Air France 447 - On topic only!

[Evan]

I'm not Shakespeare either, but methinks we have forthwith turn'd that black word "coffin corner" to banishment.

[Gabriel]

And since we are getting philosophical here, how fast is 10 knots really?

In Taliesin's photo we see that the speed is 280 knots and the barber-pole bar that marks the overspeed zone starts at some 295 knots, so the margin with overspeed is 15 knots.

But wait a second. 280 knots is 280 nautical miles per hour.
So what is moving 280 nautical miles each hour about what?
Nothing. The indicated airspeed of 280 knots is not a real speed, because it's not the rate of motion of something about something else. The airspeed shown in the display is the answer to this question:

"If we were at sea level at standard pressure and temperature (which we are not), what speed should be flying to produce the same lift that we are producing now?"

But as said, at 35000ft you are not at standard sea level conditions. Not even remotely close.

In fact, the air density up there at cruise is about 1/4 of what's down here at standard sea level conditions. And then, the mass of air that flows around the wing up there is about 1/4 of what would be flowing at the same REAL speed at sea level.

Other things left equal, 1/4 of mass of air flowing means 1/4 of the lift. However, the lift MUST be the same up there and down here because in both cases it must bear the airplane weight, and that's something that doesn't change with altitude (almost).

Other than the air density, another thing that affects the lift is the REAL airspeed. And it affects it more strongly that the air density because the lift goes with the square of the speed.

So, if we are flying in a zone where the air density is 1/4 of the standard one, how much faster should we fly to produce the same amount of lift than at sea level? The answer is twice as fast (you know, 2 squared is 4).

So, because the air at sea level is 4 times as dense as that up here, at sea level we should be flying hat half our REAL speed to produce the same lift. And that's the answer to the speed indicator's question:

The speed shown in the airspeed indicator at 35000 ft is about 1/2 of the real airspeed (called True Air Speed, or TAS).

So, when the indicator shows that you are flying at 280 knots, your TAS is about 560 kts (you are moving about the air 560 nautical miles each hour). And when the barber-pole starts at 295 knots in the indicator, it starts at a real 590 knots of TAS, so the speed margin is 30 knots of TAS.

Now, should we measure the margin between our current airspeed and the start of the overspeed in terms of indicated airspeed or TAS?

The answer, in my opinion, is in TAS (and that's why I explained the graph using TAS). Why? It takes the same "effort" (net force, acceleration) to make the same change in TAS, NOT in indicated airspeed.

For example, during the take-off roll, the speed increases about 5 knots per second (that's the acceleration). If the take-off happens at sea level, indicated airspeed and TAS are the same. So it would take it 3 seconds for the speed to increase 15 knots and 6 seconds for it to increase 30 knots.

Flying at 35000ft, where the TAS is twice the indicated airspeed, at the same acceleration it would take the 6 seconds corresponding to 30 knots to increase the indicated airspeed from 280 kts to 295 kts, not the 3 seconds corresponding to 15 knots as it would seem at first.

And that, assuming that you could produce an acceleration of 5 knots per second in a level flight, which you can't because it would require to put as many pounds of thrust as there were for take off, something that is not available up there and not even close, and to reduce the drag to that of the take-off roll, something impossible too.

I bet that, flying some 2000ft below the absolute ceiling as AF447 was doing, it would take an airplane quite a bunch of seconds (maybe a minute or so) to accelerate those 15 knots of indicated airspeed (30 knots TAS) after firewalling the throttles, pint at which the overspeed warning would sound and the pilot has quite a margin to recover well before the situation becomes anything close to dangerous.

While the stall speed approaches the cruise speed as the altitude gets higher, it's quite typical that that cruise speed will be just 0.02 to 0.05 of mach (i.e. of the speed of sound) below the overspeed limit regardless of the altitude. So the cruise speed isn't even necessarily in the middle of the window. Typically it will be much more towards the "overspeed" limit.

On the other hand, the margin above the stall will never be lower than that at final approach (typically it will be quite higher), and I have not listen you complain because of the aviation industry putting us systematically at the edge of death just before each landing.

We have like 50 years flying jets like that (since the 707 and DC-8, or maybe even earlier) and I don't think that it has caused lots of problems, if any.

Note: This post was regarding the coffin corner issue at great, and specifically NOT related to AF 447, where the coffin corner had nothing to do at all.

[MCM]

Taliesin,

A few big assumptions you've made there!

That said, yes, AF447 was not operating in coffin corner.

Gabriel,

That all makes sense about energy. What happens when you go through a windshift, and headwind decreases by say 30kts in 10 seconds (yes, that does happen). Do you have the thrust available to re-accelerate?

How about when the Outside Air Temperature increases by 20 degrees in a second or two?

