AF 447 Thread No. 7

[PJ2]

Hamburt Spinkleman;

There are reasonable things upon which informed, experienced and knowledgeable speculation or assumptions may be made. This accident is no different and many very good people who have engineering, mathematical, flight safety and airline operational and piloting experience have contributed freely and added to understanding. This includes a number of pilots who have flown this aircraft, some under similar weather conditions and who know this environment.

Along the way through these eight, possibly nine threads some exceptional thinking has been offered which has helped others who are less familiar with these aircraft and these operations but who nevertheless suspend judgement and comment, in favour of an abiding curiosity and thirst for knowing how others do this work.

In keeping with discussing what may be reasonably known regarding the displays available to the crew, post-event and so as to avoid leaving false impressions for others who are also reading here and studying this accident, that there was no flight information available to the crew other than the deck angle and the seat of their pants, the displays and all information provided by the two PFDs and NDs were almost certainly available with the exception of the airspeed and that only for a period of under thirty seconds.

The IRUs, DMC's and at least two FCPC's and the FCSC's were functioning, the aircraft was fully-powered, all the data needed for controlled flight was available and no primary system losses such hydraulics, engines or engine controls had failed.

A loss of airspeed data does not result in the loss of displays and there are no data indications, ECAM or ACARS messages indicating such a loss nor are there any indicated associated failures which would cause any or all screens to go dark.

Thus, attitude information (pitch & roll), IVSI, altitude and heading information were all available. On the Systems Pages all information would be available including flight control positions. The ISIS, a somewhat independent attitude, speed & altitude (but not VSI) system was also available except for speed for a short while.

In the thirty-six other similar loss-of-CAS events no screens were blanked nor was other information normally provided on the two PFDs and NDs lost to the crew.

[Lyman]

What did the pilots mean, then, when they reported to the Captain, that they had no displays? And when did the displays return? Did they?

[Old Carthusian]

I gave up reading this thread thinking everything had been discussed to death and that the primacy of the pilots in the accident had been fully understood. Unfortunately it refuses to die. All evidence and information points to a temporary malfunction which was compounded and turned into an accident by an inadequate response from the flight crew. This links into what I believe are relevant cultural issues at Air France particularly in the area of crew training. It also ties in with human reactions to unexpected situations and failures of CRM and lack of use of SOPs. But it is a human issue not a mechanical issue. As PJ2 has mentioned (nice to see you back by the way) this is not an issue with flaps or spoilers at all or with availability of information.

[PJ2]

Old Carthusian;

Re, "But it is a human issue not a mechanical issue. [As PJ2 has mentioned...] this is not an issue with flaps or spoilers at all or with availability of information."

Given what we have, (and it isn't unreasonable to doubt that what we have is all that is available), this is a performance accident. The far more critical aspect of this accident is "Why?" At present I think it is not complicated.

Lyman;

Re, "What did the pilots mean, then, when they reported to the Captain, that they had no displays?"

The only reference to displays is when the PF states, "We haven't got a good display...", "...of speed", and the PM states, "We've lost the speeds", (IR3 p29, p74, English version). No other references to loss of attitude, or altitude information is made. Contrarily, a number of references to altitude and pitch are made which indicate that the displays were functioning:

"Reading the three instruments (the two PFD’s and the ISIS), the PNF noticed that the airplane was climbing and asked the PF several times to descend." - IR3, p74 English Version

Regarding their contributions to the captain's understanding and the captain's responses:

"In the absence of relevant information from the copilots, reading the information available on the screens (pitch attitude, roll, thrust, vertical speed, altitude, etc…) was not sufficient in itself for the Captain to become rapidly aware of the airplane’s situation. He did not then ask questions that could have helped him to understand the sequence of events." - IR3, p76 English Version

Re, "And when did the displays return?"

The displays did not disappear. But the CAS indication did and it reappeared just under 30 seconds after the initial UAS event:

"At around 2 h 10 min 34, the speed displayed on the left side became valid again and was then 215 kt; the speed on the ISIS was still incorrect.". - IR3, p74 English Version

...and from IR3, p75:

"At 2 h 11 min 06, after several attempts to call, the PNF was anxious again about the absence of the Captain. This anxiety probably increased the stress for the PNF who was faced with a situation that he didn’t understand.

A second later, the speed on the ISIS became valid again. ADR 3 being selected on the right side PFD, the speed for the PF also became valid again. It was then 183 kt and the three displayed speeds were consistent. This brought no comment from the crew." - IR3, p75, English Version

I am well aware of the underlying assumptions behind these tentative conclusions but from experience with and knowledge of the airplane they are not unreasonable.

[CONF iture]

The displays did not disappear. But the CAS indication did and it reappeared just under 30 seconds after the initial UAS event
Later, after the aircraft was stalled, the CVR could indicate that other informations were temporarily unavailable, including vertical speed, attitude on the PFDs, altitude.

2:11:58 J’ai un problème c’est que j’ai plus de vario là - J’ai plus aucune indication
2:12:19 Les ailes à plat - l’horizon l’horizon de secours
2:12:30 Je suis en train de descendre là ?
2:12:42 En alti on a quoi là ?

I find too little has been said in 4 minutes time, especially from the Captain when he was back. French guys after all ...
Too little said or too little published ?

[canyonlight]

Thank you.

[PJ2]

CONF iture;

Re, "Later, after the aircraft was stalled, the CVR could indicate that other informations were temporarily unavailable, including vertical speed, attitude on the PFDs, altitude."

Yes, I can believe that. Under full stall conditions, a number of sensors, and therefore their downstream services, would not be working correctly. The NCD messages/indications and the "intermittent" stall warning are but two examples.

[AlphaZuluRomeo]

Hamburt Spinkleman & PJ2 : :D

[CONF iture]

Slats were retracted at time of impact. That is known from DFDR recorded data, ... That is a fact.
The only fact I can see is that none of the traces for slat position and flap lever position is published.

The NCD messages/indications and the "intermittent" stall warning are but two examples.
It is interesting to note that the vertical speed trace never reaches the NCD status, but still the PF seems to indicate that he totally lost his V/S reading.
As long as the V/S was selected from IR the V/S trace was making sense, when the V/S was selected from ADR the V/S trace was becoming somehow erratic.
We may wonder why the V/S was occasionally selected from ADR (A33Zab said something on that, I have to find it back) does it have anything to do with the IRs capability ... what could be the consequences on the PFD's attitude display ?

[PJ2]

CONF iture;

The best that can be said regarding a slat/flap extension theory is, it is possible that such a selection was made at some point but improbable that it actually was. I think this because that is a significant operational emergency decision and would be noted in the IR3 even if the parameter has not been provided. There is no mention of slats and only one technical mention of flaps in IR3.

As Turbine D has quite reasonably asked however, of what significance is this in relation to the accident?

Vertical speed is a baro-inertial calculation. Inertial and Air Data systems have different advantages in the calculation. IR calculation is better in dynamic maneuvers and ADR calculation is more stable over time, (no IR drift issues). Input is automatically selected in accordance with reliability. If ADR data is unreliable it returns an "NCD" status.

Perhaps this may explain comments regarding behaviour of the IVSI display, perhaps not. Various data-source selections were apparently made (can't recall what they were, and also FCPC was switched off/on as well IIRC) - as I mentioned, system behaviours in these circumstances are much less understood than in normal circumstances and, like the earlier examination and sorting out of the ACARS messages, a lot of permutations/combinations would have to be examined before a definitive sense of what was being displayed on the PFDs at any moment during the post-apogee phase. The crew does comment on (loss of) altitude and rate of descent during this phase so something was displayed.

[Organfreak]

Yesterday, I foolishly posted:
Perhaps due to my lack of advanced technical knowledge, but I have never been able to get my mind around the idea that they thought they were in OS.
Why not? Because of the highly unusual deck angle. How in the world could anyone suspect OS when tipped back like that???????????????????????????? I ask you.


A very kind and knowledgeable list-member has privately educated me on this (link my choice, not his). They didn't know WTH the airplane was doing unless the instruments were reliable on all counts.

Sensory illusions in aviation - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/Sensory_illusions_in_aviation)

Of course this is probably old news to experienced drivers. :oh:

[Lyman]

That is what instruments provide. Life itself. It is uncomfortable, but very occasionally one must utterly reject what one's senses say, and rely totally on gadgets.

And that is why Flaps and slats (spoilers) are important, and why a few of us are pisssed off at BEA for witholding data. As I said, the Captain had to climb the aisle to re-enter the cockpit. Note his first comment? "Er....what are you doing?"

A Frenchman does not confront another, generally, with such a terse comment.

There is a gob of very pertinent data simply missing from the releases, end of...

[Organfreak]

Lyman/Bearfoil/Bill:
...a few of us are pisssed off at BEA for witholding data.
and:
There is a gob of very pertinent data simply missing from the releases.....

I'm waiting to get pissed-off/very disappointed if the final report is missing any data. In the meantime, I don't believe it's fair to accuse them of that. 'Interim' means just that. I do agree in general that there appears to be holes in it, but guessing is, IMHO, pointless! Might you be what they call a loose cannon?

[jcjeant]

Hi,

CONF_iture
The only fact I can see is that none of the traces for slat position and flap lever position is published.We certainly will have more knowledge about this accident when (if ever) the FDR DATA will be published ... instead of those graphics from the BEA belonging more to cartoons than technical graphics

[PJ2]

Organfreak;

The excellent link you have provided discusses important aspects of pilots who are non-instrument-rated flying in, or into poor visual conditions. One must believe the instruments at all times because the alternative under some conditions is a loss of control.

You are right, most pilots are indeed familiar with these things...it's drilled into all pilots from very early on in ab-initio training/testing/licencing process.

Some may recall the John Kennedy Jr. accident in which control was lost at dusk, a time of day when the horizon is poorly defined.

Put bluntly (but not overly-dramatically), a non-instrument-trained or rated pilot who enters cloud or continues to fly in deteriorating visual conditions, once entered, has about a minute-and-a-half before the spiral dive (http://en.wikipedia.org/wiki/Aircraft_dynamic_modes) develops. The way out is to immediately level the wings but if one has no horizon and doesn't know how to read the artificial horizon instrument or attitude information is missing (displays blanked), the dive will not be stopped.

Not among the accidents listed in your link is the Gulfair A320 (http://www.iasa.com.au/folders/Safety_Issues/others/GF072Final.html)loss-of-control accident at Bahrain in 2000. Of particular interest is the "Appendix E, Perceptual Report (http://www.bea.aero/docspa/2000/a40-ek000823a/htm/pdfs/appendixe.pdf)". Also not listed among these accidents is the Afrikiyah A330 accident at Tripoli which some have discussed these same phenomenon. But we'll probably never know given both political and infrastructural problems which began with the recent civil war in Libya around the time of the accident.

In considering the notion that has been expressed regarding AF447 that all displays were blank, (no information displayed to the crew and not just a missing speed indication for a half-minute), almost certainly the result of loss of attitude, altitude, vertical speed, airspeed, heading etc (on a cloudy, moonless night) would have been a high-speed spiral dive and an extremely high speed impact with the sea. We know that such is not the case.

It isn't direct evidence of course but given the way these things go it is again a reasonable support for the view that the PFD/NDs were displaying most if not all their information except for the CAS.

[Lyman]

Captain PJ2.

"In considering the notion that has been expressed regarding AF447 that all displays were blank, (no information displayed to the crew and not just a missing speed indication for a half-minute), almost certainly the result of loss of attitude, altitude, vertical speed, airspeed, heading etc (on a cloudy, moonless night) would have been a high-speed spiral dive and an extremely high speed impact with the sea. We know that such is not the case. "

With great respect, Sir.

From BEA Interim Report Number 3

"At 2 hr 12 min 02 the Pilot Flying said: 'I have no more displays......'


The Pilot not Flying then said: 'We have no valid indications....' "

Now that's from page ten, and the statements are made 17 seconds after the Captain has re-entered the cockpit. Further dialogue? I would wager so...

[jcjeant]

Hi,
"At 2 hr 12 min 02 the Pilot Flying said: 'I have no more displays......'
The Pilot not Flying then said: 'We have no valid indications....' "
Dunno .. you must have a different report N°3 .. :hmm:
This is from the BEA report N°3 (downloaded from the BEA site)

http://i.imgur.com/PkDWB.jpg

[PJ2]

Hi Lyman;

...just plain old PJ2 is fine, ;-)

Re the wager, being a pilot I don't bet on things but thank you for reminding me of the information (on page 10!).

The loss of the VS parameter & displayed information is a possibility that I discussed in a previous post in re the "baro-inertial calculation and if supplied by a malfunctioning ADR, could return "NCD" and not be displayed. From the IR3, Appendix 1, p94:

@ 02:11:45
The vertical speed is no longer
calculated by the IR (Inertial
reference) but by the ADR. It is
about -10,000 ft/min.

This is coincidental with four ACARS messages indicating events occurring at 02:11:45, (and recorded/time-stamped between 5 & 29 seconds later) which also coincides with the disruption, as discussed above, of a number of parameters including the VS traces:

02:11:45 FLR/FR0906010210 34111506EFCS2 1,EFCS1,AFS,,,,,PROBE-PITOT 1X2 / 2X3 /1X3 (9DA),HARD
02:11:45 FLR/FR0906010210 27933406EFCS1 X2,EFCS2X,,,,,,FCPC2 (2CE2) /WRG:ADIRU1 BUS ADR1-2 TO FCPC2,HARD
02:11:45 FLR/FR0906010211 34220006ISIS 1,,,,,,,ISIS(22FN-10FC) SPEED OR MACH FUNCTION,HARD
02:11:45 FLR/FR0906010211 34123406IR2 1,EFCS1X,IR1,IR3,,,,ADIRU2 (1FP2),HARD
02:11:55 WRN/WN0906010211 341200106FLAG ON CAPT PFD FPV
02:11:58 WRN/WN0906010211 341201106FLAG ON F/O PFD FPV
02:12:02 PF states “I don’t have any more indications”; PNF, “We have no valid indications” Thrust levers to IDLE; N1s @ 55%
02:12:17 PF makes ND inputs; AoA decreases, speeds become valid again, stall warning sounds again
02:12:47 WRN/WN0906010212 341040006NAV ADR DISAGREE
02:13:32 PF: “we’re going to arrive at level one hundred”

When the PF says (likely to the captain), "I have a problem it's that I don't have vertical speed indication" and "I have no more displays" I do not think he means that his entire PFD is blank. From his comments, he still has attitude & altitude information and the altitude information is unwinding at a thousand feet every three seconds or so at that point in the stalled descent.

The overarching point, to which I am addressing my comments, that the displays were blank prior to the apogee and during the entry into the stall, and are somehow causally linked to the ensuing loss-of-control, stall and descent, is incorrect. Post full-stall and at high-pitch/high-descent rate, any number of indication & system anomalies are likely going to occur, which may lead a crew who may be continuing to believe that the aircraft (the wing) is still flying, to significant cognitive dissonance and an inability to construct a model of aircraft behaviour such that the way to a recovery could be seen. These are CRM as much as SOP matters, a subject which I expect to be dealt with in depth when the final report is issued.

[Lyman]

jcjeant

from Interim Report #3 F-GZCP. Downloaded and copied from my Apple ipad.

At 2 h 12 min 02, the PF said, “I have no more displays, and the PNF “we have no valid indications”. At that moment, the thrust levers were in the IDLE detent and the engines’ N1’s were at 55%. Around fifteen seconds later, the PF made pitch-down inputs. In the following moments, the angle of attack decreased, the speeds became valid again and the stall warning triggered again.

//////Interim Report n°3
On the accident on 1st June 2009 to the Airbus A330-203 registered F-GZCP
operated by Air France
flight AF 447 Rio de Janeiro - Paris
Bureau d’Enquêtes et d’Analyses pour la sécurité de l’aviation civile
Ministère de l’écologie, du développement durable, des transports et du logement//////

Old Carthusian. Thanks, you are right, and I would include the text on the page from which this comments arrive, also. It describes a a/c severely in UPSET, but from which a recovery is made. Also, the PF's side of the panel is not recorded, as you can see, and I do not doubt him, PJ2 has inferred proper displays as you have. That is fine, but let's not forget that we have actual evidence, in the two pilot's actual voices.


jcjeant what pages are you copying from #3? Are there different iterations of the report? Mine is of 114 pages long, downloaded some while ago. Do not tell me it has been altered/modified?

[Old Carthusian]

Lyman
I am not sure of the relevance of your comments here. The phrases you quote are taken out of context and should be read in conjunction with the rest of the transcript (further on in the report p94-96). From the comments made it is clear that the instrument displays were providing information.

[Organfreak]

I checked page 10 as Lyman cited. Here is the pertinent paragraph as copied and pasted by yrs truly:

At 2 h 12 min 02, the PF said, “I have no more displays, and the PNF “we have no valid
indications”. At that moment, the thrust levers were in the IDLE detent and the engines’ N1’s
were at 55%. Around fifteen seconds later, the PF made pitch-down inputs. In the following
moments, the angle of attack decreased, the speeds became valid again and the stall
warning triggered again.

It can be seen that on the chart posted by jcjeant (page 94), that time point of 2:12:02 is not even mentioned.

[Lyman]

OC I posted the comments stand alone, because PJ2 had inferred that they did not exist; context comes with chronology, certainly, but we were talking two different time frames. I don't see the problem.

PF and PNF made those comments as Captain re-entered, and I would point out the lack of further discussion as remarkable. The a/c was LOC and dropping madly. The report is to blame for the disjointed and disconnected conversations, not I. Why don't the comments I quote show up with the comments jcjeant posts? Where is the chronology, the complete record?

[gums]

I rejoin the fray after PJ2 has broken silence, and A333Z continues to provide detailed diagrams and systems info. Attaboys all around. And I think some of the inputs from we non-commercial drivers have helped. At least I hope so.

While wondering what the final report will conclude, my personal feeling is it will come down to training and stall recognition/recovery by HUMANS, not the machine.

- From the data traces I see no extraordinary efforts by the crew to recover from a stall such as using flaps, spoilers, etc. In other words, I do not think the crew knew that the plane was fully stalled, and the control inputs do not reflect appropriate stall recovery techniques/procedures.

- I flew the Deuce ( F-102) early in my career and it had no obvious stall entry warnings or "feelings". All that happened was the jet began to descend rapidly ( sink rate) and it had exceptional directional stability. Seems the Concorde had similar flight characteristics.

Recovery was very easy. Move the stick forward/relax back pressure in order to reduce AoA . Additional power helped, but the underslung motors on the 'bus seem to induce a nose up moment versus the centerline motor of the Deuce. Same true for the Viper, and the thing would simply start to descend with full back stick pressure once it reached the design AoA. Unless that puppy was in a deep stall, then recovery was the same - lower the friggin' nose!

- While looking at the reports and data, I conclude that the 'bus is a very stable and well-designed jet. So a stall may not be easy to recognize, especially if the crew is trained that "you can't stall this airplane". FBW and "protections" for another time, another thread, IMHO. But it may come up in the final report.

So I will make my prediction that the final findings will conclude:

- Crew applied inappropriate control inputs for the existing flight condition when data was lost ( regardless of the failures of sensors and such)

- Flight instrument displays and aural warnings were confusing.

- Crew entered a stall and did not recognize the stall nor apply appropriate control inputs to recover. In other words, the plane did exactly what the crew was demanding.

- The report will demand better training and crew coordination during unusual flight conditions.

Am I way off base?

[Lyman]

PJ2

02:11:45 FLR/FR0906010210 27933406EFCS1 X2,EFCS2X,,,,,,FCPC2 (2CE2) /WRG:ADIRU1 BUS ADR1-2 TO FCPC2,HARD

Here we are again at WRG. Followed by loss of displays, as noted by the pilots seven seconds later. Comments? The time is the same as that when the Captain re-entered.

[Machinbird]

Sure seems like someone is intent on spinning the hamster wheel up to record speed.:yuk:

[PJ2]

Lyman, re, "I don't see the problem." Nor do I, bearing in mind the broader question regarding the availability of the displays which I think is in keeping with the second of the three phases of this accident as outlined in IR3, (p72 - 76).

Re, "Here we are again at WRG. Followed by loss of displays. Comments?"

No, none.

There is nothing in the "WRG" indication in and of itself that is unarguable evidence for a loss of all displays.

Forgive me but we can go back and forth many times, as seen, and I've made my comments and asked my question.

I am entirely open to new information and would change my approach according to such new information but failing that I think it is not complicated pre-apogee, and post-apogee it is largely unpredictable (in the sense of citing cause > effect), in terms of human, aircraft and system performance. One must go where the available evidence leads, using one's experience, background and "sense of things".

[jcjeant]

Hi,

Gums
- While looking at the reports and data, I conclude that the 'bus is a very stable and well-designed jet. So a stall may not be easy to recognize, especially if the crew is trained that "you can't stall this airplane". FBW and "protections" for another time, another thread, IMHO. But it may come up in the final report.I repeat ...
You have no datas in the BEA report .. only cartoon graphics ...
For have a good idea of what was the actions of the pilots we need the FDR DATAS ! and accurate (sharp) graphics instead chalk drawings on a board like the BEA ones ...
Those will stop to a end the spinning wheel ....

[Turbine D]

Well, Lyman,
I learned a long time ago, as a Metallurgical Engineer doing failure analysis, you don't have to look at everything at 10,000X to find the defect that caused the failure. Often, most failure analysis can be determined looking at things from a much broader perspective, 1X. Analysis of an airliner crash is nothing more than a failure analysis, crudely as it may seem in these terms. Honing in on flap & slat setting or spoiler activity is looking at the problem at 10,000X or greater magnification. You need to back off and look at AF447 at 1X, once again. The BEA has graciously given you a considerable amount of data to study and understand. It is what they knew to be factual at the time they released IR #3. Not every organization charged with determining the cause of an accident would provide what you have received as of now. So begin to look at what you received as fortunate rather than incomplete.

What you should be focused on is the 1X. It starts with this from the BEA IR#3:

The Look Ahead

Cloud mass reflectivity depends on the type of air mass and on the season. Cumulonimbus reflectivity is not the same in temperate regions and below the equator. An oceanic cumulonimbus reflects radar waves less than a continental cumulonimbus cloud of the same size and height.10

Gain, tilt and the ND scale enable pilots to adjust the weather radar. Gain defines the level ratio between the signal received and the signal emitted according to the distance of the echoes. The CAL position of the gain control sets radar sensitivity at the standard calibrated level of reflectivity. The equivalence in precipitation is thus associated with a colour of the echoes presented on the ND.

Adjusting the ND scale enables monitoring at varying distances of the aircraft.

Heavy precipitation that returns most of the radar signal may also hide another disturbed area situated behind.

