinion if that makes you happy. I think there is motivation for those folks to sugar-coat things, or minimize the distress.
Yep, I do.
Now, to clarify - a stall can be a non-event. I know this from direct experience. However, it’s not always that. Characterizations of stalls as either “non-events” or “screaming horror” are both inaccurate. Stalls can be either, but most fall between.
The BEA report, based upon the flight data recorder, reports bank angles of up to 40 degrees. Holy crap, that would freak me out if I was on a commercial flight! The passengers had to notice that, there’s no way not to short of being in a coma.
LSLGuy’s responses are one of the main reasons I love this message board.
Thanks for the detailed responses LSLGuy. Very informative. Helped me understand this issue much better.
I did one in a single engine Cessna.
My second dad was a private pilot and was flying with me in a single engine plane. I was about eight years old at the time. Without warning the guy stalled the plane.
I lost my mind in terror.
He managed to catch my attention and, holding a pen between us, let go of the pen and it just floated there.
As an adult that wouldn’t work but to my 8-year-old mind that was so cool I forgot I was about to die.
Go figure.
Kids are strange.
Guy was a dick for doing it that way (and an asshole in general but not important to this story).
That said it taught me stalls can be no big deal.
I also realize (and I want to be clear on this) that stalling a single engine Cessna is a far cry from stalling a 737. No doubt there are similarities but in the end they are different creatures and not very comparable.
There might be a brief moment of it, but zero-g isn’t really characteristic of a stall.
Don’t take this anything but a general note of interest. It wasn’t that scientific. I just simulated a full stall in an Airbus 321 with all air speed indicators failed in Flight Simulator X from 38,000 over the ocean from Rio De Janeiro. The real plane was an Airbus 330 which is fairly similar but not exactly the same model. I may try it again later if I can get a sim of an A330 for exact match.
That was a rough ride down that followed a similar profile to the actual crash and it was much harder to overcome than I thought. Those airliners are fly by wire which means the pilots don’t have direct control of the control surfaces. Pilot movements go through the computer to translate desired results into coordinated control surface movements. In a full stall, the airliner wouldn’t give me enough control authority to get the nose down to build up airspeed and get out of the stall. It just went into a semi-flat spin. I was able to recovery after about two minutes down to 11,000 feet but it was difficult even sitting at a computer willing to try anything and not in a spinning cockpit. The autothrusters kept kicking on as well when I didn’t want them to. In short, the computer was trying to take over and help continue the crash. I can see how hard that would be for pilots to get out of especially if they are in a cockpit that is moving around that violently. It also illustrates that it was probably far from peaceful for the passengers as well.
The real flight data recorders can be fed into a sim to give an accurate simulation to show what it probably looked like from the inside but I doubt anyone is going to want to do that.
I am getting the impression that this plane, and maybe a lot of modern planes, is operating near the limits of its performance under normal conditions. And there is not much spare capacity when the aircraft is pushed outside of normal operating conditions.
But the only thing I can think of the might have averted the crash a whole lot more engine power, which might have pushed enough airflow over the controls to escape from the stall. That is, if they ever thought to stop trying to pull the nose up. I’m also guessing that if they applied that much power, it might be enough to tear the airframe apart.
I like wild speculation and uneducated guesses. I would probably be a lousy crash investigator.
It seems to me they’d be very likely to do this on order to try out possible solutions and revise training if indicated.
I do remember seeing a claim that this had been done with the Sioux City crash – the one where the engine fragment had severed all the hydraulics and ended up in a crash landing that killed quite a lot of those on board. The story was that every pilot who tried this ended up “dead” – no one came near to bringing down the plane into anything approaching a controlled landing.
Full-flight (Level D) simulators can faithfully replicate 40 degree rolls both by visual input and tilting the jacks to replicate the accelerations. Whether a training program requires it for their pilots or not, a simulator (even before AF447) can simulate the conditions leading into a high altitude deep stall, but the flight physics once in the stall are largely theoretical when in an uncontrolled stall since it is, obviously, incredibly dangerous to actually stall each and every aircraft type in order to gather the data necessary! A lot of controlled stall data is derived from the flight testing done by the aircraft manufacturer during certification of the airframe. These types of simulators are aircraft-specific, btw, and take a couple years to design and build for a new plane type (and a few months for simply a new sim of an existing plane), and cost several million dollars. You can be pretty sure the Air France pilots were receiving their training on these types of simulators (in fact, I know that they do!)
We can be pretty certain that the BEA will use this type of simulator to replicate this crash, and evaluate potential/probable pilot responses to see if the crash was survivable or not. The NTSB did this for the crash into the Hudson, for example.
Airlines are already incorporating more high altitude stall training into their simulator sessions, and this accident investigation will provide more details on how to train other pilots for these types of situations.
Here’s one thing which I don’t understand at all: given that they pilots seemed to clearly recognize that they were in a stall situation, why did they ever bring the thrust levers back to idle? Isn’t it pretty clear that you want to at least attempt to power out of this?
“At 2 h 12 min 02, the PF said “I don’t have any more indications”, and the PNF said “we have
no valid indications”. At that moment, the thrust levers were in the IDLE detent and the
engines’N1’s were at 55%”
Supposing the pilot had a button that turned all controls to direct input, ie. no computer modulation, would the plane become totaly unflyable?