[Gabriel]

Gabriel,

That all makes sense about energy. What happens when you go through a windshift, and headwind decreases by say 30kts in 10 seconds (yes, that does happen). Do you have the thrust available to re-accelerate?

How about when the Outside Air Temperature increases by 20 degrees in a second or two?

"PAN PAN, Jetphotos 1234 descending from 350 to 330, unable to hold altitude", honoring any stall warning (if it happens) with the needed reduction in AoA and coupled with a turn a bit off-course and an increased scan of the TCAS data, or something like that.

In the unlikely event (have you ever seen it?) that a stall warning happens at the same time than a an overspeed warning (meaning that the situation put you beyond the coffin corner), then honor the stall first. There is more margin on the overspeed side.

All of the above is from raw common sense. Not that I do know what one is "officially" expected to do in such situation. So your (or other's) confirmation/correction would be much appreciated.

[Evan]

"PAN PAN, Jetphotos 1234 descending from 350 to 330, unable to hold altitude", honoring any stall warning (if it happens) with the needed reduction in AoA and coupled with a turn a bit off-course and an increased scan of the TCAS data, or something like that.

In the unlikely event (have you ever seen it?) that a stall warning happens at the same time than a an overspeed warning (meaning that the situation put you beyond the coffin corner), then honor the stall first. There is more margin on the overspeed side.

All of the above is from raw common sense. Not that I do know what one is "officially" expected to do in such situation. So your (or other's) confirmation/correction would be much appreciated.

MCM, I'm curious about this. I know you're a Boeing pilot, but on the A330 in normal law while flying on autopilot, what happens in the scenario you presented? Does the autopilot disconnect when alpha-protect becomes active? If the autopilot does not disconnect (because the AoA remains below alpha-protect), I imagine the A/T would apply maximum thrust, and the autopilot would "give up" when it can no longer hold the designated flight level without entering the stall protection regime. On a Boeing? Does the autopilot disengage at the stall warning? Or does it disengage when it cannot hold altitude at a safe AoA? Or would the autopilot continue to attempt to hold flight level with excessive AoA?

My understanding is that autopilot will never command an AoA at or above stall warning. If so, I think the real danger in hitting the stall side of coffin corner lies not as much in the approach to stall as in the sudden transition from managed autoflight to reactive manual flight. While AF447 was not in coffin corner, it does show us how dangerous the transition can be.

So perhaps the immediate risks at coffin corner are really overspeed on the high end and loss of autoflight on the low end. Followed by the risk of pilot error.

One thing I am hoping might come from this is a sort of blending approach between automation and manual flight in these unexpected situations, if that is possible.

[Fear_of_Flying]

"PAN PAN, Jetphotos 1234

*Off-topic* So I was curious about the origin of "PAN PAN PAN" and looked it up. It comes from a French word (not for "bread", which is kind of what I thought it sounds like), but for mechanical failure. The article went on to explain that various bacronyms have been applied to PAN, such as "Possible Assistance Needed". Well, I had never heard of a bacronym before, either, so this was a learning experience (a bacronym is when you come up with suitable words for the word that already exists). Then came the most interesting part. The article listed uses of PAN PAN PAN in modern aviation, and one of them was from May 5th of this year. I hadn't heard about this incident, and I'm not sure if there's a thread here or not - I don't see it if there is. But here's the summary, kind of funny in a scary sort of way:

An Air Berlin Airbus A330-200, registration D-ALPA performing flight AB-9721 from Palma Mallorca,SP (Spain) to Munich (Germany), was on approach to Munich about 15 minutes prior to landing, when the crew declared PAN PAN reporting they were fatigued and needed to perform an automatic landing. The airport needed to temporarily invoke procedures to protect the ILS category III zones to permit an automatic landing. The aircraft continued for a safe landing on runway 26L about 15 minutes later at 10:27L (08:27Z).

(From Aviation Herald)

Too tired to land the plane? That's a new one.

[MCM]

Pan Pan is not a particularly rare event FoF. The would have been plenty used since the 5th of May.

They just achieve radio priority, and ultimately (if required) priority handling.

The incident you mention is an interesting one. I am lead to believe that the order of events was slightly different to that reported. I may well be wrong, but I believe the pilots reported that they were fatigued and going to do an autoland. The airport would have said fine, but the ILS critical areas are not protected. The crew may have decided they wanted them protected, and by declaring a PAN ATC are required to comply and protect them.

Its an interesting policy they have about requiring autolands, and I'm sure it will be closely examined by airlines everywhere.