Representation of the weather situation by crews is thus mainly linked to the use of the 3 setting parameters and their knowledge of radar, particularly of its limitations.

Onboard radar does not directly detect dangers to be avoided and has specific limitations which require active monitoring from the pilots and constant analysis of the images presented to limit the risk of underestimating the danger of the situation. It should be noted that, at the time of the accident, the presence of ice crystals at high altitude was not considered to be an objective danger and that crews were not made aware of this.

In cruise mode above 20,000 feet, a slight downwards adjustment of tilt, depending on the scale selected, is recommended so that the ground echoes only appear on the ND at the edge of the furthest distance circles. This method enables the simple and practical application of the height/tilt rule of equivalence providing the optimum tilt adjustment.

When pilots monitor the weather situation, gain can remain in CAL position. In the confirmed presence of storms and during their avoidance, a manual adjustment can be used for comparison with the CAL image.

A scale of 160 NM enables the change in the weather situation to be assessed and anticipate route changes. A scale of 80 NM is used for avoidance. Short scales must be periodically discontinued in order to observe distant weather conditions and to avoid an impasse amid the disturbances.

Red or magenta zones as well as fringe-shape echoes must in this way be by-passed from windward by regularly adjusting the tilt and the range. The avoidance decision must be taken before the echoes are at 40 NM.

The operator recommends avoiding flying less than 5,000 ft above or below a storm cell. It provides a formula for pilots to estimate the separation height between the top of a detected cell and the airplane. This formula uses the distance and the tilt points from which the zone echo disappears. Above 23,000 ft, it is recommended to fly more than 20 NM from these zones.

How did the AF447 flight crew manage their radar? Did they sense what was ahead?


The Beginning of The End

Neither of the two copilots formally identified the stall situation that the airplane was in, either via the aural warning, or by recognising the buffet, or by interpreting the high vertical speed and pitch attitude values. It should be noted that buffet is the only indication of the approach to stall at high altitude on other airplanes whose stall warning threshold does not vary with the Mach.

In the absence of relevant information from the copilots, reading the information available on the screens (pitch attitude, roll, thrust, vertical speed, altitude, etc...) was not sufficient in itself for the Captain to become rapidly aware of the airplane’s situation. He did not then ask questions that could have helped him to understand the sequence of events.

The stall warning lasted 54 seconds continuously, during which time neither of the copilots made any reference to it.

Despite several references to the altitude, which was falling, none of the three crew members seemed to be able to determine which information to rely on: for them, the pitch attitude, roll and thrust values could seem inconsistent with the vertical speed and altitude values.

The investigation brought to light weaknesses in the two copilots: the inappropriate inputs by the PF on the flight controls at high altitude were not noted by the PNF through an absence of effective surveillance of the flight path. The stall warning and the buffeting were not identified either. This was probably due to a lack of specific training, although in accordance with regulatory requirements. Manual airplane handling cannot be improvised and requires precision and measured inputs on the flight controls. There are other possible situations leading to autopilot disconnection for which only specific and regular training can provide the skills necessary to ensure the safety of the flight. Examination of their last training records and check rides made it clear that the copilots had not been trained for manual airplane handling of approach to stall and stall recovery at high altitude.


By the time the Captain returned to the flight deck, it was too late, unless he recognized the situation immediately, he didn't. What ever happened from there on was futile, there was no time given the situation.

So flaps, slats or spoilers , or not, they didn't make a difference when they left the flight envelope and were in the stall descending at 124 miles per hour without recognition of being in a stall situation. I would suggest you, one more time, read pages 87-101 of the BEA IR#3, carefully. You will see (read Pg. 94) they briefly deployed spoilers for 3 seconds before retracting them and it does show on the cartoons as jcjeant calls them. Look at the bigger picture, Training (lack thereof), knowing the airplane and rapid situation recognition.

As for the final report, I think you will be disappointed, not every little nit will be contained nor should it be, only the pertinent data.

[jcjeant]

Hi,

Tubine_D
The BEA has graciously given you a considerable amount of data to study and understand. It is what they knew to be factual at the time they released IR #3You can call this spinning wheel or any name but the BEA no released the DATA (listings) of the FDR ( it's what I call "a considerable amount of data" )
This was a early material available to publish ...

[Lyman]

PJ2 Turbine D Organfreak, Machinbird.

Howdy. It is an honor to be in such company, and I consider it my good fortune to be allowed to participate here. We all bring different skills, experience, and demeanor. I am a low time commercial pilot who had the good fortune of owning a small aviation business at one time. I am in awe of the quality of discourse here, and to any extent you have stayed is testimony to your patience, or masochism.

This is no ordinary wreck, it is a criminal investigation. Due negligece, and various other human shortcomings, 228 souls are dead. 447 transcends the usual, and while it is tragic, it is nothing to discard without several microscopes to hand. No one is well served by being too easily satisfied. I understand everyone's approach, and also the frustration of those who believe they know what happened. If anyone can direct me to the Court's website, I will say goodbye. Failing that, I hope you can understand MY frustration at any sign of "laissez faire".

"Standard of Care". "Duty of care". Without these, and what you all know of aviation, no one would fly. anywhere. Ever. My opinion? Air France should be shut down, and reorganized. The corporate culture there is largely to blame for the accident. How do I know? I don't have to know. They know. No one is innocent in this industry, bargains are struck daily, and competition is fierce.

It costs money to park equipment and rnr air sensors. Motive? Dough. Result? these dead souls. My family flies that airline from time to time, their people are wonderful, and their equipment is modern, and safe. Knuckleheads who think they are smart, and can game the odds, these are the enemy. You know them, and not only in aviation do they cost lives.

Hamster Wheel? Small price to pay, if the Truth is outed. Don't give up. Don't let the knuckleheads skulk away, to risk your family, next time.

[RR_NDB]

Hi,

gums:

So I will make my prediction that the final findings will conclude:

- Crew applied inappropriate control inputs for the existing flight condition when data was lost ( regardless of the failures of sensors and such)

- Flight instrument displays and aural warnings were confusing.

- Crew entered a stall and did not recognize the stall nor apply appropriate control inputs to recover. In other words, the plane did exactly what the crew was demanding.

- The report will demand better training and crew coordination during unusual flight conditions. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-52.html#post7104218)

The "global picture" analysis seems correct. 1X as Turbine D properly emphasized: Often, most failure analysis can be determined looking at things from a much broader perspective, 1X. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-52.html#post7104250)


IMHO, under "unusual flight conditions" you need:

1) Reliable System "outputs". In your second point you put "...confusing"

2) "Graceful Degradation" and "Fault tolerance" from the System. There are indications the crew lost important (System) outputs during the "long" trajectory from apogee to SL: Under full stall conditions, a number of sensors, and therefore their downstream services, would not be working correctly.
(http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-51.html#post7102605)

3) In your point #3 the "entering into the stall", "not recognition of it and lack of "correct perception" does not (could) suggest the need for a review of certain aspects of the "man machine interface"?

4) "Unusual flight conditions" were not partly derived from "UNUSUAL conditions" PF and PNF (clearly lacking proper training) were submitted from the environment outside and INSIDE the cockpit?

My comment on your question:

Am I way off base?

Is:

The "man machine interface" will be an important factor in this accident or the training, CRM, etc. should "compensate" it's (sometimes complex) characteristics?

PS

What i mean by "characteristics" is some "well known" issues observed in these "advanced flying Systems".

So a stall may not be easy to recognize, especially if the crew is trained that "you can't stall this airplane".

PS2

"Advanced flying Systems" currently uses "RIDICULOUS DESIGN" with respect to "air speed" anomalies (by non availability of adequate sensors) during transients (like occurred in this accident) . Still applying erroneously the very basic concept of Redundancy. And relied on (non properly trained) pilots to use "band aids" in (sometimes) "possible difficult situations" typical of daily commercial flights.

PS3

And I think some of the inputs from we non-commercial drivers have helped.

For sure!

PS4

Crew applied inappropriate control inputs

A fact!

Question:

In this case (a complex one) pilot (PF) error will be the cause? All "inputs" he received (before and during the flight) like training and System "outputs" will be considered important (for his actions)?

[RR_NDB]

Hi,

PJ2:

The far more critical aspect of this accident is "Why?" (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-51.html#post7102448)

There are (yet) many (possible) and important reasons to answer "Why".

At present I think it is not complicated.

Please comment.




Old Carthusian:

Unfortunately it refuses to die. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-51.html#post7102420)

Why (it refuses...)?

[Old Carthusian]

RR NDB
The reference is to the non too disguised attempt by some to attribute the accident or parts of the accident to the machine and not the human, cultural, management and training issues. One, of course, welcomes a spirit of inquiry but one is forced to question why so many otherwise rational people are so excised by the possibility that the flight crew got things wrong and that this was the only reason for the crash. I am afraid that I keep coming back to the adage - we try to make the machine foolproof but God keeps on designing better fools (not that I think that the flight crew were fools, just that is the adage). No matter how well a machine is designed someone will find a way to evade it's safety features and wreck it. The man/machine interface strikes me as another red herring I'm afraid.

[rudderrudderrat]

Hi Old Carthusian,
One, of course, welcomes a spirit of inquiry but one is forced to question why so many otherwise rational people are so excised by the possibility that the flight crew got things wrong and that this was the only reason for the crash.

The Flight Crew Training Manual has this quote in Abnormal Attitudes Section:
"The effectiveness of fly-by-wire architecture, and the existence of control laws, eliminate the need for upset recovery maneuvers to be trained on protected Airbus aircraft."

I think that might be changed.

[chrisN]

I was talking to a commercial (bizjet) pilot at the weekend, who told me that sim training has changed considerably post 447. There is now much more on approach to stall/buffet (but not into stall), and emphasis on recovering controlled flight including nose down if that is what it takes, and not on minimum height loss only as it used to be.

Second hand, I know, but there must be some AB and other ATPLs here who can say if that is a general thing.

[CONF iture]

I think this because that is a significant operational emergency decision
It is part of FCOM SUPPLEMENTARY TECHNIQUES following an aural stall warning, nothing more nothing less :
If below 20000 feet and if in clean, select FLAP 1
Such action from the crew would be an indication on how they were or not understanding the situation.

as I mentioned, system behaviours in these circumstances are much less understood than in normal circumstances and, like the earlier examination and sorting out of the ACARS messages, a lot of permutations/combinations would have to be examined before a definitive sense of what was being displayed on the PFDs at any moment during the post-apogee phase.
That is why I would not rush to state that the CAS was the only parameter to momentarily disappear.

I don't suppose you would be very disappointed if BEA doesn't publish all 1,300 parameters from the FDR? Those who claim BEA is withholding data will never give up.
For now the BEA is selective ... but the Judge is withholding data from the proceedings.
Have you heard anything to the contrary ?

[jcjeant]

Hi,

BEA chalkboard graphics less FDR data is partial and inaccurate information
FDR data (listings) + sharp graphics are valuable information AFAIK and could reduce the field of speculation
I can understand that the full CVR will not be released to public for many obvious reasons .. but why not the FDR ?
What are the reasons ?
One "Why?" more ......

[PJ2]

Hello RR_NDB;


PJ2:Quote: The far more critical aspect of this accident is "Why?"
(http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-51.html#post7102448)There are (yet) many (possible) and important reasons to answer "Why".


The context of the question, "why?" is Phase 2 (as stated in IR3) of this accident: Why the pitch-up and then the continued pitch-up post-apogee? So, yes, there are remaining many possible and important reasons in answer to the question, "Why?" in this context but there are also a number of explanations which are less probable as to why. The focus on aircraft system anomalies after entry into the full stall and descent will not answer the question regarding Phase 2. The extreme stress and confusion/cognitive dissonance indicated in Phase 3 resulting from aircraft/system behaviours that were inconsistent with the crews' view that they were still flying despite the stall warnings (crew comments in IR3 regarding "high speed") and the recurring stall warning and airspeed indications, likely contributed to the significantly-decreased chances of assessment of the problem and swiftly reducing opportunity for recovery.

Quote: At present I think it is not complicated.
Please comment.Well, I have been commenting all along since last May up to these series of posts on why I think this is not complicated but I will try to make that uncomplicated!

First, "not complicated" doesn't mean that details of both human and aircraft behaviour are dismissed as sidebars or non-causal factors...not at all. But some theories and notions of what happened are more in accord with what we know and therefore more relevant than others which we have seen. It is not always easy to know when one is merely jamming a notion into the small box of known facts so that a theory fits a preconceived notion, or whether one is "on to something" and just needs to ask more questions.

But as more becomes known, one is able to parse such interesting activity and ideas and then tentatively set aside some notions and turn towards others which are more plausible, at least until something else comes along which hints or even proves that one might be off-base so that one can start again. It isn't a perfect way to do things but it works.

"Not complicated" here means, this is a loss-of-control accident which resulted from stalling the aircraft. Why did that happen?

That happened because the airplane was pitched up and held there until the stall. Why? This was done immediately, without announcement, standard crew coordination, (CRM) or following SOPs. Why? The PM made no interventions. Why? One simply never pitches a transport aircraft up to 10 to 15degrees at cruise altitude and holds it there because intuitively knows what will happen...a rapid loss of energy. We have to assume the PF was earnest in his response yet the response was not correct. Why?

Some details for this exact point would be the prior crew discussion regarding climbing and the FMGC MAX ALT message. That was a clear indication that the airplane was already very near its altitude limit in terms of available thrust to maintain cruise flight. Why go higher where the available thrust would not maintain level flight let alone a zoom-climb? That would be a detail. The climb happened versus not: Why?

In Phase 3 when the "stall, stall" with Master Warning and audible warnings were heard why was the stall warning not respected as per the UAS QRH checklist? A number of ideas have been posited and discussed and may explain this but the key point is, it happened, vice not. Why?

What is going to be very complicated is understanding the human dynamics and factors which led to certain behaviours and not others. We already know that training regimes, simulator scripts, awareness of high altitude flight dynamics and swept-wing transport behaviours as well as system behaviours are being taught everywhere but we may only be able to make guesses at the psychology of the flight deck in this accident and this is where we must be extremely cautious and respectful of the crew and of our lack of knowledge.

[Organfreak]

@PJ2:

Masterful summation.

:D

(I'd hire this guy!)

[Lyman]

To characterize post apogee Pitch inputs as "mostly nose-Up" is not fair. The a/c wandered NU/ND between -3 and +17 several times.The co-pilot input ND several tmes. He had no ability to see the AoA, as we know. To constantly insert AoA in the record, knowing the pilots had no access to it, is dishonest on BEA's part. Again, several times the Nose was down well below cruise values, and continued to descend rapidly; increasing speed, and further disorienting the PF/PNF. It IS fair to ask "Why no push on the Stick to gain perhaps 20 degrees ND?"

It was PNF (LHS) who was doing the flying when the Captain entered and said "Er, what are you doing" LHS continued to fly during the initial phase of three pilot command. I can suggest that pilots reread the chronology on pp93-94 with a fair and objective viewpoint, and keep reminding yourself you have PITCH only, not AoA. It is quite clear none of the pilots had a clue as to where to put the nose, keep it, and pull out. STALL was not in the mix. And remember, the STALLWARN is WARN, not STALLED, necessarily. At one point, both sticks were to the stops, left, and the a/c was Rolled right 4 degrees, and when one stick relaxed, it immediately Rolled right to 7 degrees. I still think there was a problem with the airframe, Rudder/Lateral, that favored Right Roll, chronically.

I think it is reactionary to dismiss the possibilities of 'desperate' measures, taken by crew. Also to 'assume' 'most' displays were 'working'. If they had been, why can I not also say, "Then why the confusion on deck?" Alot of pointing, alot of statements, alot of confusion, and consternation. This includes instrumentation prior to STALL, why would that be excluded.

I would think that a fair position would be: "With so much confusion, how could the instruments possibly be working? There is no record of displays, just as there is no record of the pilots solutions. These have to be inferred, as CONFiture has pointed out in his response to PJ2. I am frankly more puzzled by the preStall record, not the utter LOC afterward. When there is no reason to say "He simply pulled up, we don't know why", then there is no reason to claim incompetence.

Especially when given the circumstances. The record of instrumentation and displays rests on the Pilot's verbal record. Relying on the DFDR to bolster an argument when no such data was available to the crew is simply dishonest. We were not there, yet so many make accustaions as if they were. And yet I am accused of Hamster Wheeling? I am suggesting the record is incomplete. WAY incomplete.

[HazelNuts39]

For now the BEA is selective ... but the Judge is withholding data from the proceedings.The sole objective (of BEA's investigation) is to draw lessons from this occurrence which may help to prevent future accidents. Obviously it needs to be selective in presenting the data that have a bearing on that objective, both in the interim reports and in the final report. I know nothing about the Judge's actions, and why would you blame BEA for those - perhaps you can explain?

[jcjeant]

Hi,

I know nothing about the Judge's actions,HazelNuts39 .. as you are (mostly) located in Françe .. this will give you some light about the judges action
Hope it's help .....
Some news there: (audio included)
Rio-Paris : les familles réclament une enquête transparente - RTL.fr (http://www.rtl.fr/actualites/transports/article/crash-rio-paris-les-familles-reclament-une-enquete-transparente-7742464108)

[HazelNuts39]

jcjeant,

That's about the lawyers request (Les principaux extraits...), nothing about the judge's response, nor about BEA's responsibility in this. Doesn't the judge makes his or her own decisions, independent of BEA's investigation?

[Lyman]

BEA's responsibility is non-existent, vis-a-vis the Court case. One must read the Mission statement, and realize that there is sufficient discretion in it to frustrate even the most Liberal Magistrate, imho.

If there is a superior authority to both the Judge and BEA, that is where disclosure reposes.......Even then, the Judge can frustrate disclosure. Monkeys play football.

[HazelNuts39]

To constantly insert AoA in the record, knowing the pilots had no access to it, is dishonest on BEA's part.BEA simply describes the condition of the airplane, and doesn't in any way suggest that the pilots had that information.

Between 02:11:00 and 02:11:45 PF applied nose-up and nose-down inputs to maintain a pitch attitude of 15°, which required full travel of nose-up SS at 02:11:45. He then pulled the thrust levers back to idle, which caused the airplane to pitch down to -10° against full NU SS maintained until 02:12:15. Did I refer to AoA?

[Lyman]

HazelNuts39

"Between 02:11:00 and 02:11:45 PF applied nose-up and nose-down inputs to maintain a pitch attitude of 15°, which required full travel of nose-up SS at 02:11:45. He then pulled the thrust levers back to idle, which caused the airplane to pitch down to -10° against full NU SS maintained until 02:12:15. Did I refer to AoA?"

If I ask you who was PF during this time frame who would you say was flying?

Bonin from the RHS?

[RR_NDB]

Hi Captain

PJ2:

but we may only be able to make guesses at the psychology of the flight deck in this accident and this is where we must be extremely cautious and respectful of the crew and of our lack of knowledge. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-52.html#post7105390)

I understand your rationale and must admit, it is robust. HF (being addressed) probably is the major "component" of this accident.

When investigating something "different from our expectations" (i started to learn long time ago doing corrective maintenance in a powerful NDB (RR 375 Kc/s :)) we must be open minded to all possibilities.

Your thinking (on why's) i understand is concentrated in the crew (unexpected behavior) and considers the machine "performed as expected". As a designer i always question the design. It can have 'flaws". Better when we detect first. Cost less.

I understand your thinking that crew actions were "unexpected" before and after facing the consequences of the "inadequate inputs=large stimuli to the plane, zoom climb, etc." An "entire group failure": PF, then PM and finally CPT being unable to deal with the issues and not showing any useful timely "response".

Considering the "effective aircraft" (Machine+PF) failed VERY FAST (degraded) without any "useful reaction" (showing ZERO resilience) and considering (probably) there are some (perhaps important) factual information we don't have, I prefer for now (as safer):

To consider the man machine interface "performance" as a possible important contributing factor why the crew failed completely (acting unexpectedly).

There are some reasons for this.

I sincerely hope the lessons of this case (all possible) are learned and implemented with no "trade offs".

The user wants to travel not just in "competitive carriers" operating "competitive machines". The user really need "Uncompromised Quality". Safety as a direct result. This is what the users expects indirectly (or directly).

This case, a complex one presents a great opportunity to advance further toward the objectives of every one professionally conscious on these issues.

IMHO the man machine in these "advanced flying systems" should be reviewed specially when working under "unusual conditions". Exactly when you need most it's resources.

With the information we have so far in this case we have reasons to consider this as a possible and probable important factor.

Actually the authorities should pursue this objective, in the interest of all involved in the industry.

I am anxious to see the results of the HF study and the role of the man-machine interface on crew (lack of) "performance".

[jcjeant]

Hi,

HazelNuts39
That's about the lawyers request (Les principaux extraits...), nothing about the judge's response, nor about BEA's responsibility in this. Doesn't the judge makes his or her own decisions, independent of BEA's investigation? We should try to make an effort to understand what the lawyer says in this interview
It simply says that BEA has passed the entire FDR to Airbus company
He also said that Airbus is one of the parties named in the lawsuit .. as are other civil parties that is to say the associations of families of victims and people with right
Also it is not normal in terms of law that either party has the intelligence and the other does not ...
And it's even more weird that the judge handling the case should not take a decision to correct the imbalance .. as she refused to join the FDR in the instruction process (so far)
The families associations have the right (as civil party) to have their own experts ... but how they can make expertise with no datas ???
Hope it's help to understand what happen there ...

[PJ2]

RR_NDB;

I try very hard to leave hindsight bias behind and go where the available evidence and my experience and knowledge of the airplane take my thoughts. It doesn't always work, so disagreement, based upon further information and not just opinion, is always part of the process.

Believe me it is not pleasant to contemplate what has happened to this crew but as so many have said, it is crucial to find out what happened and why so that prevention takes place. In fact many things have already changed in the past year as a result of the available data in the form of recurrent training and so on.

It is a fact that airplanes, their systems and engines rarely fail mechanically, that navigation is extremely accurate, that autoflight is very good, that CFIT's and mid-air collisions are rare thanks to GPWS/EGPWS and TCAS, that satellites have made communications over vast areas of ocean almost routine, (though not here even though such wasn't likely an important factor), and that SOPs, CRM and things like checklist design have all contributed to the remarkable safety record of the industry.

It is therefore a fact that a very high percentage of accidents are HF accidents.

This is a primary reason for safety programs such as FOQA/FDM, which also have a crew-contact element (by a pilot's peers, not by management!), which is designed to address human factors issues and prevent untoward events.

So it is extremely important to understand what happened here and why, especially in Phase 2, and with two first officers and the ill-defined command and experience gradients which, I am sensing, may possibly also had had a cultural aspect.

These are SMS, HF and organizational areas of accident investigation and prevention. So, with reference to your comment, "Your thinking (on why's) i understand is concentrated in the crew (unexpected behavior) and considers the machine 'performed as expected'. ", I was not so much focused on the crew so much as it is where the available data is drawing attention. That means that new information, when or if it arises, always has the capacity to draw attention. This does not mean that design, and the machine is not the focus but that one places such attention in the context of the man-machine interface.

Designers cannot reasonably be expected to anticipate everything that will occur in an aircraft either by mechanical incident or crew action so it must be designed to fail gracefully as some have pointed out here, and this airplane failed gracefully - a loss of speed information does not necessarily result in a loss of control or loss of the aircraft. I don't think it is at all reasonable to expect that a designer will, in the course of such anticipatory processes, design against all outcomes that may or may not obtain in a fully-developed high altitude stall by line crews.

That said, even with a full NU THS, there was sufficient elevator authority and upwards force on the horizontal stabilizer and aft fuselage to lower the nose and un-stall the airplane albeit likely over as much as a 20,000ft loss of altitude. I think that is a remarkable bit of engineering. A level D simulator may not have the exact algorithms for post-stall behaviour but nor is it entirely without data and fidelity in such conditions.

Crew confusion must be examined very closely in both Phase 2 and 3. Post-pitch-up the PM was confused by the initial two very short stall warnings for example. Who knows how that may have influenced subsequent perceptions and input? The other aspect which I expect will be examined in the final report is the behaviour of the stall warning system below 60kts but this has been widely discussed.

It remains a concern as a former transport pilot that potentially irreversible actions were swiftly, unilaterally taken without adhering to SOPs, CRM communications standards and the handling of abnormals. Cockpit discipline and TEM (Threat/Error Management) processes are drilled into crews in each simulator session and are causes for failure of the ride if not executed to high standards. These processes intervene to prevent rushed actions while providing a basis for calm, measured and coordinated responses by both crew members. This isn't some elusive ideal, this is the normal standard by which transport aircraft are flown, so any unexpected divergence from this standard requires explanations and a willingness to closely examine crew actions where the data supports that kind of an approach. That is why I seem focused on the crew.

[DozyWannabe]

Heavens to Betsy, is this still going on?

To characterize post apogee Pitch inputs as "mostly nose-Up" is not fair.

Yes it is. When talking about average positions over time the inputs were *overwhelmingly* NU.

The a/c wandered NU/ND between -3 and +17 several times.

Yes - in response to the stall. Speed decays, nose comes down, speed increases but the aircraft pitches up and stalls again due to elevator/THS position.

The co-pilot input ND several tmes.

Not enough, and for nowhere near long enough. When we ran the scenario in the sim, the SS needed to be emphatically forward and held there for at least 8-10 seconds to correct the THS position and control the aircraft. This is speculation on my part, but having seen the PFD myself I suspect the PF was "chasing" the ADI indication in the same way he'd follow the FD.

The key aspect is that in order to perform the correct recovery procedure, the pilot in control needs to first understand what the problem is, and then perform a recovery based on the information they receive. For whatever reason the flight crew of AF447 didn't even get to understanding the problem, and the PF's responses were instinctive and consistently reactive.

It was PNF (LHS) who was doing the flying when the Captain entered and said "Er, what are you doing" LHS continued to fly during the initial phase of three pilot command.

No, he was never in control for more than a few seconds at a time, because the traces show the PF taking back control almost immediately without announcing his intent to do so.

And remember, the STALLWARN is WARN, not STALLED, necessarily.

I'd say if it's been sounding for a minute, there's a good chance that you're in the stall regime. This is immaterial however, as there's no acknowledgement that they ever even heard it.

At one point, both sticks were to the stops, left, and the a/c was Rolled right 4 degrees, and when one stick relaxed, it immediately Rolled right to 7 degrees. I still think there was a problem with the airframe, Rudder/Lateral, that favored Right Roll, chronically.

I don't know how much more simply I can put this. Lateral control via the ailerons is not reliable when this or any other aircraft is in the stall regime. Basic aeronautics - ailerons work by directing the airflow over a section of the wing. If a wing is stalled, it cannot produce lift, ergo if an aileron encounters stalled air it will not respond as expected. This is why full stall training at PPL level has the pilot use the rudder to level the wings via the side-effect of roll if necessary.

I think it is reactionary to dismiss the possibilities of 'desperate' measures, taken by crew. Also to 'assume' 'most' displays were 'working'.
...
I would think that a fair position would be: "With so much confusion, how could the instruments possibly be working?

Simple - these were FOs trained in the glass cockpit era. Most of their airline hours have in fact been with autopilot engaged. On the occasions they've flown manually, it has been done with the FD engaged and with the speed tape on the left clearly marking the safe range. They've not done any manual handling on basic instruments in that type other than possibly in canned simulator scenarios. It's like riding a bike or playing an instrument - the knowledge will always be there, but unless it's practiced regularly you will not be able to perform at your best in a scenario where you're suddenly called upon to do so.

I believe that when he says "we don't have any displays", he is referring to the displays that he has come to rely on for the entirety of his time in manual control of that type, namely the FD and max/min speed tape. I believe that the basic instruments were still available, but that he had come to rely on the electronic guides to the point that he was afraid of having to go back to basic panel flying.

I believe that the initial NU demand was an overcorrection to the turbulence that left them a touch ND in pitch, which started a series of instinctive panic reactions, complete breakdown of instrument scan and failure to calmly assess the situation. I don't believe this was entirely his fault because he was not properly trained for the situation that confronted him.

[Organfreak]

Dozy,
Heavens to Betsy, is this still going on?

Hell, we're just warming up!

:bored:

[Turbine D]

Lyman,

You are still concentrating on the examination at 10,000X magnification.

Does it really make a difference who was piloting the aircraft between 2 hr 11 min 00 and 2.hr 11 min 45 and which seat they were seated in? The BEA made it clear on Page 20 of IR#3 the AoA was not directly displayed to the pilots. Further, if you read Pages 29-31 you will see some of the data you say was not presented to the pilots that was presented as they referenced it in their conversations. I will let it up to you to decide which ones, although it is pretty clear.

PJ2 & RR_NDB,

You both have hit on some very important aspects of this event IMHO. The question of "Why?" is of paramount importance in problem solving any situation or event that is out of the ordinary. The machine - human interface is always an important aspect to look at and review. There is a technique to be applied when asking "Why" to get at a root cause. It is a matter of asking it five times, giving a response between each "Why". At level five is generally found to be the root cause from which corrective action can be taken. It is a process that works. However, it is more complicated when you can't interview a key person in the problem loop. When this happen, you have to make a list and go through the process. So PJ2, lets take your first question and explore: "Why, when the autopilot disconnected, and manual flying was demanded did the PF pull the sidestick back commanding a continuing climb, Why? Well, there can't be an interview, so we have to provide multiple answers for consideration:

a. He was shocked it happened, the "Startle Factor".
b. He responded based on his training and SIM experiences
c. He responded based on his recollection of the "Memory List".
d. He --- --- --- ---

For each of these you ask "Why" four more times and what you will come up with is a rather concise short list, sometimes, one item that defines the root cause. Now I can't contribute to this as only experienced personnel intimately familiar with the aircraft (the pilots) need to compile the list and ask "Why". Once you have determined (or think you have) the root cause, you can then turn the root cause over to experts to begin the process of providing the solution, along with your input as to the quality of the solution decided as being best.

A simplified example would be this:

The power failed on climb out at 1,700 feet on a Boeing 767, Delta's Flight 810 from Los Angeles to Cincinnati on June 30, 1987.
The Captain with 29 years of commercial flying experience failed to coordinate with the First Officer and reacting to an amber light, warning him of a fuel-flow problem, pulled two round knobs cutting off fuel to the plane's two engines instead of pushing square buttons two inches away that would have corrected the problem.
"Why"?
To make a long story short, he had flown the B-727 for years and had transferred to the then, new B-767. When this event occurred shortly after liftoff, he applied from his "experience and memory", without looking, exactly what was required on the B-727 but not the B-767. The initial fix was to cover the knobs and square buttons with a plastic cover that you had to "look at" to open, but the real solution was to move the engine control panel to the overhead console. Training was also found to be defective at the time at Delta regarding entry into the B-767 which was corrected.

[Lyman]

TD

A friend drives FBW, not Bus. He was cruisin along, sippin coffee when the a/c started a sweet turn to the right. Both pilots noticed immediately, and switched off the data drive, and flew HDG to the destination. He bird dogged the event with others in company, and no one could explain what happened. He did run across others it had happened to.

So this a/c will climb "unexpectedly" and Uncommanded, (to QUOTE Airbus) when UAS happens. Instead of "Inferring" proper instrumentation for 447, and without any conclusive rejection of "AUTOZOOM", I'd like to keep an open mind. In the meanwhile, the HF, which I have been trying to discuss (from a defensive pov, I'll admit), wants a clear and microscopic overview.

So far we have a too quick PU from RHS, a consequent loss of orientation (both airframe and crew) and that's it. This wreck can be argued to have happened within several seconds of autoflight loss, since all sense of PITCH and AOA slithered away through the fingers of our impatient pilot. POSSIBLY. The entire public perspective misses that fact. For want of a breath, and a count to five, the flight was lost. Fully half of the culpability for the crash after this time frame belongs to AF training, AIRBUS complacency and overconfidence, plus an arguably criminal case against AF for deferring a non deferrable.

I haven't even cranked up the dark field, or the scanning EM. Not to mention the gas chromatography room. Sometimes it is simple. Did I just frame the dilemma, simple to impossibly obtuse?

[RR_NDB]

Hello PJ2,

Thank you for your very good answer. Sincere and constructive.

Concerning this segment:

Designers cannot reasonably be expected to anticipate everything that will occur in an aircraft either by mechanical incident or crew action so it must be designed to fail gracefully as some have pointed out here, and this airplane failed gracefully - a loss of speed information does not necessarily result in a loss of control or loss of the aircraft. I don't think it is at all reasonable to expect that a designer will, in the course of such anticipatory processes, design against all outcomes that may or may not obtain in a fully-developed high altitude stall by line crews. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-53.html#post7106123)

I would like to comment something now (and think and elaborate later):

It seems very, very important to have ALWAYS available a basic set of resources in order to allow the very basic need to just aviate when facing extreme conditions. This is mandatory for a "Graceful Degradation" of the "effective aircraft" (A/C+PF/PM). The incomplete factual information we have suggests the crew may not had this may be adding to the difficult conditions they entered (by their own errors, e.g. "entering WX", lack of perception after AP and A/THR quit leading ultimately to stall, etc.). IMHO the interface design (a real challenge due the complexity of the Systems) may be improved (using lessons from this case) if the HF study being developed and the analysis shows that. The AoA indication may be a result of this process. I am not expecting the need (or possibility) of a major redesign or change in Airbus SAS FBW "protections philosophy", obviously.

Crew confusion must be examined very closely in both Phase 2 and 3. Post-pitch-up the PM was confused by the initial two very short stall warnings for example. Who knows how that may have influenced subsequent perceptions and input?
Important facts that were "important inputs" to the crew (and their perception).

It remains a concern as a former transport pilot that potentially irreversible actions were swiftly, unilaterally taken without adhering to SOPs, CRM communications standards and the handling of abnormals. Cockpit discipline and TEM (Threat/Error Management) processes are drilled into crews in each simulator session and are causes for failure of the ride if not executed to high standards. These processes intervene to prevent rushed actions while providing a basis for calm, measured and coordinated responses by both crew members. This isn't some elusive ideal, this is the normal standard by which transport aircraft are flown, so any unexpected divergence from this standard requires explanations and a willingness to closely examine crew actions where the data supports that kind of an approach. That is why I seem focused on the crew.

Well put, i agree with you. This "organizational aspect" is paramount.

I must admit your "focus" seems correct.

Please, regard my "man machine emphasis" as a very important input to the crew "output". (e.g. the stall warning system below 60kts, etc.). And very probably to be considered as a "contributing factor" in the final report.

Despite having a "technically oriented mind" (:8) i learned very early the importance of the "organizational aspects", certainly with "higher hierarchy".

[Lyman]

If a pilot can follow each "degrade", well and good, otherwise "Graceful Degradation" can be a plot in the cemetery. Think about it. It sounds wonderful, but to synch each degrade with its counterpart, its replacement, describes a dance, no less. Here, the pilot should have hesitated to suss Attitude (so we are told). What was graceful about a cavalry charge, Master Caution, and a maneuvering a/c? ECAMS? Turbulence? Duff speeds? Say PF got it right, there followed three more instances of UAS, at a time when Airbus had not decided what to tell each crew. Reselect Autopilot? Or NOT? Careful, one chance only.

Arthur Murray can teach dancing in the studio, but two years hence, when the music is still, and the orchestra on leave, will John remember what to do with his hands? With his feet? Dancing is not natural to most pilots.

A fading away of a/c response must be synchronized with a building skill set from a pilot? Each gentle nuance recognized and bade farewell? It's a freaking machine. Fix the wrong stuff, train the right stuff, and let's keep moving. ffs.

[RR_NDB]

Hi,

Turbine D:

For each of these you ask "Why" four more times and what you will come up with is a rather concise short list, sometimes, one item that defines the root cause. Now I can't contribute to this as only experienced personnel intimately familiar with the aircraft (the pilots) need to compile the list and ask "Why". Once you have determined (or think you have) the root cause, you can then turn the root cause over to experts to begin the process of providing the solution, along with your input as to the quality of the solution decided as being best. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-53.html#post7106179)

:ok:


Will use the tool to test with some problems. :8

Please, references on it. Thank you.

[gums]

Long time, and long time for many of us.

I salute Doze for actually trying the profile in a sim, and I think that another here has access to the sim.

RR and PJ and Doze and others have conferred on PM's, so my observations are for others.

The human factor aspect of this tragedy is a big factor. Why is that, Gums?

- "You can't stall the jet, but you can overspeed the jet". Need inputs here from Okie and Rudderrat and maybe PJ2. So maybe the PF was more worried about mach than stall. And my personal choice when airspeed goes FUBAR is to not worry too much about overspeed as much as under speed. Of course, a valid AoA indication is immensely helpful, but the 'bus seems to ignore the sensors if airspeed is below "x".

Before exploring all the manuals those here have made available, I was not aware of the small AoA margin that the 'bus has between cruise and stall. I have to admit that I was shocked at the low AoA values I saw, as my experience in bent wing jets exhibited much higher AoA values before getting into trouble, the VooDoo being the big exception ( ask me about that beast offline, heh heh).

Ditto for mach limits when at 35K and above.

- The stall warning was there, but for some reason was not given attention by the crew. Did I read all those paragraphs and recordings correctly? Could they have been more concerned with overspeed than a stall? Remember, "you can't stall this jet", but there's no "protections" concerning overspeed. Hmmmm.....

- Some old pilots here harp on attitude and power to stay where you were when things went FUBAR. I am one. But seems like the jet has this auto-throttle feature, and I am not sure when the pilot can tell the "system" that throttle position commanded by the human in the cockpit should be obeyed.

Make no mistake, I flew the Viper with the electronic engine control that did things when we were real fast or slow ( not for us, but for the engine!). It was NOT an auto-throttle, and our PLA ( power lever angle) command was just that. It commanded a level of power that we mortal pilots desired/commanded.

Is there a way to tell the "system" that you want the throttle(s) to provide a direct power command and not be connected to all the flight control modes?

- "TOUCH" ----- I mean that sometimes your "touch" or your "feeling" counts as much as all the fancy indicators and displays( apologies to Doze, but trust me, I had/have touch). I flew thousands of hours with Joe Baggadonuts in three different jets and had many that could not "feel" the jet entering a stall. "Can you feel that?", I would ask. "Nope". So I then trained them to be "mechanics" and read the instruments and not horse the jet about.

I get the impression that many of the low-time 'bus drivers have never been close to overspeed or a stall. As Retired and 'bird and Smilin' and others here will attest, the plane always talks to you. The "feeling" of the jet and the gauges/indicators come together so you can confirm what your state is.

But you have to train the crews to "feel". If they can't "feel" then you must help them to interpret the gauges and such.

[rudderrudderrat]

Hi dozywannabe,

Good to read you back again and I agree with your post.

Additionally from the present A320 FCTM:
"The so-called "abnormal attitude" law is :
• Pitch alternate with load factor protection (without autotrim)
• Lateral direct law with yaw alternate
These laws trigger, when extreme values are reached:
• Pitch (50 ° up, 30 ° down)
• Bank (125 °)
• AOA (30 °, -10 °)
• Speed (440 kt, 60 kt)
• Mach (0.96, 0.1).
It is very unlikely that the aircraft will reach these attitudes, because fly-by-wire provides protection to ensure rapid reaction far in advance. This will minimize the effect and potential for such aerodynamic upsets.

The effectiveness of fly-by-wire architecture, and the existence of control laws, eliminate the need for upset recovery maneuvers to be trained on protected Airbus aircraft."

I think the last paragraph may have to be changed.

[DozyWannabe]

Good to read you back again...

Thanks, RRR - thanks guys. Although I must confess that I thought we'd be talking about something different by now and that the hamster would be having a snooze until the final report appears.

I think the last paragraph may have to be changed.

With 20/20 hindsight, I'd agree - with the caveat that the airframe never reached an orientation sufficient to trigger ABNORMAL ATTITUDE law, and to the best of my knowledge, not many have, if any. I think that the FCTM was written referring to the kind of "jet upset" that tended to happen in the early days of jet airliners, whereas what happened here was an aircraft ever-so-slightly askew but basically stable being handed to the PF, who then proceeded to control it into a stall.

[Clandestino]

@rudderrudderrat:

Whose FCTM is this? Stating that

The effectiveness of fly-by-wire architecture, and the existence of control laws, eliminate the need for upset recovery maneuvers to be trained on protected Airbus aircraft. is IMHO criminally negligent. FBW has known degradation and failure modes, insisting that they will always work properly when proverbial hits the fan is ignorant at the best. Chances of having to perform GPWS maneuvre in direct law are so small they can be ignored but Airbus pilot must absolutely know when to rely on protection and when it is not available.


This links into what I believe are relevant cultural issues at Air France particularly in the area of crew training.
Could be but it's not necessarily so as a few AF crews have previously successfully negotiated temporary loss of all airspeeds. Question is what made crew act so differently form every other crew facing the same problem. That HF group report will be a bomb is quite safe bet.

We certainly will have more knowledge about this accident when (if ever) the FDR DATA will be published ... instead of those graphics from the BEA belonging more to cartoons than technical graphics
(...)
I repeat ...
You have no datas in the BEA report .. only cartoon graphics ...
(...)
BEA chalkboard graphics less FDR data is partial and inaccurate informationGraphic presentation of AF447 DFDR readout is accurate, consistent and informative enough for anyone able to read it. Airline pilots of sub-average grade and above should be able to understand what is drawn. Of course, there are lot of PPRuNers who are either unable to understand BEA data or insist on misunderstanding it as graphs contravene their pet notion of Airbus being intrinsically evil. "Why" is good question.

From the data traces I see no extraordinary efforts by the crew to recover from a stall such as using flaps, spoilers, etc. In other words, I do not think the crew knew that the plane was fully stalled, and the control inputs do not reflect appropriate stall recovery techniques/procedures.You are correct but it's even worse; CM2's reaction to stall warning was surprisingly consistent and fatally wrong; first time stall warning goes off as the nose is jerked up, he just keeps pulling, second time, as the energy bled off, he pulls inanely. Third and fourth times as he released the stick, AoA decreased and air data became valid again, he pulled again, sealing the flight's fate, never commenting what and why he was doing.

So a stall may not be easy to recognize, especially if the crew is trained that "you can't stall this airplane". FBW and "protections" for another time, another thread, IMHO.If that's indeed so, it's a good thing there's synthetic voice shouting "STALL STALL" at A330 pilots. Cricket too.
That A330 or any other FBW Airbi cannot be stalled is severe misunderstanding and if the crew were really officially told so, then their trainers should be taken off duty until retrained. There's not magical about FBW alpha protection, it's just a mechanical-electronical device which limits pitch control in order to keep alpha below critical. Just like on 'Vark or Viper. It needs valid sensory input, computer and output to elevator. Something gets banged up in the chain - it doesn't work and it's easy to promptly see if it dropped offline, if one pays attention to his artificial horizon, that is. Nothing startrekkish about it.

There is now much more on approach to stall/buffet (but not into stall), and emphasis on recovering controlled flight including nose down if that is what it takes, and not on minimum height loss only as it used to be.I sincerely hope this is not the whole story.

IMHO it won't hurt to repeat what must be readily understandable to any pilot at any time of day or night: there is a whole world of difference between low and high level aerodynamic and engine performance. People who keep insisting that maybe CM2 was trying to replicate extremely low level manuever at FL330 are maybe onto something but usually go tangential when they go on blaming the training for it. For Finnegan's sake, any pilot must be fully and correctly aware at any time of what his aeroplane is capable of doing - if not, he has no business being at the controls. What was so far trained was avoidance of stall at low level because final turn stall was quite common killer and still is in the general aviation. What is commonly misunderstood in PPRuNe discussions is that airliner pilots are trained to recover from approach to stall, not fully blown stall recovery. That's because the only thing that can extract any aeroplane from low level, low energy stall is excavator.

High altitude stalls were never of concern because it was assumed that pilot would readily understand that they need to keep the speed up to avoid stall and trade altitude (of which there's abundance at typical cruise level) for speed if they got on the back of the power curve. Real life with lack of high altitude stalls seemed to confirm the notion. Than we had Pinnacle 3701. That accident was quickly forgotten as it was ferry flight, so no public outrage over passengers' deaths, and ascribed to cowboyishness of the deceased pilots. Heck, on PPRUNe there was ever-present "blame it on the technology" brigade who insisted that core-lock phenomenon was main culprit in the crash. It is correct that pilots undid themselves but they were not suicidal, they were just ignorant and that killed them. Did the powers to be push for better pilot selection and training post Pinnacle and Colgan? No. Just more hours that should serve as panacea for all the holes in the pilots' knowledge. Learn as you go.

IMHO, no amount of unusual attitude or high altitude full stall recovery training is going to prevent AF447-like (or Colgan, Armavia, Gulf Air, Ethiopian, Kenyan....) accident form recurring! We are not discussing about aeroplane being thrown form the sky with crew failing to recover, we are dealing with disoriented and confused crew that kills itself and everyone on board by doing the exactly wrong thing while believing it is right! It is not about recovery, it is about recognizing one is confused and getting to grips with reality ASAP. If this is dealt with soon enough, no radical recovery actions are needed, if not, it might be too late for airframe strength or altitude available anyway.