Gabriel - declaring a PAN and descending is obviouly what you would do. That wasn't exactly my point. My point was that this thread seems to be very much discounting the "coffin corner" concept as one that seriously affects civilian jetliners, when that is not the case. There seemed to be concentration on normal aircraft operations, but not considering the effect of increased G force (be it by turbulence or even just enroute track changes), or by environmental changes. You are correct that accelerating to that speed normally would take time, but it is environmental factors that are more likely to actually cause the problem.

Evan,

You would have more idea about what an airbus does when faced with that scenario than I do! I know very little about airbii. For what its worth, I'm not sure I agree with the the "blended" approach. I think too often confusion comes from not understanding what the aircraft is doing. Too many different, subtle modes. I personally prefer the Boeing method - either the autopilot is doing it, or it isn't doing it. At least I know for sure what my control inputs are doing.

[Evan]

Evan,

You would have more idea about what an airbus does when faced with that scenario than I do! I know very little about airbii. For what its worth, I'm not sure I agree with the the "blended" approach. I think too often confusion comes from not understanding what the aircraft is doing. Too many different, subtle modes. I personally prefer the Boeing method - either the autopilot is doing it, or it isn't doing it. At least I know for sure what my control inputs are doing.

Maybe 'blended' isn't the word, but something like a five-second warning that the autopilot is about to disengage. I realize a lot can happen in five seconds, but the pilot could always take it immediately by pressing the instinctive button. I think the sudden handover can be disorienting when a pilot is least expecting it.

My question put more simply is what will the autopilot do at cruise altitude if environmental factors suddenly put your airspeed below stall?

[Gabriel]

My question put more simply is what will the autopilot do at cruise altitude if environmental factors suddenly put your airspeed below stall?

It's not exactly what you are asking, but AFAIK, in many airplanes (all that lack envelope protection?), in situations of diminishing airspeed, the autopilot set in alt or vert speed mode will have no problem to increase the AoA all the way into the stall (or at least stall warning).
We have Turkish and Pinnacle as examples.
What I don't know is if, when the stall warning is reached, the AP will self disconnect or not.

[Gabriel]

Gabriel - declaring a PAN and descending is obviouly what you would do. That wasn't exactly my point. My point was that this thread seems to be very much discounting the "coffin corner" concept as one that seriously affects civilian jetliners, when that is not the case. There seemed to be concentration on normal aircraft operations, but not considering the effect of increased G force (be it by turbulence or even just enroute track changes), or by environmental changes. You are correct that accelerating to that speed normally would take time, but it is environmental factors that are more likely to actually cause the problem.

Then we agree 100%.
The "enroute track change" effect can be highky mittigated, nearly eliminated, by limitting the bank angle to a max of 15 degrees. I would definitivelly not make a 25 or 30 deg bank turn if I'm just 10 or 15 knots IAS above the slow speed buffet or stall warning. Agreed?

[Taliesin]

Taliesin,

A few big assumptions you've made there!

That said, yes, AF447 was not operating in coffin corner.

That's true, but I believe that my assumptions are reasonable. An aircraft with the same wing flying at the same density altitude is very likely going to have a similar weight, barring adverse winds, it would not fly at that altitude if the weight were not similar.
Either way, even if the weights are not similar, the density altitude and Mach numbers are, which were the details in question.

In the end, we agree

[Evan]

BBC reporting a bit ahead of the release of the final report:

One of the mistakes of the crew, according to investigators, was to point the nose of the aircraft upwards, after it stalled, instead of down.

The father of one of the flight's victims says that investigators have told him, ahead of the final report's release on Thursday, the reason for this incorrect move was that the flight director system gave "erroneous information" that the jet was diving.

Which is why the QRH procedure specifically instructs pilots to turn off the flight directors. THis is why you MUST adhere to procedure.

[TeeVee]

BBC reporting a bit ahead of the release of the final report:



Which is why the QRH procedure specifically instructs pilots to turn off the flight directors. THis is why you MUST adhere to procedure.

interesting: when the machine clearly makes a mistake, it is still the human's fault because he failed to shut the machine off. when does it ever become the machine's or its' engineers' fault?

[Evan]

interesting: when the machine clearly makes a mistake, it is still the human's fault because he failed to shut the machine off. when does it ever become the machine's or its' engineers' fault?

The machine isn't "making a mistake". The machine is deprived of data (airspeed) necessary to function. Therefore the manufacturer specifically directs the pilots to deactivate components that might be unreliable in that state, and take the plane into their own hands.

This is what the pilot is there for.

If he is properly trained...

[Peter Kesternich]

interesting: when the machine clearly makes a mistake, it is still the human's fault because he failed to shut the machine off. when does it ever become the machine's or its' engineers' fault?

Going down that road and giving the last responsibility to the machine will lead us to single pilot flight crews and eventually fully automated planes. And that is the day I will stop travelling on aircraft.