I was not aware of the small AoA margin that the 'bus has between cruise and stall. I have to admit that I was shocked at the low AoA values I saw, as my experience in bent wing jets exhibited much higher AoA values before getting into trouble, the VooDoo being the big exception ( ask me about that beast offline, heh heh)....and yet no airliner falls from the sky because of it. There are a lot of things common to all aeroplanes but some fine details are either applicable to subsonic, stable transport, some to supersonic, maneuverable interceptor bot not to both.

there's no "protections" concerning overspeed.Normally, there is high speed protection which, very unsurprisingly, doesn't work without valid airspeed.


But seems like the jet has this auto-throttle feature, and I am not sure when the pilot can tell the "system" that throttle position commanded by the human in the cockpit should be obeyed.To be pedantic: it's called autothrust and unlike autothrottle doesn't have servos that physically move the trust levers. It dropped off as speed become unreliable and anyway it can be disconnected by using the red button on either thrust lever. With ATHR off, it acts just as you wrote:

our PLA ( power lever angle) command was just that. It commanded a level of power that we mortal pilots desired/commanded.

[HazelNuts39]

PJ2's reference to IR3, (p72 - 76) made me read those pages again. I was puzzled by the following reference to the FPV, since the trace on page 107 shows that HDG-VS selected all the time (my bolding):
At around 2 h 11 min 42, the Captain came back into the cockpit, (...) Neither of the two copilots gave him a precise summary of the problems encountered nor of the actions undertaken, except that they had lost control of the airplane and that they had tried everything. In reaction, the Captain said several times “take that”, doubtless speaking of the FPV. (...)There also seems to be a translation error further down on the same page, where "nose-up" should read "nose-down" (à piquer in the french original): Several nose-up inputs caused a decrease in the pitch attitude and in the angle of attack whose values then became valid, so that a strong nose-down input led to the reactivation of the stall warning.Plusieurs actions à piquer provoquent une diminution de l’assiette et de l’incidence ...

[DozyWannabe]

I was puzzled by the following reference to the FPV, since the trace on page 107 shows that HDG-VS selected all the time

On both sides?

[jcjeant]

Hi,

Clandestino
Graphic presentation of AF447 DFDR readout is accurate, consistent and informative enough for anyone able to read it. Airline pilots of sub-average grade and above should be able to understand what is drawn. Of course, there are lot of PPRuNers who are either unable to understand BEA data or insist on misunderstanding it as graphs contravene their pet notion of Airbus being intrinsically evil. "Why" is good question.Do you have the FDR listing for support the BEA graphics ?
The answer is NO
Information is not complete .....
FDR data are not "emotional" like CVR (with all the speculations about each words) ... that's just technical data .. factual information
Why release fragments of the CVR and some graphics but nothing about FDR listing ?

[rudderrudderrat]

Hi DozzyWannabe,
On both sides?
The crew didn't selected their FDs off - the bars simply parked out of view when the Airspeed was considered invalid. When the airspeed became valid again, the FDs would reappear.

I can't find the FCOM reference - but I think on reappearance they default to the aircraft's present HDG & VS until reprogrammed.

[TTex600]

Dozy, I can't speak for the 330, but the narrowbodies only have one button on the FCP for FD type. If you select FPV, it applies to both sides.

Other than that, I concur with, "Heavens to Betsy, is this still going on"

[Machinbird]

There also seems to be a translation error further down on the same page, where "nose-up" should read "nose-down" (à piquer in the french original): Quote:
Several nose-up inputs caused a decrease in the pitch attitude and in the angle of attack whose values then became valid, so that a strong nose-down input led to the reactivation of the stall warning. Maybe not. Too much AOA on a surface can have the opposite effect and it can "lose its grip" and allow the aircraft to develop a pitch rate in the opposite direction. Then as AOA decreases, it would normally regain its "grip" and start pitching the nose back up creating the nose up and down bobble, but if the crew then relaxed their back stick at the bottom of the maneuver, the nose might continue further down than its normal range and allow airspeed to become valid and reactivate stall warning.

[HazelNuts39]

@TTex600:
Same on A330: Picture and description of FCU are on page 45 of IR#2.

@Machinbird:
Maybe so, but that doesn't make the English version a correct translation of the French original text.

[Machinbird]

Machinbird:
Maybe so, but that doesn't make the English version a correct translation of the French original text.
Oops, so you were basing that on the French original.:O
Google translate was no help.

[PJ2]

RRR; (like the handle...)



"It is very unlikely that the aircraft will reach these attitudes, because fly-by-wire provides protection to ensure rapid reaction far in advance. This will minimize the effect and potential for such aerodynamic upsets.

The effectiveness of fly-by-wire architecture, and the existence of control laws, eliminate the need for upset recovery maneuvers to be trained on protected Airbus aircraft."

I think the last paragraph may have to be changed.


After one's initial PPC on a new type, approach-to-stall training is not required by the CARS (Canadian Air Regs):

(e) Manoeuvres
(i) At least one steep turn in each direction with a bank angle of 45° and a change in heading of at least 180° but not more than 360°;
(ii) Approaches to stalls

For the purpose of this manoeuvre the required approach to a stall is reached when there is a perceptible buffet or other response to the initial stall entry.

The following approaches to the stall are required during initial and upgrade PPC's:
(A) one in the take-off configuration, except where a zero-flap take-off configuration is normally used in that model and type of aeroplane;
(B) one in a clean configuration; and
(C) one in a landing configuration.
One of the approaches to stall shall be performed while in a turn with a bank angle of between 15° and 30°.
(iii) Steep turns and approach to stalls are not required when the PPC is conducted using either a LOFT scenario, a scripted PPC or a fly-by wire aeroplane; and
(amended 2000/12/01; no previous version)
(A) for an initial PPC on aeroplane type, steep turns and approach to stalls have been satisfactorily demonstrated during initial training;
(B) for a semi-annual or an annual PPC:
(I) steep turns and approach to stalls that are required in the applicable annual training syllabus have been satisfactorily demonstrated during this training; or
(II) steep turns and approach to stalls are not required in the applicable annual training syllabus.

[mm43]

Several nose-up inputs caused a decrease in the pitch attitude and in the angle of attack whose values then became valid, so that a strong nose-down input led to the reactivation of the stall warning.Post #1690 (http://www.pprune.org/tech-log/456874-af-447-thread-no-5-a-85.html#post6624280) in AF447 - Thread No.5.

[DozyWannabe]

The hamster has also noted that a certain contributor is still swearing blind that the climb was uncommanded, despite all evidence to the contrary, and the hamster is heartily sick of it.

[PJ2]

thanks mm43. Takata's #1692 is also interesting in that it touches on the baro-inertial vertical speed parameter! HtB...wait!...there's more! ;-)

Dozy, gums, nice to see you back for a visit and contributions. I look in once in a while and this time decided to join in for a bit but need to return to standby mode.

[roulishollandais]

"Effective aircraft" does not mean that you have, from left to right, a box for "human", an other independant "man-machine interface" box, and a third independant box "machine/system".

The three are mixed inside the effective aircraft. It is the interest of this concept to say that. The study of instabilities as Aviation SAfety and Pilot Control (so early as 1997) shows how it is wrong to imagine the system is well or faulty and the man is faulty or well, and shows it is the best way to continue with oscillations and divergences, and to never improve.

As any dynamic system, the AF447 effective aircraft has mandatory two qualities :
1. it must be observable :ooh:
2. it must be controllable :ooh:

1. To be observable the matrix describing the effective aircraft must include all the men-parameters, and all the system-parameters, it is a very big matrix, much bigger than this one from an non fbw aircraft. The correspondant sensors have to be connected.

To be controllable, the determinant of this matrix is not allowed to be zero. Correspondant actuators must be connected.

The man-machine interface is not an isolated box, but concerns many functions of the big matrix who must not be faulty, and allows the effective aircraft to be always observable and controllable.:eek:

The human factor is very much more complicated that a short description of psychologically confused crew in one isolated box It has his place in any of these many functions of the effective aircraft matrix.:*

The designers of the aircraft have to analyse very closely if all these many functions do what is expected from them.:sad:

For instance, if it is only the system who "observes" the AoA ("incidence" in french) (AoA present in the traces) , but the pilot has to take decision, to control the aircraft, we doubt about his real ability to control the effective aircraft without these information, and we can guess something is wrong in this architecture. Then the pilot is no more able to achieve his pilot task. :bored:

It has also to be said that the law does the airline who sells the ticket, responsible, is it directly or through his agent failure.

[PJ2]

roulishollandais;

Re, "As any dynamic system, the AF447 effective aircraft has mandatory two qualities :
"1. it must be observable
"2. it must be controlable"

Je'm'excuse - Please forgive me if I misunderstand, but it is already this way in transport aircraft.

I'm not sure what you mean by "effective aircraft", but having flown Douglas, Boeing, Lockheed from 1973 and now the Airbus aircraft since 1992 these aircraft (A320 series, A330/A340 series) are eminently, (éminemment), "observable and controllable". They do not present unusual difficulties which require greater skill or knowledge than other transport aircraft today.

Nor do the accident rates indicate a large difference between aircraft types.

However, no aircraft, no design I know of is controllable or observable when it is taken into a full stall.

AF447 was recoverable even after entry into the stall but it required that the stick be pushed fully forward and held there until the wing began flying again. That would take between 15,000 and 22,000ft (I've flown this in the sim many times). This is Machinbird's "unloading of the wing" to which he referred some pages back.

Any transport aircraft in which the controls were moved in a way so that the stall of the wing is maintained as they were in AF447 would remain stalled.

Said another way, a B777 pulled up in the same manner and handled the same way as this aircraft was would also crash.

Also, I do not buy the sidestick vs control column argument one bit. Any pilot watching the pitch attitudes seen here does not need sidestick or column position to tell him/her that something extremely serious is about to happen if control of the aircraft isn't taken over immediately and the nose lowered to normal cruise attitudes.

This is the part that is very definitely not complicated.



Quite frankly, when I started, I flew with WWII guys who, if that kind of flying was ever done with the airplane he might break your arm while taking control from you. These guys were not pleasant to fly with but they knew how to stay alive in marginal conditions and made sure everyone else learned, one way or another.

Thank goodness those days are gone (because they also did stuff that scared the living daylights out of me), but today we talk about "managing" an airplane through the FMGCs and Autopilot instead of flying it. I recall watching someone actually try programming the Stadium approach to 31 at LaGuardia because he was uncomfortable flying it and that was a long time ago now.

In the present system, 99.9% of flights work well with SOPs, CRM, appropriate use of automation (according to enlightened airline policies which permit hand-flying), but the loss of such skills is nevertheless no longer a blip but a trend.

[TTex600]

Thank goodness those days are gone (because they also did stuff that scared the living daylights out of me), but today we talk about "managing" an airplane through the FMGCs and Autopilot instead of flying it. I recall watching someone actually try programming the Stadium approach to 31 at LaGuardia because he was uncomfortable flying it and that was a long time ago now.

Now it's the Expressway visual to 31. As in the Long Island Expressway. Sadly, many guys build a place/bearing/distance off of DIALS and link it to the course fix for 31. Cross DIALS at 180kts, gear down and flaps 2, dial in some vertical speed, drop the rest of the flaps, manage the speed and all you have left to do is click the AP off on final. The techno savvy FO's look at me like I'm a fossil when I click off everything at DIALS and just fly it by the seat of my trousers. :eek:

[Old Carthusian]

Clandestino
If I can perhaps explain further the situation is a lot more complex than my sentence indicates at first glance. A significant factor in all this is that the training and cultural deficiency doesn't necessarily manifest itself in a series of similar or repeated incidents and also that most pilots will not be affected by it. I suspect that every airline has pilots who whilst not being inadequate are not up to the skill levels of their counterparts. Most airlines are adept at weeding these out or conducting training in such a way that those individuals never become a threat. However, occasionally it seems an airline develops a culture which neglects training and develops a casual culture allowing a threat to develop. Examples where this can be identified would be Pan Am, Korean Airlines and China Airlines.
The incidents that happen do not occur in the same area but rather manifest themselves as lapses of judgment which a well trained pilot would not make. An airline in this situation may well have many competent and professional pilots but a laissez faire attitude. One thinks here of Pan Am's flight attendants unofficial blacklist of captains to avoid. Air France seem to have fallen into this situation - the past years incidents suggest that training is deficient and that culture is also an issue. The Air France safety audit report makes disturbing reading and what I note is that the terminology used to refer to the culture is similar in nature to the other airlines. It may be that we can consider this crew exceptionally unable but the A380 incident for example shows a disregard for basic standards at Air France that suggests otherwise.

[Machinbird]

Granted the A330 was not certified initially under US guidelines, but I thought it interesting to see how well it met US standards. As the title says, this is selected information by me.:)
§ 25.181 Dynamic stability.
(a) Any short period oscillation, not including combined lateral-directional oscillations, occurring between 1.13 VSRand maximum allowable speed appropriate to the configuration of the airplane must be heavily damped with the primary controls—
(1) Free; and
(2) In a fixed position.
It would seem that the A330 meets this standard very well...............until it gets into Alt2 Law. Then in is not heavily damped in roll.
§ 25.201 Stall demonstration has some interesting specifications:

(3) As soon as the airplane is stalled, recover by normal recovery techniques.
(d) The airplane is considered stalled when the behavior of the airplane gives the pilot a clear and distinctive indication of an acceptable nature that the airplane is stalled. Acceptable indications of a stall, occurring either individually or in combination, are—
(1) A nose-down pitch that cannot be readily arrested;
(2) Buffeting, of a magnitude and severity that is a strong and effective deterrent to further speed reduction; or
(3) The pitch control reaches the aft stop and no further increase in pitch attitude occurs when the control is held full aft for a short time before recovery is initiated.
AF447 seems to have demonstrated #1 and #3 a number of times during the prolonged stall.
#2 Buffeting apparently was not a strong and effective deterrent to the crew of AF447 and the Captain back in the cabin apparently did not recognize it either.

Is this knowledge well known among crews in general?:confused: Shouldn't it be? Suppose your stall warning system fails?
§ 25.207 Stall warning.

(b) The warning must be furnished either through the inherent aerodynamic qualities of the airplane or by a device that will give clearly distinguishable indications under expected conditions of flight. However, a visual stall warning device that requires the attention of the crew within the cockpit is not acceptable by itself. If a warning device is used, it must provide a warning in each of the airplane configurations prescribed in paragraph (a) of this section at the speed prescribed in paragraphs (c) and (d) of this section. It has been discussed a number of times in the AF447 threads, but the amazingly minimal response of the crew to the audible stall warning indicates to me that the concept of stall warning needs better human interface engineering in general.

For example, Colgan overrode his stick pusher! Maybe what is needed is not more force, but something that is harder to ignore. I really liked my pedal shaker in the Phantom, but the stick shaker is more applicable to the airline environment. Has anyone seen a better warning system? We do not seem to have quite the correct solution yet.
Here is your link to the reference: Electronic Code of Federal Regulations: (http://ecfr.gpoaccess.gov/cgi/t/text/text-idx?c=ecfr&rgn=div5&view=text&node=14:1.0.1.3.11&idno=14#14:1.0.1.3.11.2.159.33)

[RR_NDB]

Hi,

PJ2, (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-54.html#post7107998)


unloading the wing is used to prevent entering a stall, even when the aircraft is well below its normal level flight stall speed.
(http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-47.html#post7097358)

The concept is clear: Stall is a function of weight to be lifted (by the airfoil) and beyond a certain AOA there is a fast degradation. Unload means "help" the wing (to do the lift) "reducing the weight" using physics. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-47.html#post7097535)

[PJ2]

TTex600;

Yes, Expressway 31, thank you. We used the FDM program to monitor these approaches (de-identified) and then picked a gold standard, hand-flown approach, turned it into an animation and showed the troups, the idea being "the window". The numbers were as you say.

It's not a matter of being a whiz with the automation - that's easy stuff. It's a matter of thinking ahead two to six miles and keeping an eye on the guy you've been cleared the visual behind, at night, against NY city lights.

O.C., good thoughts - in my view you do capture some of the factors which I think are at the heart of this accident. I also agree with you on the Op-Safety Review - despite the caution not to view the document as connected in any way to the accident, the document means something...not everything, but it is a reasonable part of the mix.

[Clandestino]

Thanks, Old Carthusian, fully concur.

For example, Colgan overrode his stick pusher! Maybe what is needed is not more force, but something that is harder to ignore.Seemingly the stall warnings in both Colgan and AF447 cases were not ignored but the reactions to them were terminally wrong. We are not necessarily looking at perception but rather cognition malfunction.

Is this knowledge well known among crews in general?It is. trick is knowing it so well it could be remembered easily at 4 am.

[rudderrudderrat]

Hi Clandestino,

Whose FCTM is this?
A320%20321%20FCTM%20Flight%20Crew%20Training%20Manual.pdf (http://www.737ng.co.uk/A320%20321%20FCTM%20Flight%20Crew%20Training%20Manual.pdf) OP-020. P 15/16 FCTM 08 JUL 08

It appears to Airbus themselves who made the statement.

[Owain Glyndwr]

Selected Items, Part 25—airworthiness standards: Transport category airplanes
Granted the A330 was not certified initially under US guidelines, but I thought it interesting to see how well it met US standards. As the title says, this is selected information by me.http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/smile.gifThe A330 was certificated to JAR 25 Change 13, which should have been virtually identical to FAR 25 as written at that time (1980s?) I'm not sure how the FAR 25 you quote (2012?) relates to that earlier standard.

Anyway, the differences seem to be: [The emphasis and selections are mine]

JAR 25.181 Dynamic stability

contains another requirement....

(b) Any combined lateral-directional oscillations (Dutch roll) occurring between stalling speed and maximum allowable speed appropriate to the configuration of the aeroplane must be positively damped with controls free, and must be controllable with normal use of the primary controls without requiring exceptional pilot skill
Maybe marginal, but the A330 seems to meet this requirement also

JAR 25. 201 Stall demonstration has the following

(c) The following procedures must be used to show compliance with JAR 25.203
.....
(2) As soon as the aeroplane is stalled, recover by normal recovery techniquesJAR 25.205 Stall warning

(b) The warning may be furnished ......

.....

(c) ...... Stall warning must continue throughout the demonstration, until the angle of attack is reduced to approximately that at which stall warning is initiated.
Since the demonstration is supposed to be ended as soon as the aeroplane is stalled the designers would not have been expected to maintain the warning throughout 54 seconds of stall. With hindsight this requirement should be up for review.

PS. On checking, the current CS25 has removed the reference to the demonstration, which fits the bill:

Once initiated, stall warning must continue until the angle of attack is reduced to approximately that at which stall warning began.

[HazelNuts39]

#2 Buffeting apparently was not a strong and effective deterrent to the crew of AF447 and the Captain back in the cabin apparently did not recognize it either.The AF447 pilots probably mistook it for an indication of overspeed. At high altitude there isn't any appreciable difference between high-speed and low-speed buffet as both are caused by the same aerodynamic phenomenon (an oscillatory interaction between local shock waves and airflow separation).

The certification pilots probably considered the buffeting to be "a strong and effective deterrent to further speed reduction". BEA says as much in a note on page 54 of IR#2: "The stall manifests itself particularly through vibrations." (I take it that "vibrations" is franglais for "buffet").

In the video of a TV emission that was posted (http://www.pprune.org/tech-log/478681-af447-final-crew-conversation-thread-no-2-a-11.html#post7100798) a little while ago on the other thread, the AB Chief Test Pilot also seems to talk about "deterrent buffet" (at 34:50), but my understanding of spoken french is not good enough to understand what he says. Maybe one of our french contributors could help?

[Clandestino]

It appears to Airbus themselves who made the statement.

What you posted is First Choice Airways Flight training manual.

It is possible "Airbus themselves" made the statement, which was then copypasted into the FTM, yet it is equally possible someone at FCA was the source of the (mis)statement.

While I think Airbus pilots have to be trained to deal with unusual attitudes and approach to stall recovery in laws other than normal, this has no bearing on AF447 accident. Failure to recover did not result from lack of skills but inability to understand what is going on and consequently apply the correct procedure.

[rudderrudderrat]

Hi Clandestino,

Thanks for the decode of FCA. I only noticed the Airbus logo on the top of the page, however my airline's FCTM has exactly the same words so I conclude they are Airbus recommendations.

Failure to recover did not result from lack of skills but inability to understand what is going on and consequently apply the correct procedure. I agree.

On previous type conversion courses, we stalled the simulator to the nose drop and then recovered. When we had stopped using the manual pitch trim, the elevator feel was incredibly heavy, buffet could be mistaken for turbulence, controls became sloppy. It was impossible to prevent the nose dropping below the horizon using elevator control alone.

I've never stalled the A320 sim. I've flown at Alpha max & seen TOGA Lock, and heard "Stall Stall" briefly. etc.

The AF447 crew presumably had never been exposed to a full stall in the sim and hence they failed to recognise the symptoms.

[AlphaZuluRomeo]

Hi

@ rudderrudderrat & Clandestino, re:
The effectiveness of fly-by-wire architecture, and the existence of control laws, eliminate the need for upset recovery maneuvers to be trained on protected Airbus aircraft.
(my bold)
In Alternate (or Direct) law, an A320 (or other FBW Airbus) is no more a protected aircraft.

I admit this is a bit far fetched... and would prefer this sentence to be removed or, at last, better explained.

AFAIK, no such sentence is in the A330/A340 FCTMs :)

[Mr Optimistic]

Is there a consensus yet as to the altitude needed to effect a recovery by a typical crew (ie not the optimum with hindsight type of recovery), say from the dynamics AF447 had at 30k ft ? I believe the law they were in retained over-g protection but with no visible horizon or AoA meter would recovery rely on watching speed and pitching up to the stall warner when flying again ?

[Machinbird]

Quote:Machinbird
For example, Colgan overrode his stick pusher! Maybe what is needed is not more force, but something that is harder to ignore.

Seemingly the stall warnings in both Colgan and AF447 cases were not ignored but the reactions to them were terminally wrong. We are not necessarily looking at perception but rather cognition malfunction.
As I recall, the Colgan crew had just been discussing tail stall which may have primed them for their wrong reactions.

When you are the one flying and something grabs your stick, it is a natural human tendency to fight back. What we need is something that leads us to the correct action.(Even at 4:00 AM :})

Does anyone have any thoughts along that line? Perhaps a Sim routine combined with a standardized stall warning device that creates an almost Pavlovian response.:confused:

I say almost Pavlovian, because I remember the Abidjan A-310 accident where dropping the nose in the face of a false stall warning on takeoff caused an accident. Obviously the stall warning reaction has to be tempered with some common sense.

[Machinbird]

Is there a consensus yet as to the altitude needed to effect a recovery by a typical crew (ie not the optimum with hindsight type of recovery), say from the dynamics AF447 had at 30k ft ? I believe the law they were in retained over-g protection but with no visible horizon or AoA meter would recovery rely on watching speed and pitching up to the stall warner when flying again ? Before the hamster wheel started turning, it was becoming clear that the actual pull out of the resulting dive would create less altitude loss than the recovery from the stall.

It seems that the Basic Fighter Maneuvering (BFM) turn strategy was relevant to the pullout altitude loss.
From Hazelnuts last cut at the problem, it seems that the pullout could benefit still more from a little more power during the recovery to level flight. Maximum structural g was not being reached (probably due to induced drag at the higher AOA recovery).

[OK465]

It seems that the Basic Fighter Maneuvering (BFM) turn strategy was relevant to the pullout altitude loss.

...it seems that the pullout could benefit still more from a little more power during the recovery to level flight.

Mach:

I'd be real careful with advocating this.

You're not looking for a sustained QTT (quickest tightest turn) here.

Just something approaching a minimum radius maneuver. Turn rate is irrelevant.

[Machinbird]

Turn rate is irrelevant.

I'd be real careful with advocating this.
Thanks for the caution OK465.
In that turn rate is pointing your velocity vector away from the down direction, it isn't irrelevant.

I'm actually interested in identifying some generic simple to use guidelines that pilots could use. Obviously you will not want to be adding power above your cornering velocity, but the concept of adding power while having your nose pointed down to improve your pull out is a bit counter-intuitive to most pilots.

It looks like cornering velocity for a particular aircraft might be a good thing to know. No one plans on having to recover from a dive, so no one seems to be teaching it.

I am not advocating anything but a look to see if we could do a better job in performing dive recovery should the need arise. How/when/what to teach would raise a host of other issues that would probably be premature to discuss.

[RR_NDB]

Hi,

Machinbird:

standardized stall warning device that creates an almost Pavlovian response. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-55.html#post7108993)

For example:

A "vibrating (modulated, AM and FM)" pilots seat special cushion (integrated or stand alone as a retrofit).

:8

PS

Differentiating CLEARLY low speed and overmach by "sophisticated (and reliable) "multiple input parameters data processing".

Something like "seat of the pants" Mk II :}

[Mr Optimistic]

Thanks. I recall Hazelnut's post and the relatively limited loss of altitude (well to my non-pilot eyes) anyway. A really interesting recent post mentioned 20k feet(thanks Dozy great post !) perhaps based on sim. Figured there must have been other attempts so wondered what the consensus was and what the best strategy is. knowing how much nu can be attempted when time is really pressing doesn't look that obvious when in an unusual attitude.

[HazelNuts39]

I'm actually interested in identifying some generic simple to use guidelines that pilots could use. It depends. When you have plenty of altitude you would play it safe to avoid secondary stalls.

When the ground is approaching fast, my strategy as a non-pilot engineer would be (I've been shot down earlier mentioning this): Push the sidestick fully forward until the stall warning stops. Then immediately SS to neutral, slowly start pulling, carefully increasing the pull force until stall warning is encountered, then immediately SS to neutral until stall warning stops, then start pulling again, etc.

For the reasons you mentioned I wouldn't add power.

P.S.
Depending on how fast autotrim moves the THS, manual trim should be considered.

[DozyWannabe]

Depending on how fast autotrim moves the THS, manual trim should be considered.

In the A320 sim it was actually relatively speedy. You could probably physically move the wheel faster manually, but I suspect the servos moving the THS would take roughly the same amount of time in any case.

[Mr Optimistic]

Hazelnut, will potential recovery options be part of the accident investigation and reported eventually ?

[HazelNuts39]

will potential recovery options be part of the accident investigation and reported eventually ?No, I don't expect that will be the case. We are engaging in an academic discussion that is only remotely relevant to AF447 whose pilots apparently never realized they were in a stalled situation.

The first priority is and should remain staying out of a stall. In that spirit BEA has made a number of recommendations, for example "that EASA and the FAA evaluate the relevance of requiring the presence of an angle of attack indicator directly accessible to pilots on board airplanes."

[OK465]

To further this academic discussion I would venture this in general...

(edit: in response to the use of the esoteric term 'corner velocity' :))

IIRC 'Va' for a 330 varies from around 320k at around 28,000 down to around 260 at sea level. This encompasses quite a wide range of TAS, all 'limited' to 2.5 g.

From a practical standpoint, I'm not sure how likely it is that anyone who put themselves, or was put into a dive situation would be able to deftly modulate thrust to maintain a speed (let alone mental concentration on it), whose value was varying nearly as rapidly as the requirement to modulate the T-levers, on the way down.

Dive recoveries have been taught in every heavy I checked out in. The recovery procedures are recommended by the manufacturer and worked fine in all the aircraft I'm familiar with.

In this same vein, as CONF says, the manufacturer established stall procedures that involve extending the slats (Flaps 1) below 20,000 if clean. Were one to have a chance to practice this, one might see the decidedly positive benefits of following the manufacturer's guidance.

One first has to recognize that the guidance is applicable.

[DozyWannabe]

IIRC 'Va' for a 330 varies from around 320k at around 28,000 down to around 260 at sea level. This encompasses quite a wide range of TAS, all 'limited' to 2.5 g.

That G limitation only applies in Normal Law however. Once you're in Alternate, roll is direct and the calculations can no longer be applied to limit the maneouvre.

@HN39

While you're right that preventing a stall in the first place must remain a priority, I believe that Airbus and Boeing collaborated on providing proper stall recovery training for airliner crews. What bothers me about the AoA indicator is that while it would certainly be a help in a situation where the pilot knows how to use it, we're dealing with a situation here where at least one pilot was unable to use the basic panel to diagnose and recover from a stall.

[NeoFit]

@ Mr Optimistic

Is there a consensus yet as to the altitude needed to effect a recovery by a typical crew ...

I am very glad to see how many answers you have had.

A Tech & physical view is my first answer but, instead of a fastidious (and probably wrong explanation) I have better to give you a short report (http://www.bea.aero/docspa/1994/yr-a940924a/htm/yr-a940924a.html):


...in good weather conditions, the aircraft suddenly started to climb adopting a steep pitch attitude and stalled. The crew managed to recover control of the aircraft and came round to land.
...
The throttle levers were then quickly brought back to the idle position. At the same time, the trimmable horizontal stabilizer started to move in a nose-up direction.

As the aircraft descended through 1,700 feet, at 10 h 43 m 22 s, with a speed of about 195 knots, the Captain asked for flap extension to 20°. The VFE, the speed limit authorized for this new configuration, is 195 knots. When the flap control was set to 20°, the thrust levers advanced and engine thrust increased.
The flight crew countered the nose-up effect resulting from the increase in thrust by using the pitch controls, with the auto-throttle (ATHR) remaining in automatic mode.
The throttle levers were then quickly brought back to the idle position. At the same time, the trimmable horizontal stabilizer started to move in a nose-up direction.
The nose up effect that resulted was countered by the flight crew through gradual nose-down action on the elevators. When the trimmable horizontal stabilizer reached its maximum nose-up value and the elevators also reached their maximum nose down value, the throttle levers, according to the FDR readout, moved rapidly to their stops.
In a few seconds, the flight path started to rise and the pitch attitude went to 60°. Witnesses saw the aircraft climb. It banked sharply to the left and the right and stalled before adopting a strongly negative pitch attitude ( .33 degrees) towards the ground.
The maximum altitude reached was 4,100 feet, while a minimum indicated speed of 35 knots was recorded. The stall and ground proximity warnings sounded during the descent.
The flight crew managed to regain control of the aircraft, with the lowest point being around a height of 800 feet, that is 240 meters from the ground.
In a few seconds, the flight path started to rise and the pitch attitude went to 60°. Witnesses saw the aircraft climb. It banked sharply to the left and the right and stalled before adopting a strongly negative pitch attitude ( .33 degrees) towards the ground. The maximum altitude reached was 4,100 feet, while a minimum indicated speed of 35 knots was recorded. The stall and ground proximity warnings sounded during the descent. The flight crew managed to regain control of the aircraft, with the lowest point being around a height of 800 feet, that is 240 meters from the ground.

4,100 minus 800 equal 3,300 ft stall recovery
(of course, air density is more important near the ground)

[OK465]

Doze:

That G limitation only applies in Normal Law however.

The 2.5 g limit applies in all laws, including Boyle's, Charles's, and Burke's (Gene Barry) law and the law of supply and demand. :)

In one particular law you can actually exceed it. Starts with a 'D'. :}

[DozyWannabe]

You know what I meant - the aircraft's systems will no longer prevent a pilot from exceeding it outside of Normal Law. ;)

[Mr Optimistic]

Neofit , thanks. some ride.

[Mr Optimistic]

Still with significant nd and thrust at idle ? Think 20k ft was intimated in an earlier post which is what caught my eye. With a startled crew in poor conditions one minute isn't very long to interpret and initiate a course of action.

Post disappeared.

[DozyWannabe]

@PJ2 - not sure where your post's gone, but...

Going back to my notes, it took about 16-18,000ft to recover the way we did it. My writeups are on the last page of thread 6 and the first page of this thread.

We did not reenable ADR or FPV during the recovery, so I don't have the numbers available that you do. This was purely seat-of-the-pants stuff, applying ND until the THS returned more-or-less to the centre and holding the dive to gain speed, correcting the adverse roll with rudder inputs and gently pulling out of the dive once we figured we were going fast enough to be flying again.

Our experiment was purely qualitative - proving that the aircraft could be recovered if the situation was correctly diagnosed in time and that the aircraft's systems would in no way impede the recovery.

[PJ2]

Re altitude loss and other data, Dozy, what was your experience in the sim? Our sim exercises usually took about 20,000+ ft. for recovery.

Dozy, edited/re-posted for clarity.

With weights, CG, SAT mirroring AF447 and a bit of turbulence, following loss of airspeed (all 3 ADRs out), the sim was pitched up at FL350 and held in the climb until stalled, (THS reached 13.6deg). Shortly after the stall we returned the ADRs for use during the balance of the exercise, (to see the FPV during the stall).

Post-apogee (approx FL360), full forward stick was applied and held.

At FL330 the pitch was 8deg ND.

At FL310 the AoA (using FPV) was approx 40deg and the VSI was 18000fpm +. Pitch was about 14deg ND which was all the pitch that could be obtained.

Pitch slowly reduced to about 10degND still with full forward stick. As it was held the THS unwound and returned to normal settings.

We could watch the AoA reducing as the FPV slowly climbed "up" the PFD from past the red ND warning arrows below 30deg pitch marks.

Thirty seconds after the first Stall Warning passing through FL270 the AoA had reduced to 30deg, descent rate was 16000fpm.

Ten seconds later at FL255 the AoA was 12deg, CAS was 250kts, VSI was 7400fpm.

At FL245 the stall warning stopped 40 seconds after it began, the AoA was 10degND, M0.658, VSI 7000fpm down, CAS 278kts.

From an AoA of 40deg to 10deg took 24 seconds and about 6000ft. This exercise took about 22000ft; some were less.

Overspeed was never a problem nor was a secondary stall if one was gentle, (took about another 6000ft IIRC)

[Mr Optimistic]

Sorry to butt in but can i ask how you judged how much pitch up to apply?
:rolleyes: Hmm post back again...

The unfolding sequence over 2 years has given the profession plenty of time to consider how one would have dealt with this situation so in a sense you must have figured out what you hope you would have done. if not 'primed' for action allow a minute to stabilize the brain and that must add another 10k ft at the rod experienced. i realise the issue was how not to get to that situation in the first place but understanding the option space does inform any potential mitigations .

[PJ2]

Mr. Optimistic;

Normal cruise pitch attitude is between 2.3 and 3deg depending mostly upon weight. A pitch up to 5deg pitch attitude (+2.5deg) results in about an 800 to 1500fpm climb and a gradual loss of energy if held long enough. The UAS QRH checklist and the FCTM cautions strongly against holding such pitch attitudes for long and advises to get the QRH out quickly and set pitch and power. The FCTM also states that the Memory Items are not to be done if the immediate safety of the aircraft is not impacted.

For the purposes of the exercise there was no "judging" of how much to pitch up. We pitched up high enough to stall the aircraft. Fifteen degrees would do it, sometimes we were higher.

The overriding impression of these sessions was how quickly things occurred and how fast was the altitude loss.

[Mr Optimistic]

Sorry Pj2, cross posted. Thank you.

[PJ2]

Mr. Optimistic;

Does "option space" refer to the time available to assess a situation, decide upon a course of action, assess the response and secure the airplane? I hadn't heard of the term but likely it applies!

The danger of subtlely being taken in by hindsight bias is always there even when one really tries to avoid it; - we simply know what happened beforehand. The best that can be hoped for in these practical experiments is some sense of possible recovery time. "What would one do?" is not part of the exercise because we already really, really know what we would do!, and we weren't there... ;-)

But two questions remain unanswered: "Why the initial pitch-up?", and "Why wasn't the stall warning heeded post-apogee?". We've all seen various explanations as to why and are waiting to see how the BEA HF people come to terms with these two inexplicable matters and how such compares with our own thoughts and suggestions.

[Mr Optimistic]

Yep that is what i had in mind about option space, the time and actions available to the people required to act. I understand that the lhs did the wrong thing, and the rhs should have spoken and acted, but the safety of the industry cannot rely on people always being on top form or of analysing to the degree bystanders like us can.
If you are going to design a clever aircraft, need to help the crew in situations like t
his.

[DozyWannabe]

Mr. Optimistic:

Other way round - the PF was in the RHS on this occasion. I'm not fully understanding your point either - the aircraft's systems told them that they were approaching stall, and continued to warn them as they went into the stall regime itself. It warned them that they were in Alternate Law 2 without protections. The ADI told them that they were too nose-high, the speed indications (when they came back) indicated they were too slow and the altimeter was unwinding at an alarming rate.

How much more help can you expect the aircraft to give?

[gums]

Salute!

If I recall, we determined that the jet was not in a "deep stall" such as the Viper had or the T-tail jets. In other words, the cee gee and center of aero pressure was much like most airplanes.

So I have a hard time thinking it would take more than 10,000 feet to recover.

Granted, I was used to flying at the limits and sometimes over the limits, but I can forgive the basic heavy pilots in that regard.

As PJ has asserted, being smooth and getting the nose back up after reducing the AoA is the key to minimum altitude loss. You can't do this as a "mechanic". You have to have "touch" and be in "touch" with the jet ( I can just see Doze snorting, heh heh). I don't see any way to do this in a sim.

A reliable AoA indication seems a great thing to help you get back to acceptable flight conditions. The FBW implementation on the 'bus appears to have a flaw with regards to AoA. In other words, you can override the "system" and get into the situation that AF447 did. I don't get it. I lost one friend over the desert that buried the nose too far on an attack pass and the gee/AoA limiter gave him the best aero possible but he was just too low/steep. Sierra happens.

I also speak from personal experience in straight wing jets and bent wing ones and deltas. The mach buffet feels different than the stall buffet. One is more of a buzz and the other is more of a "shake". Then there's aileron reversal if it's a mach problem as others have described here about the shock waves.

Go read about Yeager's experience and his use of the horizontal stab. Once supersonic his elevators didn't work, so he manually cranked the stab. Seems an engineer warned him about that possibility( I heard the story directly from Yeager one day in 1979 at the 16th Sqd, First in the F-16) So all the USAF jets after that had the solid horizontal stab that worked for elevator and kept working in the transonic/supersonic regime.

I have to admit that this is a sad story, and I pray that I fly as a SLF with a crew that does more than drag us from DFW to DEN without flying a Citabria every now and then.

Gums

[Mr Optimistic]

I guess like me that you are an engineer. yes we all discuss the evidence available but look at the outcome. The 'system' failed. Sorry about the lhs/rhs issues, it has been discussed so often i forget where the pendulum stopped.

How do you train aoa instrumentation if it is never to be used unless in upset!

The warnings they had didn't tally with the situation they thought they were in so the significance of the warnings was lost. The stall warnings in their minds were erroneous and had to be filtered out.

[PJ2]

Re, "How do you train aoa instrumentation if it is never to be used unless in upset!"

Precisely.

There is far more to this suggestion than just installing the guage.

[HazelNuts39]

The mach buffet feels different than the stall buffet. One is more of a buzz and the other is more of a "shake".If you are responding to my post #1084, I suspect we are talking about different things. The airplane encountered "low speed" buffet at 02:10:53 at Mach 0.68. I believe that buffet was caused by "an oscillatory interaction between local shock waves and airflow separation", rather than the stall buffet as you describe it..

[mm43]

I believe that buffet was caused by "an oscillatory interaction between local shock waves and airflow separation" Or, a combination of Center of Lift [max] and roll angle causing the lift separation and 5 second pitching, which persisted for about 15 seconds before pitch and AoA parted company.

[Linktrained]

PJ2 Sim Stalls #1107
I assume that these were reproducing 447's performance using TOGA. A lower power ( Flight Idle ?) might have helped in the earlier stages.
How in the limited time (which WE now know was very limited) could PF have explained what he was having to do, to PNF. There was no further Drill to be actioned.
Of course he should not have stalled... But having got there, someone had to try to get out of it.
The Captain might have come to the Flight Deck without being called, due to the change of apparent gravity with the change of pitch of the aircraft - not I think, noticeable on a simulator.
( None of my Training or being checked was done on a simulator - we just had to use the real thing, with four fans and a Flight Engineer.)
LT

[OK465]

How do you train aoa instrumentation if it is never to be used unless in upset!

With all respect, how do you train someone to use an oil pressure gage?

When necessary, you assess whether or not it is in the marked proper range and note whether it is increasing, decreasing or steady. You're trained.

What you do then is up to you.

[DozyWannabe]

You have to have "touch" and be in "touch" with the jet ( I can just see Doze snorting, heh heh). I don't see any way to do this in a sim.

Why do you say that? I don't scoff at the idea that pilots should have a feel for the aircraft in the slightest, I just don't agree with the idea that the Airbus FBW control design makes that impossible. In fact all jetliners have had artificial feel of some sort from the DH Comet onwards.

In other words, you can override the "system" and get into the situation that AF447 did.

No-one "overrode" anything - the system dropped into a failure mode without protections because the airspeed sensors temporarily failed, and the evidence points strongly towards departure from controlled flight as being pilot-induced due to escalating misreading of the situation.

How in the limited time (which WE now know was very limited) could PF have explained what he was having to do, to PNF.

The point is that from the moment the UAS indication appeared, the PF did not in fact immediately have to do anything. They were at cruise, their speed was OK, they had a little bump from turbulence, but that wasn't going to send them into a significant dive. The correct thing to do according to CRM would have been to let the aircraft ride itself out of the turbulence and then discuss what to do with each other while monitoring the aircraft's response. Instead, the PF seems to have instinctively yanked the aircraft into a climb and developed a near death-grip on the sidestick until they hit the water, without telling anyone that he was doing so.

[Linktrained]

I agree.
But I was trying to follow the Sim Stall and thinking " What if, at apogee PF HAD suddenly realised ( At Last ! ) we are stalling...Merde.... And PJ2's experience on the Sim might be..."
I should have been clearer.


OK465
The oil pressure instrument on my car is backed up by an oil pressure warning light which comes on if the usual pressure is missing.
Similarly the THS could have a warning when it exceeded certain a figure. This might have made it easier for any of the three pilots to notice and wonder why.

LT.

[gums]

Salute!

No problem, Doze. I need to express myself more clearly and referring to a post you made a year and a half ago.

- By "feel" and "touch" I am talking about vibrations, noises, thumps and bumps, and even the airframe shaking so badly that it's hard to read the instruments. Ask Retired about flying the F-4 at high AoA.

I am not talking about "artificial" feel in the controls. You know, springs, dampers, bobweights, etc.

Really good pilots will KNOW when something is not right. They might detect a different noise from the motors. They might feel a vibration that wasn't there a second ago. In other words, they are tuned to the aircraft, regardless of the flight control implementation. (On private, ask me about one day I "felt" something and saved the jet)

- The disregard of AoA by the 'bus system in some modes bugs me. I fully appreciate that at very low dynamic pressures, the AoA sensors are close to useless. But seems to me that the 'bus ignores the AoA when the "system" determines that airspeed is invalid, or speed too low. Sad, because the wings certainly know if they are stalled, so the sensors should still be valuable if the crew is in doubt.

So I was not clear about "override". You are correct. The system allowed the pilot to exceed the stall AoA and the pilot may have still thought that he couldn't stall the jet.

Interestingly, our first operational FBW jet ( over 4,000 built) even "protected" the pilot from overriding the system limits until AoA was over 30 degrees. Then we could manually control the horizontal tail and "rock" outta the deep stall by holding a switch with one hand and rocking with the other hand on the side stick.

++++++++++++

I fully agree with Doze and others ( last paragraph of his last post). Why the pilot pulled back and then held the stick back for so long is a mystery. My opinion is he was more concerned about overspeed initially, but after that......

Gums

[Lyman]

HazelNuts39

Howdy. Is there any possibility at all that RHS was following in his scan the V/S (SELECT) at 5000fpm descent at a/p loss, confused it with the Nose Down he saw at takeover, and imprinted it in his initial S/A? Therefore the NU?

Perhaps Farley can explain how difficult it is to hover the Harrier. A friend tells me its like doing jumping jacks on a bowling ball. I tend to use that thought when I envision a/p loss on 447. It was difficult.

Flight is a continuum, one takes a break from the flow at great risk. I try not to be too dismissive of how difficult it turned out to be.

As to the DFDR/CVR contents? To consider any part of it to be "off limits" to anyone with appropriate expertise is to deny the concept of a Democracy.

Especially at Court. The Plaintiff is denied access, whilst the defendant has a copy on the coffee table? Please, some perspective.

Hi Dozy.

[DozyWannabe]

Really good pilots will KNOW when something is not right. They might detect a different noise from the motors. They might feel a vibration that wasn't there a second ago. In other words, they are tuned to the aircraft, regardless of the flight control implementation.

Yeah, but you can't design airliner safety concepts around the assumption that the pilots will be "really good". Average ability must be assumed, alongside failsafes to catch pilots having a "below average" moment.

But seems to me that the 'bus ignores the AoA when the "system" determines that airspeed is invalid, or speed too low.

It doesn't "ignore" AoA during UAS, otherwise Stall Warning wouldn't work. It can't use AoA data for protections during UAS because the redundancy calculations are defeated. Low airspeed is a tough one. As was the case a while back, I'm prepared to bet that similar cut-off points are used on most modern airliners, not just Airbii.

[HazelNuts39]

mm43;
The "high frequency component of an amplitude increasing to until about 0.1 g peak-to-peak,"(...) "vibrations that might correspond to buffeting" that BEA speaks about on pages 43-44 have a period of about 0.5 seconds.

[Lyman]

Doze.

@DozyWannabe
Yeah, but you can't design airliner safety concepts around the assumption that the pilots will be "really good". Average ability must be assumed, alongside failsafes to catch pilots having a "below average" moment.

That is precisely the beginning of the Hamster Wheel that will perpetuate these wrecks. The Spiral of "Better Automation", "Diminished Ability".

Do you see the connection?

[DozyWannabe]

Howdy. Is there any possibility at all that RHS was following in his scan the V/S (SELECT) at 5000fpm descent at a/p loss, confused it with the Nose Down he saw at takeover, and imprinted it in his initial S/A? Therefore the NU?

Not really, because you just cross-check with the altimeter, and as long as it's not unwinding drastically, you rationalise the V/S display as a turbulence-related blip.

Perhaps Farley can explain how difficult it is to hover the Harrier. A friend tells me its like doing jumping jacks on a bowling ball. I tend to use that thought when I envision a/p loss on 447. It was difficult.

Difficult, yes - but as difficult as hovering the Harrier? I don't think so. The Harrier is notoriously tricky to control in hover as it's not very stable and you're only about 50ft or so off the ground, whereas the A330 is *extremely* stable, and in this case was handed over at cruise altitude, with a reasonable speed setting. The AF447 PF could in all likelihood have kept his hands off the stick and nothing untoward would have happened.

As to the DFDR/CVR contents? To consider any part of it to be "off limits" to anyone with appropriate expertise is to deny the concept of a Democracy.

This isn't about denying content to those with appropriate expertise, it's about air accident investigation *needing* to be a special case to prevent possible further accidents between the time of the accident and the court case. I was accused of naivete on the other thread for saying that Airbus had to be given the FDR data for technical reasons and not for the benefit of their legal team. I'm sure their legal team have seen some of it, but until the final report is released and the trial begins, I don't think they'll have looked into anything the BEA consider non-relevant (and remember, the distinction of relevance is made by the BEA, *not* Airbus).

It's also interesting that you use "Plaintiff" and "Defendant" in that manner - strictly speaking, the crews' families, Airbus and AF are what Anglo law would consider co-defendants. With the evidence as it's laid out in the 3rd Interim Report, it seems that all three are likely to come in for a degree of censure - mishandling and poor CRM for the flight crew, failure to expedite replacement of the Thales pitot tubes for AF and Airbus, and poor management and training practices on the part of AF.

I've said my piece on automation elsewhere. Automation is a tool, nothing more - if that tool is being used to skimp on training and development of handflying, then it is the fault of those using automation for that end, bot automation itself.

[jcjeant]

Hi,

Gums
Really good pilots will KNOW when something is not right. They might detect a different noise from the motors. They might feel a vibration that wasn't there a second ago. In other words, they are tuned to the aircraft, regardless of the flight control implementationIndeed ... A good example is the pilot Hans-Joachim Marseille "The Star Of Africa"
He flew with gym shoes instead regulatory boots
He has said to Willy Messerschmitt:
I made ​​one with my plane .. I must do no more to focus on instruments .. all my concentration may well be on my target

It's also interesting that you use "Plaintiff" and "Defendant" in that manner - strictly speaking, the crews' families, Airbus and AF are what Anglo law would consider co-defendants
The court of justice in charge of the case is located in Françe and apply french laws .. not Anglo laws

[Machinbird]

Re, "How do you train aoa instrumentation if it is never to be used unless in upset!"

Precisely.

There is far more to this suggestion than just installing the guage.
The natural place to be using AOA is on approach. That is where to begin training to use it. Once the indication is worked into your scan, you can begin to use it elsewhere in the flight envelope.

[DozyWannabe]

The court of justice in charge of the case is located in Françe and apply french laws .. not Anglo laws

I'm well aware of that, I was just making the point that Lyman's insinuation that the non-release of raw FDR data to the public and/or the crew/family lawyers was equivalent to the prosecution withholding evidence from the defence in a criminal trial was false.

[RR_NDB]

Hi,

PJ2:

There is far more to this suggestion than just installing the guage.


Due it's importance (to the wing "function") i think it "self teaches". And (perhaps) should be used in training.

I admit, it collides with Airbus SAS philosophy of "stall prevention" through "protections".

But (may play an important role) can be vital in (rare) extreme cases.

Mac

(Controversial, :confused: )

[RR_NDB]

Hi,

Bear:

The Spiral of "Better Automation", "Diminished Ability". (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-57.html#post7110867)

Important point!

Mac

PS

Is training capable to "prepare" when applying this approach (to the limits)?

[Lyman]

@DW

QUOTE
I'm well aware of that, I was just making the point that Lyman's insinuation that the non-release of raw FDR data to the public and/or the crew/family lawyers was equivalent to the prosecution withholding evidence from the defence in a criminal trial was false.UNQUOTE

[B]NO[B] In juris civile, Plaintiff and Defendant are no less adversarial. Perhaps more so. "NON RELEASE". That means what, like "Withheld?"

Witholding evidence can be charged in itself as a felony. Even a jurist can be recalled for indiscriminate actions.

Let me ask a non threatening question, Doze. Say you are in Upset, and you see your Pitch as 7 degrees. You have no idea your AoA, it is at 40 degrees. You ok with that? Any problem? You cannot suss anything of help, and your g is around "1". Still ok? Right, AoA is ok in the computer quiver, but not in the Pilots'?

You seem to have forgotten that at the time of this UAS, Pilots were told to Disregard the STALL WARN. Also, the a/p was not yet accused of "Uncommanded climb".


Also @DOZEWANNABE:
(FBW) "It doesn't "ignore" AoA during UAS, otherwise Stall Warning wouldn't work."

Can you say that the SW did work? Evidently not, and for the same reason the pilots could not Suss the STALL. The AoA was not available. Could one explanation be that the pilots did not even acknowledge the WARNING because they all tacitly took it to be bogus? You have a better question than "How could they ignore the STALL WARN?"

[Turbine D]

Lyman,

Read my post in the other AF447 thread regarding release of information...

TD

[HazelNuts39]

HazelNuts39: Howdy. Is there any possibility at all that RHS was following in his scan the V/S (SELECT) at 5000fpm descent at a/p loss, confused it with the Nose Down he saw at takeover, and imprinted it in his initial S/A? Therefore the NU? Maybe you should ask a pilot which I'm not. The trace of V/S Selected on page 111 prior to A/P disconnect shows 0 fpm with regular 'spikes' to -5000 fpm (the much debated 'zipper'). After A/P disconnect the spikes disappear and the trace shows 0 fpm for about 15 seconds and then starts jumping up and down to +5000, -5000, +5000, 0, +11000 where it remains for about 20 seconds.

The FD was not available immediately after AP disconnect, but briefly reappeared about 10 seconds later.

I'm afraid I don't understand your question.

[Lyman]

Thanks TD I have. Both you and Dozy are absolutely correct. Save one: At its completion, the Final Report is not given to the Judge, it is provided EASA, isn't it? Thence to parties deemed important by the authority. BEA has no power, they are "Party to" ?

HazelNuts39. Sorry, I just wanted to know your opinion re: its availibility to PF, not his reaction to it. Your response is perfect. From it, I conclude the zipper may have held his attention as to V/S and influenced his inputs, perhaps well into the "climb". My opinion only.

Further, the display of V/S select, appearing (post a/p loss) "erratic", may actually have been accurate, but further confused the pilot as to his "climb".

I only suggest that the "zipper" can't just be discarded as irrelevant.....
Had he not judged his V/S accurately, how was he to suss his actual ROA?

[RR_NDB]

Hi,

PJ2:

The overriding impression of these sessions was how quickly things occurred and how fast was the altitude loss. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-56.html#post7109949)

Therefore, a "practically impossible" stall recovery, given the "environment" they were in and their "possibilities" to tackle the task.

"Why the initial pitch-up?", and "Why wasn't the stall warning heeded post-apogee?"

You are assuming this will be only "understood / explained" by HF study results?

Mac

[gums]

Salute!

@ Doze:

Redundancy is required, no problem with that. OTOH, graceful degradation is also required. So why throw out the one sensor that actually indicates the ability of the wing to provide lift or lack thereof? By the time the AoA sensors are invalid I can guarantee that the plane's wings aren't providing useable lift doofers.

'bird has it right. Start off using the AoA for approaches, as it inherently compensates for weight and such to provide the best aid in the cockpit.

Being a USAF jock, we used a baseline speed and added knots for fuel weight and such. The U.S. Nasal Radiators used the AoA indexer lights and such. So I was impressed once flying the A-7D, which was a Navy design which the USAF adopted and improved. Gotta tellya that an approach was much safer and easier than what I had been used to.

Sure, a good crew will compare the speed based upon weight with the AoA, but once close, AoA was the primary aid for an approach.

Once gear up, our Viper AoA "bracket" went away. Due to our operational requirements and design of the FBW system, pulling full back gave you the best AoA for the gee you were commanding. By the time you reached the max AoA you were in one Gee flight ( we hit 9 gees at 15 degrees).

For the 'bus, I can see an AoA indication in cruise configuration that is small and basically unimportant. But when the system starts to revert to backup modes, then the thing might be the best friend you have, especially for a pitot-static problem. And then there's the flight path marker (FPM) produced by the inertial system, which show exactly where the jet is going without regard for any aerodynamic sensors.

I can not imagine seeing the FPM at the bottom of the indicator or HUD and not thinking about being fully stalled while holding back stick, can you?

Gums

[DozyWannabe]

From it, I conclude the zipper may have held his attention as to V/S and influenced his inputs, perhaps well into the "climb". My opinion only.
...
I only suggest that that the "zipper" can't just be discarded as irrelevant.....
Had he not judged his V/S accurately, how was he to suss his actual ROA?

Having now spent time in an A320 simulator, I can tell you that in MANAGED mode, the corrections applied by the autopilot are displayed in that window. "-5000" appears, but only for a fraction of a second as the slight corrections are applied. I'd be very surprised if that interrupted his instrument scan (of which the FCU display should only be an occasional part) and led him to conclude they were descending for two main reasons.

1. They are a regular occurrence in MANAGED mode, and lots of corrections are expected in this scenario. He may have been relatively low-hours for an airline pilot, but this is one scenario where spending a lot of time with the automatics in would have led him to discount that as anything to worry about.

2. He did not start climbing until the automatics disengaged, at which point (as you'd expect), the "zipper" stops.

[RR_NDB]

Hi,

Considering we fly near the "corner" i ask:

Why not to implement (ASAP) a "resource" to indicate Stall type.

The system (IMHO) is able to process and inform reliably on low speed or overmach stalls.

Thiells 727 man machine interface mislead 3 crew. (Redundant misleading of Climb and Alt)

Current Systems, it seems can indicate precisely Stall type:

The mach buffet feels different than the stall buffet. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-56.html#post7110003)


Why the pilots (sometimes near the corner) should only guess?

Mac

This "guessing" (wrong) probably was the most important factor in AF447 loss.

[Lyman]

From PJ2 Describing SIM, 447 profile, STALL

"The overriding impression of these sessions was how quickly things occurred and how fast was the altitude loss."

If RHS could have been debriefed, would he have stated something similar?

Substituting "GAIN" (altitude) for "Loss?"

DOZY, I refer to SELECT, not MANAGED. I have a further question, but probably for an engineer. What is the damping for a 5k fpm ascent/descent? Is the bellcrank, A/Rod protected from sequential deflections, followed by a withdrawal of input? Are the inputs density compensated? I am sure they are. Having seen the deflections of a preflight Stabilator check on F-16, they are emphatic, to say the least.

Another thought. Could the .5 second cycle .1g amplitude "vibration" have been a stammering elevator select? OR THS stutter? Just a mechanical question, nothing more.

[DozyWannabe]

Lyman - read what I'm saying. In MANAGED mode, the corrections applied by the autoflight system are displayed in the SELECT window of the FCU panel. I want to make clear at this point that we're not talking about control inputs of a magnitude to effect 5000fpm descent rate immediately, simply that this is the value the software uses in certain configurations. Lower to the ground, you'll see -1500 appear in the window for a split second as corrections are applied, but you don't start descending at 1500fpm.

[Lyman]

Not surprised we are not communicating well. You haven't touched on my question. Perhaps let's just give this dialogue a rest. I'll just read. You're the Pilot rep.

[HazelNuts39]

Could the .5 second cycle .1g amplitude "vibration" have been a stammering elevator select? OR THS stutter? Just a mechanical question, nothing more. The buffet onset envelope is determined in flight test before the type is released into service. It is scheduled in the AFM, the FCOM, the QRH and in the FMEGC. Stall warning is set to occur at an AoA 1 degree less than buffet onset. The recorded 'signature' matches that previously obtained in flight tests. Why would the elevator begin to stammer and/or the THS stutter at the precise moment that the airplane enters a regime where buffet is known to occur?

[RR_NDB]

Hi,

Before the day the Systems becomes PERFECT and EVERY possible situation is simulated and TESTED, as a designer i would prefer to deliver planes with "small resources" (like AoA) capable to save POB.

gums:

But when the system starts to revert to backup modes, then the thing might be the best friend you have, especially for a pitot-static problem. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-57.html#post7111040)

:ok:




Mac

PS

Is seems something like the AoA indicator (proven in EXTREME conditions) lies in the "Tectonic Fault" dividing who:

1) Think a System (and the crew) is capable to work always as expected.

2) Known (battle hardened) that sometimes you must be able to do the "necessary" to survive.

Airbus SAS or Air France or authorities (or AF447 crew) made what was possible:

1) Maintained hundreds of A/C relying on OBSOLETE AS sensors

2) Didn't replace the WORSE obsolete sensors in time

3) Didn't anticipate possible PROBABLE ("recurrent") scenarios.

4) Didn't act with it's authority capabilities

5) Didn't even understand what "was happening" yet before entering WX :{

Perception "means" survivability:

Thousands of professionals IMO didn't act proactively.

And Murphy Law again put the toll for that collective failure (merely a probability result).

PS

Well, if AoA gauge has no place in the cockpit at least his data could be used to "inform" the stall type. :}

[mm43]

@HazelNuts39
"vibrations that might correspond to buffeting"I agree with that, and I should have written before the wine glass was empty! Possibly the point I was trying to make, was that the onset of the 5Hz 0.6g (p/p) at cockpit vibrations occurred as the 0.2Hz oscillation of the pitch attitude and AoA commenced. The amplitude of this "buffeting" has risen in tandem with the increasing amplitude of the 0.2Hz longitudinal pitch/AoA oscillation.

I wonder if PJ2 noted anything similar in the SIM? Possibly not, as that aspect of the airframe dynamics was then outside the known flight envelope.

[roulishollandais]

@PJ2
I'm not sure what you mean by "effective aircraftThis excellent 1997 book that Machinbird advised me, shows 51 references of "Effective aircraft" to explicite the definition: Aviation safety and pilot control: understanding and preventing unfavorable Pilot-Vehicle Interactions (http://books.google.com/books?id=w56Y9ayzq44C&pg=PT32&lpg=PT32&dq=%22Effective+aircraft%22+aviation+safety&source=bl&ots=qBmOBbaT1g&sig=Qv95W-oBXB2SNG9p65nTGbolxGw&hl=en&sa=X&ei=nUt3T6rZGMTF0QWHgcHJDQ&sqi=2&ved=0CDMQ6AEwAQ) isbn=0309056888...National Research Council (U.S.). Committee on the Effects of Aircraft-Pilot Coupling on Flight Safety (http://www.pprune.org/search?hl=en&biw=1272&bih=495&tbm=bks&q=inauthor:%22National+Research+Council+(U.S.).+Committee+on +the+Effects+of+Aircraft-Pilot+Coupling+on+Flight+Safety%22&sa=X&ei=nUt3T6rZGMTF0QWHgcHJDQ&sqi=2&ved=0CDQQ9Ag) - 1997 - Transportation - 208 pages
Implications for Design of the Effective Aircraft Dynamics Reduce time lags in the high-frequency effective aircraft dynamics. To reduce tendencies for ...

](no) unusual difficulties which require [...] (particular) skill or knowledgeThese characteristics are required for certification, but have nothing to do with system "controllability"
eminently, (éminemment), "observable and controllable"Observability and controllability don't have to be eminent, they just are mathematic.
Nor do the accident rates indicate a large difference between aircraft typesI assume you refer to A/B ? I have no royalties in any, nor in other manufacturer, nor in any insurance/aircraft financer. I am only involved in flying pleasure, pilot safety, and mathematic accuracy.
However, no aircraft, no design I know of is controllable or observable when it is taken into a full stall.False :
1. In any pilot school over the world, stall is teached to get private pilot. Would this manoeuver be uncontrollable and the international civil aviation allow that ?
2. AF447's A330 finally seems to stall like many others aircrafts stalled : you just have to push to stop the stall .Gums quoted F-102, Concorde, Viper having analog stalls.Yourself said it is the case also with B777 and A330 D-sim. Is that uncontrollable if the pilot learned that ?
3.The french pilot school where Air France teached his young pilots (perhaps Bonin), IAAG/EPAG uses for VFR and IR learning the Socata TB20 Trinidad. This aircraft stalls in just descending, unstalls just with pushing... Would this manoeuver be uncontrollable and the french civil aviation allow that if it was ?
4. All the aerobatic pilots, including fight pilots regularly stall and spin, controlling perfectly their flights.. We like to look them and we could not imagine one second their flight is not controllable. Are they ?　
AF447 was recoverable even after entry into the stall but it required that the stick be pushed fully forward and held there until the wing began flying again. That would take between 15,000 and 22,000ft (I've flown this in the sim many times). This is Machinbird's "unloading of the wing" to which he referred some pages back.
And Presto! we have some numbers and they show that (http://www.pprune.org/members/306485-machinbird)
If you do not mess it up you would do surprisingly well
The "on the threshold of stall warning" trajectory with specific excess thrust (T-D)/W = 0.15 results in level off after 15 s (was 18s) at FL76, Mach 0.51, 295 kCAS, az=2,24 g.
The final loss of height is not your 22000 FT, nor FL200, nor 15000 FT,
but 5000 (max) +(FL100-FL76)=7400 FT
Said another way, a B777 pulled up in the same manner and handled the same way as this aircraft was would also crash
Are the B, C, D, etc pilots hammered with "B, C, D, etc. can't stall" ?
sidestick vs control column
Both may be tried, used. if correct. The problem of observability, governabillity, man-machine interface is not which actuator, but how it is connected to the effective system. I have no royalties in any manufacturer, and I choose to put philosophy in library not in aircraft.
This is the part that is very definitely not complicated
However, it is more complicated when you can't interview a key person in the problem loop. When this happen, you have to make a list and go through the process
In the present system, 99.9% of flights work well with SOPs, CRM, appropriate use of automation (according to enlightened airline policies which permit hand-flying), but the loss of such skills is nevertheless no longer a blip but a trend.

That is very dangerous ! Every sixty seven hours you have a four minutes crash ! I choose to fly with "these guys [...] (who) knew how to stay alive, (many of them are here, thank you !) and not in your "managed airlines" and their effective aircraft.

roulishollandais

[mm43]

Originally posted by roulishollandais ...
That is very dangerous ! Every sixty seven hours you have a four minutes crash !
Now, now, now ...:=

PJ2 forgot to hold the "9" key down after the decimal point, but I doubt he would forget to push the stick forward.:ok:

[Lyman]

So what I'm looking to eliminate is the zipper as Alpha Prot, considering there was turbulence, UAS, and an actual SW for the Pilot flying. His input did not have time to actually change the AoA, so I'd be looking for some combination of local W/S to duff the AoA vanes. Since STALLWARN is available in ALTLAW2, it wouldn't make any difference whether in autoflight or manual.

just askin'

Whether the Law degrade was due imminent STALL (actual or AoA A/S sensed) or UAS, or computer malfunction, would it be important to know what ND value would obtain given autoflight's need to correct, WARN? Would it be 5000fpm/ND?

Because that would be almost like foul play, eh? "We're about to STALL, Your Aircraft"........."Oh, and LAW CHANGE"........"SeeYa".........

[thermostat]

MM43, thank you for providing this valuable link to the manual. Very interesting. I used to have the 320 version but no longer do.

[PJ2]

Hello Lyman;

In re the autopilot disconnect and, "It was difficult. " (post #1124 (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-57.html#post7110848))

No, it was not. And a disconnect is most certainly not like "hovering a Harrier".

An autopilot disconnect is a non-event.

If the airplane is descending/ascending slightly when the disconnect occurs, a gentle nudge on the stick (millimeters in terms of movement of the top of the stick) brings the airplane back within the expected requirement, which here is maintaining a cruise altitude.

A disconnect simply is unimportant to a pilot - and absolutely should be a non-event. One takes over, commands the airplane and makes it do what you want, returns it to what you were doing before the disconnect, and when available, re-engages the autoflight system. It's what we do.

The terms "Select" and "Managed" have specific meaning for Airbus pilots. They refer to an autoflight mode where control of the aircraft is either through the MCP (Mode Control Panel - autopilot panel) or the FMGC.

When the AP disconnects and there is a rate of climb or descent occuring, (including a small deviation such as occurred here), the vertical speed window will display VS (as opposed to other modes such as FPA or just dashes in "managed" mode). In other words, the display defaults to VS and will show what the rate was at disconnection but isn't a "command"...it's just information at that stage.

In terms of response, no pilot is going to slavishly follow an FD command to "descend at 5000fpm" because a flight director says so. He is instead going to "look through" the temporary FD "command" and get his airplane gently back to cruise or whatever it was doing prior to the disconnect and get on with the drills/checklists. No big thing. Really.

Hello, Machinbird;

I do appreciate your comments and the fact that you present such a different view of 'the wing' than I have as a transport pilot - it is very engaging and well worth the discussion, thank you.

Re, "The natural place to be using AOA is on approach."

I like knowing and using AoA. I used to interrogate the ACMS Alpha Parameter functions on the ACARS units all the time and keep the AoA displayed down on the pedestal especially during heavy turbulence. So it can be done (though it isn't taught!), and likely easily with software and pixels (rather, raster scan!).

So I do not argue against an AoA presentation to the pilots...I like it and agree that training on it would be straightforward in terms of the mechanics of learning and applying the information. I am not personally arguing against it but I'm recognizing some of the impediments to making the change and the likelihood of such change in the face of certification regs, training/standards/checking and the like. Airline budgets are thinner than they ever were and "safety as a business case" is not an atypical approach, it is THE way it is done. Agree or not, cost and cost control are enormous factors in daily operations.

The one argument which I think may be convincing is in knowing that one was stalled but the counter-argument is, the blaring "Stall, Stall" and loud warning chirp didn't get their attention...would AoA have? What else does a pilot need to tell them their aircraft is stalled?

The industry has been largely successful without AoA so the argument that AoA would have "saved" AF447 vice other standard handling procedures (not stalling, pushing the nose down once stalled), is not convincing. It wouldn't have saved Colgan, it may have saved the Airborne Express DC8, it may have saved the FedEx ATR72 at Lubbock and (IIRC) a B727 where the pitot heat was turned off, but in the millions and millions of hours flown in all types, lack of AoA has not been an issue until AF447.

So the argument must demonstrate that an AoA indication would have made a difference to the outcome, for this crew and while it may have, we just don't know. The captain may have noticed from his vantage point and called out "push down!"...we don't know. I agree that one accident prevented makes such change highly desirable but again, what shall we argue next? I absolutely know through data that there are some things that can be done right now that would prevent CFIT and overrun accidents in the approach and landing phase but short of exorting the stabilized approach, very little is left up to technology and presented information.

But we could have an electronic "energy-status bar"...showing the potential landing distance, even on contaminated runways as a digital presentation relative to the aircraft's energy level on the approach.

Now that kind of presentation would be used thousands and thousands of times because that is the number of non-stabilized approaches that are occurring in airlines' FOQA data.

The argument for AoA is not "natural" but is nevertheless a good one...the more info the better, but it comes with much more than just putting in the guage and both manufacturers and airlines are going to look at the cost.

It is the way this business is done, much to my (and most others who work in it) frustrations.

My comment about "how quickly things unfold" is an observation on the amount of time it took to lose control. Time is always "relative to perception and familiarity". When a pilot is highly trained, highly experienced and very familiar with his/her airplane, the flow of even huge amounts of information and lots of events flow much more slowly, perceptually, because the mind anticipates much more effectively and easily, (depending upon other factors such as distraction, fatigue) than if one is relatively inexperienced in such circumstances.

In terms of this loss of control, while it unfolded literally over a period of about 40 seconds, oddly that is tons of time to do something, but it is not a lot of time for assessment, discussion, action.

This is a really important point: Most things that happen in an aircraft that cause trouble or an accident take a few seconds and one has a bit longer for response. This is what training (and training and training) is for - to reduce the time needed to assess accurately and respond. The response to the UAS was instantaneous without, in my view, good reason to do so and without the requisite crew coordination. One simply never does that in a transport aircraft, period, unless it is a TCAS, GPWS warning. The SOP is, first establish control of the flight path, (in other words, don't give up a stable flight path for one that is unpredictable, uncertain, unannounced and unplanned); once control is assurred, call for the drills and/or checklists, all the while keeping a very close eye on the aircraft.

By the time the aircraft had stalled, they had backed themselves into such a corner of confusion that there were no breadcrumbs to get back to stable flight because they completely lost situational awareness even though the nose was pointed up, the stick was full back and the altimeter was unwinding at about a thousand feet every three seconds.

PJ2

[mm43]

thermostat

Thanks for reminding me about the A330 Instructors Support Manual (http://www.scribd.com/doc/35605266/INSTRUCTOR-SUPPORT-Airbus-A330#page=1) link.

[RR_NDB]

Hi,

DozyWannabe:

the aircraft's systems told them that they were approaching stall

What type of Stall?

and continued to warn them as they went into the stall regime itself.

SW was presented several times erratically, not adequately processed by the System.

It warned them that they were in Alternate Law 2 without protections.

Ok

The ADI told them that they were too nose-high

Ok

the speed indications (when they came back) indicated they were too slow

Faulted at critical moments

and the altimeter was unwinding at an alarming rate

After FUBAR "threshold"


How much more help can you expect the aircraft to give? (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-56.html#post7110002)

IMO, the A/C (it's man machine interface) could (may be, should):

1) Provide an "immediate"* information (aural, whatever) to inform origin of trigger problem (actually a "frequent" and recurrent problem). It would have helped.

2) Protect* (limiting to REC MAX FL) from zoom climbing to the "corner" entering "the coffin" :} .It could have helped.

3) Provide* information on "dangerous trend" like AoA increasing at such rate. It could have (redundantly) alerted PM on PF "fatal" persistent NU.

4) Better Stall (characteristics) indication. Technically feasible. It could have "helped".

5) More elaborated Aural and Visual information to help "HF aspects" during "crisis". It could have increased their chances reducing confusion, stress, etc.

In this specific point we may mention: Antoine de Saint Exupéry's "It seems that perfection is reached not when there is nothing left to add, but when there is nothing left to take away"

Probably something that can be applied during extreme conditions in many designs.

Mac

PS

Man machine interface designers would have an opportunity after the results of the HF study of AF447.

(*) Cost: "near zero"

[RR_NDB]

Hi,

Just remembering:

The "unprotected" B747SP AVERAGED ROD of 30,000 during 20 seconds.

They experienced as much as 5g. And practically lost left horizontal stabilizer / actuator.

:eek:

Mac

[RR_NDB]

Hi,

gums:

I have to admit that this is a sad story, and I pray that I fly as a SLF with a crew that does more than drag us from DFW to DEN without flying a Citabria every now and then. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-56.html#post7110003)

Why not use your car? :} Just 800 sm.

Mac

PS

My longest leg (in a Winnebago) was 1000 miles. By car is a lot easier. :)

[roulishollandais]

Originally Posted by PJ2
In the present system, 99.9% of flights work well with SOPs, CRM, appropriate use of automation (according to enlightened airline policies which permit hand-flying), but the loss of such skills is nevertheless no longer a blip but a trend


That is very dangerous ! Every sixty seven hours you have a four minutes crash ! I choose to fly with "these guys [...] (who) knew how to stay alive, (many of them are here, thank you !) and not in your "managed airlines" and their effective aircraft.
roulishollandais

Originally posted by roulishollandais ...
That is very dangerous ! Every sixty seven hours you have a four minutes crash !
Now, now, now ...:=

PJ2 forgot to hold the "9" key down after the decimal point, but I doubt he would forget to push the stick forward.http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/thumbs.gif


@mm43:
it is still very dangerous if PJ2 did forget a final "9" , and we replace thanks to your intervention, mm43, 99.9% by 99.99% : 1/10 000 means my children would have every 667 hours a :\ 4 minutes crash
Would you accept that mm43 ? no, no, no ...:=

And PJ2 would need 22 000 FT (or 35 000 ?) to push and recover, instead of 7 400 FT !!! no, no, no ...:=

roulishollandais

[jcjeant]

Hi,

PJ2
The industry has been largely successful without AoA so the argument that AoA would have "saved" AF447 vice other standard handling procedures (not stalling, pushing the nose down once stalled), is not convincing. It wouldn't have saved Colgan, it may have saved the Airborne Express DC8, it may have saved the FedEx ATR72 at Lubbock and (IIRC) a B727 where the pitot heat was turned off, but in the millions and millions of hours flown in all types, lack of AoA has not been an issue until AF447.
but in the millions and millions of hours flown in all types, lack of AoA has not been an issue until AF447
Indeed .. millions of flight hours and not get in trouble ...
This is always the case .. until there is a trouble ...
The cargo door locks of Boeing B747 .. millions of flight hours and not get in trouble .. until a door opened in flight and passengers ejected ...
Same plane type ... explosion .. cause problems in fuel tank
The B737 ... millions of flight hours .. until the rudder jammed ....
And in fact all the improvements of flight safety has always risen from the
similarly
Millions of flight hours with no problems .. and then .. suddenly the "apocalypse" and the eternal question:
Why ?

[Lyman]

As an Anthropologist, jcjeant, I say without fear of contradiction, that in my experience, Man is the only sentient being who predictably misses that which is obvious.

and, sadly, sometimes deliberately.

[PJ2]

roulishollandais;

You say, "And PJ2 would need 22 000 FT (or 35 000 ?) to push and recover, instead of 7 400 FT !!! no, no, no ..."

Sorry, where did you get 7 400ft from?

PJ2

[RR_NDB]

Hi,

jcjeant:

Why ? (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-58.html#post7111411)

In all 3 cases you mentioned there was "tangible" failures, even on TWA800.

Now, if HF have to explain the crash we may say the causes are getting more "intangible"

Since the beginning BEA told the equipt. worked as designed.

:E ,

Mac

[gums]

Salute!

I gotta go with Roul ( newbie on this thread best I can tell) and RR.

I can not claim 20,000 hours ( mostly cruising at 35K and monitoring the autopilot, and taking a break now and then to the bathroom or whatever). I will match my 4,000 hours against anyone here. Except for the ocean crossings, I rarely flew much autopilot except to find a let down chart or something that I had not prepared for. Oh yeah, every two or three hours crossing the ocean I had to snuggle up to a tanker at about 30 or 40 feet away and get some gas. BFD except in a thunderstorm or at night.

The human interface is extremely important. The basic pilot skills are extremely important. If the robots were really good and had some judgement, then I would be O.K. with them. But they are not.

Somehow we have lost a sense of airmanship. It's as much a skill as an attitude. The new planes are more like a video game than an old plane that required touch and feel and basic flying skills. I flew all kinds. The big thing I learned was that basic flying skills counted more than all the computer assistance and such when things turned to worms.

I shall go with Doze and others' opinions that had the pilot just let the plane alone for several seconds, that we would not have this thread or the fatalities. Many contributing factors and design aspects here. We shall see.

I shall not absolve the crew's actions.

But I cut them a little slack due to the Airbus reversion mode sequence and some confusing warning indications. I also submit that the 'bus FBW system is not as "safe" as all the PR implies.

Gotta go....

Gums

[RR_NDB]

Hi,

Anthropologist, Bear:

Man is the only sentient being who predictably misses that which is obvious. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-58.html#post7111416)

Why?

Mac :confused:

[jcjeant]

Hi,

RR-NDB
Since the beginning BEA told the equipt. worked as designed.I hope you noticed that in the other cases (B747-B737) the equipt worked as designed (bad design of course) .. and it was "apocalypse" :uhoh:
Working as designed is not a proof that the end result will be good

[Turbine D]

roulishollandais,

I think one has to be realistic when citing percentages. The gold standard is Six-Sigma - 99.99966% free of defects. You can calculate what this might mean in terms of successful flight hours. However, I should point out this does not include the thing that nature might throw at you, variables that are beyond any control. When you take into account the machine and all of its various component makeup, the human factor and all that could and does imply, and the nature of flying in an atmosphere that can be and is unpredictable at best, Six-Sigma can be a goal, an admirable goal, but may take considerable time to reach, if at all obtainable. Think of all that must happen to make it so. However, the human aspect can overcome some of the shortcomings of the other two variables, as has been pointed out by various pilots, based on proper training, proper instincts and hands on learned abilities...

TD

[RR_NDB]

Hi,

gums

The basic pilot skills are extremely important.

Rich data base, Intrinsic creativity enhanceable by adrenalin.

Nothing compares.

The big thing I learned was that basic flying skills counted more than all the computer assistance and such when things turned to worms.


In "normal conditions" automated Systems are designed to work near perfection. Who can perform so good during long hours "yaw dampening"? And performing as an autopilot? Just impossible.

When "worms are in sight" the System MUST HELP. This is ABSOLUTELY necessary in a good design (and relates DIRECTLY with the "interface")

It's inconceivable one need to interpret when facing difficult conditions "what the plane is telling" through unreliable, massive, erratic or intermittent indications of it's machine to man, interface.

PR (and sales people) concentrates on the "beauties" of new designs. What we know (in real world) is another face. Every mechanism has limitations.

In AF447 case during this fascinating and rich effort a large group of motivated professionals in a high Synergy addressed most aspects involved, here in PPRuNe.

Certainly HF aspects (closely related to the "interface") will be the "key" to explain crew actions that sadly put the A/C in a "terminal state" in a "near terminal speed".

Mac

PS

I also submit that the 'bus FBW system is not as "safe" as all the PR implies.

May be we could put like:

Airbus SAS approach is not what PR implies. Generating (clearly generated this even in pilots) a perception of something less dependent on "crew skill" than other approaches.

IMHO it's the opposite: In order to be able to interpret (and act) when the 'worms are in sight" the 40 seconds PJ2 mentioned could be too long.

In terms of this loss of control, while it unfolded literally over a period of about 40 seconds, oddly that is tons of time to do something, but it is not a lot of time for assessment, discussion, action.

The "capable pilot" here makes the difference.

Yes, we shall see. We don't have all required information. We need to wait to conclude.

Every judgment before (all relevant factual information) is at least, risky.

[Lyman]

PJ2 NeoFit came up with 3300 feet to recover the a/c from full STALL (SL).

The 7400' was elsewhere, Did we not see the video? Full Stall to safe landing in camera.

jcjeant, In Construction, it is called "As Built", or "After the Fact". Also, there is Design/Build, which is too sloppy for Airbus. It worked well enough for Clarence Johnson,

RR_NDB. Because sometimes Humans prefer "Correct" over "True". "Correct" requires a tacit wink/nod, though, and falls short of 100% SAFE. Culture requires compromise, and sometimes compromise is deadly. For instance, I ask three questions, and get an answer for one unasked, ignoring the valid three.

In the scheme of things, all here are safe, for now, and should consider thinking of some scary things, instead of reassuring each other. Ignorance is not bliss, it is merely ignorance. And faux comfort.

[john_tullamarine]

The gold standard is Six-Sigma - 99.99966% free of defects.

Indeed, but 6σ applies to the folk who step back a pace and contemplate the population.

For the individual on the day, the reality is more binary ... either you are in strife or not. From the perspective of standards, either the system accepts the population risk as acceptable (and the occasional mishap as inevitable) or provides some semblance of training (admittedly at a cost) to give the individual some reasonable hope of employing the cognitive things which humans are good at.

Like PJ2, I hold the view that a bit of sim time for crew exposure to the improbable may (not necessarily will) provide some benefits.

[Turbine D]

Hi John,

In a way, that is what I was trying to say, but maybe wasn't clear enough. Indeed, training can/does help overcome the realities presented that are not normal.

Thanks and regards,

TD

[RR_NDB]

Hi,

jcjeant,

I hope you noticed that in the other cases (B747-B737) the equipt worked as designed (bad design of course) (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-59.html#post7111483)


1) 737 PCU malfunction: Only after Colo. Spgs crash ("testability problem") they started to detect a hidden problem. i don't know if we can classify this as a "design problem". Far more subtle than the Pitot's problem.

2) 741 Flight 811: Indeed a design fault with MANY incidents before this case.

3) 741 TWA 800: IMO doesn't fit as a design problem.They improved the design reducing the risk of a similar case.

Mac

We need to be cautious dealing with "design problem" classification. Everything could fit in that category. E.g. "lack of redundancy is a Design Problem? Not necessarily.

The use of Redundant OBSOLETE sensors (triple redundancy) that "FAILS" near simultaneously IMO is a SERIOUS DESIGN PROBLEM. The use of just one PCU (737) compared to the (redundant) 727 IMO is not a Design Problem. And in the 737 the lack of redundancy magnified the (hidden) problem.

The first case (COS 35 final) was shocking. The plane just rolled diving vertically from ~ 1,000 ft AGL when in the final. I can't imagine the feeling of both crew. Gusty winds masked the analysis. And it was a subtle (not easily testable) problem. Few days before the crash there was "strange" behavior involving rudder. Intermittent failure in a "closed loop" System.

Testability is a complex issue. For some problems only TIME can test.

[RR_NDB]

Hi,

john_tullamarine:

Like PJ2, I hold the view that a bit of sim time for crew exposure to the improbable may (not necessarily will) provide some benefits. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-59.html#post7111516)



At least could help to emphasize what can happen. I think, worth the effort. For some. Not necessarily all.

Mac

[john_tullamarine]

I can't imagine the feeling of both crew.

But now contemplate the situation had a crew either in flight experience or sim exposure to something similar.

Might they have been able to bring that experience to bear in time, for instance, to try asymmetric thrust as a counter in combination with aileron input ?

An effective initial solution doesn't always have to be elegant ... but it may need to be implemented very quickly without too much Monday morning quarterbacking if it is to have any real world probability of working ...

[Lyman]

John T.

I don't think we can fairly or even logically reject that there were ad hoc and unusual attempts made at recovery. The panel here would have none of it, in general, but let's not forget Bonin, (RHS, PF) was a licensed Glider pilot, as was Sully. I have always been impressed with the rating, for obvious reasons.

We'll see. Or not. I have been unwilling to make the leap I believe necessary to simply assume that an experienced Glider pilot would be so thick in PITCH.

Energy.

[lomapaseo]

99.99966% free of defects

I have no idea to what this applies :confused:

certainly not to individual parts like screws, pistons, solonoids, switches and jeezuz bolts.

Maybe with a good quality control program and sampling you might get 75%, but an awful lot of stuff on planes rarely gets the sampling to even validate the quality control.

The idea to achieve safety is to provide backup, like redundancy, alternate paths and trained pilots. That way it takes a lot of things going wrong to result in a catastrophic outcome (death and destruction)

again my belief is opposite a lawyer's

"if not for"


and instead I would still insist on

"in spite of"

as the way to achieve safety in a complex machine of multiple systems called an airplane etc.

[gums]

Salute!

The Colorado Springs disaster was, indeed, sad. Also lost another jet of the same type near Pittsburgh if I am not mistaken. The stoopid rudder actuator would stick and only a determined effort punching the pedals could offer any hope of recovery. Sad, sad, sad.

(flew into the Springs and out last week and that crash was on my mind)

Folks here that have never flown a swept wing to the edge of the envelope and beyond may not appreciate how important yaw control is. "Retired" can tell all about that when flying the Phantom - lock stick between legs and only use rudder to roll the sucker to counter adverse yaw, and the jet is shaking and the buffet is very severe. Didn't have that problem in the Deuce due to basic aero characteristics of a delta, nor in the Viper- aero and flight control systems much better in the late 70's and 80's.

I side with PJ and RR for the most part. Training and the human interface are two factors that must be addressed by the 'bus folks and the BEA report. I will take PJ's experience in the type over anybody else here. I will take my own experience in 5 or 6 jets over anyone else here, as well.

late here in Florida, and gotta go.

Gums

[Machinbird]

The final loss of height is not your 22000 FT, nor FL200, nor 15000 FT,
but 5000 (max) +(FL100-FL76)=7400 FT
Dutchroll, these were speculative numbers for a last minute escape from the full stall. The pullout is dependent on achieving near maximum performance, and the stall recovery altitude/time is entirely speculative. The emphasis should be on speculative. Remember, this is for near maximum performance. Without an AOA indicator, I see no way to come close to these numbers. OK465 seems to believe it will be a problem teaching people when and if to add some power during pullout to improve performance. He is probably correct.

My intent in coming up with these speculative pullout figures is to remind people who may be in similar situations to not give up! You might think you are toast, but keep thinking and keep flying to the limits of your ability. You just might get a pleasant reward. The light at the end of the tunnel is not always a train.:}

PJ2s post #1107 (http://www.pprune.org/7109943-post1107.html) is the result of a number of simulator runs of the AF447 scenario and involved a high altitude initiation of the recovery. This is essentially valid data within the unknowns of how well the Sim matches reality in an area it isn't really certified for. Perhaps PJ2 can tell us something about the power setting during the recovery. (High power would be a worse recovery condition than idle power.)

If the recovery is initiated lower, it is likely that the pitch rates nose down will be higher, the descent rates a little lower, and thus the time/altitude loss to unstall the wing may be somewhat less.

Just like the Viper has a sticky spot in its pitching moment curve, the A330 may also have a sticky spot. There may be techniques for achieving a higher nose down pitch rate, but it is not worth risking a test crew to find them. With proper fixing of the myriad Swiss cheese holes that led to AF447, we should never see a similar incident involving an airliner during my lifetime (and I plan on living a long time).

[Old Carthusian]

I rather think we need to be very careful in talking about 'deteriorating skills' because of increased automation. A cursuory examination of past accidents will show many, many examples of where piloting skills were lacking and these on unautomated aircraft where piloting skills were required. Automation has improved aircraft safety immeasurably. Several 'solutions' have been suggested on this thread but all run up against issues. Changing the technology is neither easy nor necessarily desirable. It must be realised that this is a human caused accident and removing the pilot would have probably been the only way to have avoided this accident. Remember the adage 'We keep on making things foolproof but God keeps on designing better fools'. I will state again - no matter how well you design a system someone will find a way to evade it and wreck it in a totally unexpected way. That is the reality of this accident.

[PPRuNeUser0198]

Why did the crew not maintain 85% thrust and 5 degrees pitch up to maintain a safe speed?

[Machinbird]

Why did the crew not maintain 85% thrust and 5 degrees pitch up to maintain a safe speed?T-Vasis
That is the $64,000 question. There are two basic lines of thought on this.


They mis-applied UAS procedures applicable to lower altitudes.
They were distracted by a roll PIO to such an extent that they lost track of what was happening with the aircraft.

We are all hoping that BEA's Human Factors experts can figure this out with some clarity.

[Lyman]

3. It is not known the condition of the pilots' displays. The Data divulged on this thread is from a source unavailable to the crew during their ordeal.

4.From the awkward interaction between the two (then three) pilots, some problem can be taken re: at least the Flying Pilot's displays (screens).

5. There is no conclusion re: some interesting observations made by the pilots themselves: "What was that?" "I feel some crazy speed"....etc. "ER....What are you doing?"....et al.


6. Additionally, these noises and artifacts are not released as found, and given the BEA make no comment with exposure of their evidence, the conclusions here are completely unsupported.

7. With his very first input, the Pilot was greeted with a STALLWARN, cricket, sufficient one would think to start a cascade of confusion from a cascade of data, some important, some less so, some completely unnecessary.

It is quite possible that 5 degrees NU was completely insufficient, and without knowing the data supplied to the crew, the "book" may have been worse than the reality.

It is painfully clear that the situation was challenging, despite the protest of those who believe it was just another night at the office.

Hopefully, and with a complete record, more will be known. It is possible what the pilots had to deal with will never be known. To the extent that leaves the crew open to ridicule, such is life.

In a community that trumpets perfection, it would appear ironic that most are happy with group think and "consensus".

[Machinbird]

I rather think we need to be very careful in talking about 'deteriorating skills' because of increased automation. A cursuory examination of past accidents will show many, many examples of where piloting skills were lacking and these on unautomated aircraft where piloting skills were required. Automation has improved aircraft safety immeasurably.......
OC, deteriorating aircraft handling skill level for pilots of automated aircraft are a fact of life. The reasons are myriad but are frequently discussed on a number of threads on this forum. We have always been crashing a lot of un-automated aircraft through the years, mostly for the reason of pilots attempting to fly beyond their abilities/training.

Airlines see pilot training as a business case expense, IMHO there will be no dramatic investment in better/more pilot training.
(There is probably room for simple desktop trainers to maintain basic instrument skills in the budget. That is where they will likely get the best bang for the buck. As a suggestion, airlines could put several of these in pilot lounges and provide incentives for using them.)

That leaves better automatic systems to keep the pilots in their 'executive cocoon' while the computers re-configure the aircraft to keep on trucking. This is controversial of course, but allows safety to be improved with minimal cost. How many accidents can you think of that resulted from pilot mis-application of emergency procedures?

[Lyman]

POV only.

From an historical perspective Airlines safety has everything to do with two things.

Dramatic improvements in Powerplants, and a better understanding, and management of Weather.

Both had to do with economics. UAL: (on the DC6) "RADAR equipped".

Time in Transit. Less fuel with a concomitant increase in pricing. Safety is in there, but mostly as a left handed marketing tool. It is poor marketing to allow clients to think too long of safety.

The result of the sacrifice of these folks will be in catching AirFrance with trousers down. For money. Is there another way to see it? To the extent that punishment is harsh, and the deal improves for passengers/crew, it is a sad but important take-away.

[PJ2]

Machinbird;

In re the sim exercise, power was at various settings (though never TOGA). Also, there were variations in altitude needed for recovery. In one exercise we "did nothing" and the aircraft happily remained in stable flight at FL350 as power remained in "TOGA LOCK" (last setting before disconnect) and the pitch at 2.5deg, approximately.

I suspect some tests of hand-flying the A330 in Alternate Law would be part of the investigation to see how sensitive the aircraft is, to further understand the initial actions on the stick of the PF.

As I've mentioned a number of times throughout these threads and because not all here seem to comprehend, the air is much thinner at cruise altitudes and so with something the mass of an airliner one must be gentle with the controls and make tiny stick adjustments in manual flight. This fact was first stated in Davies' book, "Handling the Big Jets", in 1967 - 1970.

gums, you're right about the sim not really being able to teach this kind of thing - you can't feel "the mush"...the slight delay in responsiveness in both changes in attitude and flight path even though the subtle indications may be accurately simulated. These kinds of things need to be taught in the airplane at cruise altitude. The airplane hand-flies very well in cruise in Normal Law but I haven't flown it in Alternate Law so can't comment on that. Alt Law is supposed to be wysiwyg. I've hand-flown other transports at cruise which don't have C* Laws or the "add-on's" which provide the usual protections we see in these aircraft, (the B777 has some protections built in and I suspect the B787 has more), and they're sensitive so it is important to know how the A332 responds when hand-flown at altitude in Alternate Law. I suspect with gentle handling it responds as any other similar type, (but "gentle handling" it appears, is not always natural and must be taught)

I stress too, that, in accordance with most air regulations (I have cited the CARS here in the past but the JARS I suspect are similar), the simulator is only used to teach the approach to the stall and not stall recovery. The full stall is never taught and is not required to be taught, for a number of important reasons, (this has been extensively discussed in other AF 447 threads.)

In the approach to the stall, the SOP is to apply TOGA power, lower the nose and minimize altitude loss. This still applies! The standard response to a full stall is to first get out of the stall and that means reducing the AoA to below the wing's stall AoA, then recovering by (gently!) controlling the loss of altitude and re-applying thrust as needed.

For some here who still don't appear to comprehend these things and who won't and can't support their recent statements that recovery from a full stall in transport category aircraft from cruise altitude, (as occurred here), takes "3500ft" or "7400ft", unloading the wing to reduce the AoA to below the stall AoA takes an enormous amount of altitude, and, while stalled, the altitude loss is extremely high, again as seen in the data.

[PJ2]

machinbird;

In your response to T-Vasis:


They mis-applied UAS procedures applicable to lower altitudes.
They were distracted by a roll PIO to such an extent that they lost track of what was happening with the aircraft.

A third possibility, (perhaps a sub-possibility from either of the two above that you've mentioned), may be the PF's relative inexperience with high altitude manual flight, (in Alt 2 Law, roll is more sensitive than pitch), resulting in an unintended stronger pull on the stick than he desired, which he then began wrestling with at the same time he wrestled with (and nicely got under control) the roll oscillations. The two "stall warning" blips surprised both crew members, (What was that? - PM), but of course both warnings were legitimate as the AoA for the ECAM Stall Warning had been momentarily exceeded. Subsequent stick and pitch actions are less understandable but obviously plausible to the crew. Why?

[Machinbird]

A third possibility, (perhaps a sub-possibility from either of the two above that you've mentioned), may be the PF's relative inexperience with high altitude manual flight, (in Alt 2 Law, roll is more sensitive than pitch), resulting in an unintended stronger pull on the stick than he desired, which he then began wrestling with at the same time he wrestled with (and nicely got under control) the roll oscillations. The two "stall warning" blips surprised both crew members, (What was that? - PM), but of course both warnings were legitimate as the AoA for the ECAM Stall Warning had been momentarily exceeded. Subsequent stick and pitch actions are less understandable but obviously plausible to the crew. Why?
PJ2
You will get no argument from me here although IMHO this belongs under (2)-They lost track of the aircraft due to roll control problems.
(in Alt 2 Law, roll is more sensitive than pitch) Your observation on the force mis-match between proper pitch and roll inputs is significant from an aircraft handling POV. This is an undesirable combination.

[PJ2]

Hi Machinbird;

Re, roll sensitivity vs pitch in Alt 2 Law and your observation, "This is an undesirable combination. "

No, the difference is a matter of degree, not of characteristic. Both are eminently controllable but the flight controls also respond to large, quick inputs just as any aircraft would. These inputs were, for whatever reason and by all accounts, very large and very quick...sufficiently so as to cause the stall warning to trigger twice. Again irrespective of the reasons which may be found in understanding of high-altitude flight, training, experience, HF and technical/aircraft, this qualifies as mishandling the airplane. 'Why', is, as we many realize, is the question, which implies at this stage nothing regarding the PF or the airplane or ?, except a need to understand.

[roulishollandais]

99.99966% free of defects I have no idea to what this applies

Congratulations ! Never use a number without to know exactly which reality it covers. specially when it is probality or statistics.

PJ2 gave us unvolontary an opportunity to show that. When he used the number 99.9% it was a synonymous of high safety in his mind :(, as it is in horoscope or lottery, and tried to sell us some sterilized speech. In the cockpit, it would have his place, but we are here on the ground, all together, to get a little smarter with higher air safety, after AF447

I immediatly choosed to show what was wrong, and applied 99.9% to flight time, as he was in the defender's position, and the result was 67 hours safe vs 4 min crash, which is very unsafe.

The gold standard is Six-Sigma - 99.99966% free of defectsContinuing this way, with 99.99966% free of defects, it means 3.4 / 1 000 000 defects.
For 4 minutes crash, you have 19 608 hours safe (in his whole life, PJ2 has flown around 20 000 hours)
For 1 minute crash, 4 902 hours are safe...

[gums]

TNX for the support, PJ. Ya gotta head south this summer and sip a brew on my deck in Colorado. same for the 'bird.

Seems the 'bird is still positing a roll PIO and concentration on that while pulling back on the stick. The jet appears to have done a slow turn to starboard until impact, regardless of pilot stick commands. Can't find any trim for roll in the 'bus manuals you have provided, but looks like the jet simply commands zero roll rate if you let go of the stick in the roll axis. Is that right? My little jet had that "hat switch" on top of the stick, so you could actually command some roll rates even with "hands off". Same for baseline gee.

I fully understand the lack of training in post-stall recovery in the actual airplane, especially at high altitude. I also understand that many pilots have a problem recognizing the mach buffet compared to stall entry indications unless they have experienced both in a real airplane, but seems to me that a sim could implement some "shakers" and "buzz" things to help. My bent wing experience before the Viper helped me to see the difference, and yeah, it was "feel" and "touch". OTOH, my straight wind experience in two jets that could reach the critical mach without being in a steep dive was more interesting - wing rock, aileron buzz and even aileron reversal due to the shock wave effects, nose tuck that could not be overcome without speed brakes and reduced power ( elevator couldn't work due to the shock wave). Not fun at all.

After two years here, I am leaning back to my original position about the jet being in a stall all the way down. Not the classic "deep stall" as we see in the Viper and the T-tail airliners, as I originally suspected. Seems the 'bus has plenty of nose down pitch authority even with the THS cranked to the stop.

I still have trouble believing it would take more than 10,000 feet to recover from the stall they were in. Speed was above 100 knots actual dynamic pressure, and just reducing the AoA to zero for a few seconds should have produced useable lifties from the wings. So I figure a recovery in less than a minute and at 20,000 feet or so. The lower the plane got, the easier the recovery due to reduced mach and increased air density.

The mass of the airliners was a lot more than anything I ever flew, so I can imagine things taking longer, but not 35,000 frigghin' feet. I also understand that entry to an unusual flight condition would present problems recovering before getting into real trouble. Good old momentum and such. We called it "overshoot", but our control surfaces were lots larger for our gross weight than any commercial airliner.

Am waiting for the "final" report and recommendations. Will be disappointed if the human factors are not addressed, and not just training, but the displays.

[RR_NDB]

Hi,

(in Alt 2 Law, roll is more sensitive than pitch) (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-60.html#post7112412)

PJ2, this characteristic was intentionally implemented in the A/C? If so, why? If not, it could be improved?

For example, take into account Altitude as an input parameter in order to "help" the crew perform better the (less common) cruise level manual handling? That can occur in adverse conditions (severe turbulence, degraded (intermittent) avionics, etc.)

How you compare the difference (quantifying) in sensitivity (roll x pitch in Alt II) versus Alt I (if there is) and versus NL. The reason of my question is:

To understand "how different" was the plane PF "received" when was "inserted" in the loop and started to apply the corrections he decided (after processing the inputs he is receiving).

The "decision making" PF made (his first reaction) is unknown, obviously. But, we can infer, something important (to him) under the conditions he was facing, led him to.

[rudderrudderrat]

I concur with your description of the "feel" of an aircraft in #1162.
In Normal Law - there is no such feeling of how much control surface deflection was required to satisfy your request.
Can't find any trim for roll in the 'bus manuals
Correct.
In Normal Law, it is trimmed automatically by deflecting the ailerons sufficiently so that at stick release there is zero roll rate.
In ALT LAW it's like a Boeing in roll. Stick deflection demands a proportional aileron deflection. The only trim available is rudder - to yaw the wing sufficiently so that aileron deflection is not required to hold the wing level. The roll rate is faster.

In Alt Law, Pitch control "feels" the same as Normal Law - you select a delta g, it changes attitude and stays there on stick release - automatically trimming as the speed washes off.

[roulishollandais]

Hi PJ2,

Warning : Machinbird called many times to high vigilance about this calculation. I agree totally to the calculation but also to the warnings, and here is copy of his last warning.
Dutchroll, these were speculative numbers for a last minute escape from the full stall. The pullout is dependent on achieving near maximum performance, and the stall recovery altitude/time is entirely speculative. The emphasis should be on speculative. Remember, this is for near maximum performance. Without an AOA indicator, I see no way to come close to these numbers. OK465 seems to believe it will be a problem teaching people when and if to add some power during pullout to improve performance. He is probably correct.

My intent in coming up with these speculative pullout figures is to remind people who may be in similar situations to not give up! You might think you are toast, but keep thinking and keep flying to the limits of your ability. You just might get a pleasant reward. The light at the end of the tunnel is not always a train.http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/badteeth.gif


You will find most of the details of the 7400 FT calculation, iin these Thread 7, posts :

- Hazelnuts39 #945 (phugoid)
- Machinbird #950 "10,000 feet is about the last point one could hope to begin recovery"
- Machinbird #955
- Hazelnuts39 #960
- Machinbird #962

The result is :
with begin of recovery at 10 000 FT
-5 000 FT (max) (for unloading the wing)
- good performance with 0.15% thrust for recovery :
loss of height = FL100-FL76)= 2 400 FT

Finally the total loss of height is - 5 000 - 2 400 = - 7 400 FT

:)

[Machinbird]

Continuing this way, with 99.99966% free of defects, it means 3.4 / 1 000 000 defects.
For 4 minutes crash, you have 19 608 hours safe (in his whole life, PJ2 has flown around 20 000 hours) http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/cool.gif
For 1 minute crash, 4 902 hours are safe... http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/badteeth.gifFortunately, due to redundancy concepts, we can have small failures and not have an accident. Our actual airline accident rate on an hourly basis is far better than 6 sigma. (For most airlines at least)

Compare that with the overall Navy accident rate during most of my career--3.5 to 2.5 accidents per 10,000 flight hours. Makes one wonder if you will personally get involved in the statistics during your career.:hmm: Then add combat hazards to that.:{

[RR_NDB]

Hi,

Machinbird:

Fortunately, due to redundancy concepts, we can have small failures and not have an accident. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-60.html#post7112634)

Did you investigated Kapton (wiring) related incidents, more frequent (AFAIK) in naval aviation?

Posted reply (http://www.pprune.org/tech-log/480601-hf-keying-when-refuelling-represents-fire-hazard-2.html#post7108205) on "HF keying during refueling" thread (on F4 fuel probes) and sent PM on "unloading the wing on an airliner"

[PJ2]

roulishollandais;

First, in your post #1148 (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-58.html#post7111250) you have completely missed my point in favour of making your own. Further, in your choice to parse the metaphorical expression, "99.99%" rather than try to understand what was being conveyed by the post you have demonstrated a complete absence of understanding of how flight safety works and why this industry is so remarkably safe and you have refused to engage in a pleasant dialogue which would have led to further discussion and perhaps even further comprehension.

My "99.99%" was a metaphor...a figure of speech indicating high, versus medium or low correlation between SOPs and their adherence, and the very high safety levels of our industry. You took this expression literally, applied mathematics to a metaphor and drew precise and entirely incorrect conclusions which have no basis in the kind of activities, processes and preventative strategies which bring about such high levels of safety in this industry. Should we briefly examine a counterexample, you take the "ten to the minus ninth" standard and apply it to one variable, (my number of hours flying) and ridiculously try to extend such "chances" to the industry. That's not really how it all works, you see.

Your approach utterly fails to take into account this complex and effective safety system which is decades old and to which many engineers, human factors specialists, pilots' associations, manufacturers and airlines have immeasurably contributed and so, beyond a detail of what "99.99" means statistically, the point you are trying to make fails to make any contribution to a broader understanding of what is behind the millions of hours of unremarkable flight and why this accident has so captured the imagination and interest of so many.

Regarding your calculation of 7400ft, you have set up a straw man case which bears no relationship to the discussion at hand yet the outcome for your straw man argument, "7400ft" is applied to the case, again at hand. I don't know how you do that and still claim some correlation between the two but of course you are free to do as you please notwithstanding relevance to the discussion.

For your "model recovery" (which may have basis in fact but which, again, bears no relationship to the present exercise), you begin the recovery at FL100. If one approaches (not enters) the stall at FL100, recovery may occur within your calculated altitude if the recovery from the approach-to-stall is handled smartly. In fact in simulator exercises, typical altitude loss using the "Approach to Stall" procedures was held to between 500 and 800ft depending on a number of factors. Further, I suspect that had these procedures been applied just prior to the apogee, (lowered pitch to 5deg below the horizon), TOGA thrust, respect the stall warning (in Alt law), the event may not have developed into a full stall. Once the AoA was increasing to 30 and 40 degrees, that procedure would NOT work and recovery from the stall would be necessary, which means pointing the nose down to between 15 and 20 degrees ND and accepting the extremely high rate of descent.

However, the simulator case at hand and our real case, AF447 was in full stall, and at FL100 was still descending at approximately 17000fpm with an AoA of 45deg. Your case and your argument is therefore wholly non-applicable; the aircraft is not going to recover in "7400ft".

You may pit your willingness to argue minutae such as "99.99%" against the experience of several here who know their stuff and against the numerous simulator exercises all of which run counter to your notions of recoverability. That, as you say, is your choice. But I submit that the issue here is not one of being right come what may but is one of understanding both the aircraft and high-altitude flight and what needs to be done by the crew to swiftly unload the wing to reduce the AoA below the stall when a transport aircraft is in a full stall at cruise altitudes.

It is completely incorrect to suggest that a fully-stalled airliner at cruise altitudes of the kind we are discussing here will recover from such stall with a 7400ft loss of altitude from said altitude. An appreciation of the subtleties of this very dynamic set of circumstances would go a long way in support of a collegial exchange of views.

The only case in which your understanding may be marginally in accord with the facts is in strictly controlled flight test conditions where the approach to the stall is part of a test regime...note I said "Approach to the Stall" because NO ONE intentionally stalls a transport aircraft at any altitude any more, period. The Flight Test manual for these exercises states that such tests will be done at FL140 or higher. The one case which has occasionally been introduced here is the Perpignan case but this crew collectively decided to approach the stall test on final at 3000 AGL and rushed the process. The recommended height for such tests is for a reason, and the stuck AoA vane is one clear reason why.

[Lyman]

It was mentioned that "Approach to STALL" recovery at Colgan prior to, no altitude loss was the goal. You mentioned a recovery with ND; at Colgan, losing more than 100 feet was disqualifying. I tend to agree with your presentation, from 350, any recovery from Stall will use up more than half the total altitude, and may not even be possible, especially if no AoA indicator to 'help' guard against secondary STALL?

I cannot understand why the focus is on something that is at best tangential, and when grumpy, totally irrelevant. To have lost control in the first place makes the STALL recovery argument silly. Whatever the cause, to deplete her energy in that fashion suggests that it will not be suddenly available to be careful with that which may be regained in a 20k dive?

[RR_NDB]

Hi,

Machinbird:

Fortunately, due to redundancy concepts, we can have small failures and not have an accident. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-60.html#post7112634)

Actually a good "Redundant Design" can support (degrading) even major failures.

Our human body is an example: We are adequately redundant: Duplicated lungs, Kidneys, arms, etc. The Designer (:) implemented this SAFETY feature extensively in nature.

The "Fault Tolerance and Graceful Degradation" was certainly the major achievement of this "Design Approach", actually by " Species Evolution".

On big Systems we may comment Fukushima disaster on that:

Their "APU" failed completely (when they most needed it) after the "protections" were triggered (by the G's) and the ABSOLUTELY NECESSARY cooling of the reactors ceased what lead to a "Thermal Runway".

They had to use Chinooks (risking) during subsequent days at the emergency peak of the event.

Why it failed? The March 11 wave was taller (than the expected maximum). The "equipt" operated as per design. :} . Like the 747 cargo door issue (flt 811) and a multitude of other sad examples.

The Engineers responsibility is enormous: Quality (the result of the design) depends totally from a good Specification.

In aviation, as many here commented the pilot must operate "inside" the "machine limits". Example: The Pitot's current limitations. The first threat F-GZCP "received" was the path (with a "last minute" slight deviation) towards a WX other crews deviated. This exceeded the limits of an important element of the System, (Pitot's) triggering a cascade of events. A rare chain so complex, being regarded as directly related to (complex and "subjective") Human Factors.

The Redundancy applied in modern airliners is carefully studied to the Industry requirements.

And frequently reduce the consequences (after major failures) or even transform potential serious accidents in "manageable incidents".

It can tolerate major failures? Examples abound. And the crew (a good preparation and integration to the equipment) is a VITAL "part" of the System (The effective aircraft: System+crew).

Gen. Chuck Yeager in October 1947 successfully landed his bird (X1) after facing a TOTAL electrical failure just after separating from "mother ship".

The fuel valve" (a redundant safety feature) was an important factor to save him and the bird. And the toll (of the failure) was put in him. Who accomplished (partially) the mission. Reducing it to an incident.

PS

One magnificent example was Capt. Sully achievement. (I remember when i crossed Hudson river in a rented motor home). He even flew "offset" the bridge aligning the bus when in the short final. The comment of his wife after learning the fact shows how Sully was the perfect "element" of a well designed System, his bird an the crew.

[Turbine D]

PJ2,

I am sorry my mention of Six Sigma and its goal of 99.9966% free of defects has been misinterpreted by some and not understood by others. To somewhat clarify things, Six Sigma is a business management strategy to improve the quality of process outputs by identifying and removing the causes of defects (errors) and variability in processes. It is not confined to just improvements of manufacturing processes. In the airline industry, there are many processes which in one way or another contribute to the safety aspects of flying, which is very, very good considering the complexities.

Processes usually do not perform as well in the long term as they do in the short term and there are many reasons why this is often the case. Figuring out why is the task and identifying improvements is the goal. To accomplish identification and zero in on meaningful improvements, the Six Sigma process has over 30 management and method tools that can be used. Most do not require a statistician background, just good common sense and knowledge of the process to be improved. For example, the generation of a key list or lists is a tool.

Now pertaining the A-330, there is a memory list for pilots to memorize regarding the occurrence of unreliable airspeed. It defines what a pilot should do if UAS is encountered and as I recall, it emphasizes low altitude critical situations, takeoffs and landings while also mentioning altitudes above 20k feet. It does not mention the words "at cruise" or "high mach". Now in cases of UAS at cruise and high mach, most pilots have figured it out, what to do to prevent LOC, but more recently, not all have done so. Would an improvement to the memory list to include what to do "at cruise" help future situations that might occur? This is just a small example of a Six Sigma tool being used to improve the output of a process. Then one can think about training processes and what can be improved there. People, generally bean-counters, say training cost a lot of money, it does. However, spending some extra money can result in much greater savings if errors can be prevented. GE, a big Six Sigma advocate had reaped over $1B in savings using the Six Sigma process, and that was 9 or 10 years ago.

It is interesting to note that at least two airlines do use the process to some degree with reported good results, Air Canada and Alaska Airlines. Many other companies in the aerospace industry use the process, Boeing, Honeywell, GD, Lockheed Martin, Northrop Grumman to name a few. Even the USAF, USN and USMC are using the tool to improve their processes.

That is what I meant by Six Sigma...

[Mr Optimistic]

I shall have to read these posts in slow time. in the meantime however, let me say that in no branch of science or engineering have i ever heard or had recourse to 6 sigma confidence levels. that is nonsense. however is a 20k margin for altitude loss for 447 reasonable assuming cognition started when the captain returned?

[Organfreak]

To the Fabulous PJ2,
I think part of the problem is a severe language barrier. Just a thought.

[gums]

Thanks, Rudder, it's what I thot. Our simple FBW could be trimmed for a gee ( plus 3.5 gee and about minus 1 gee), which several guys did to have a zero gee hands-off stick for gaining energy. Roll was likewise, and a hands-off normally was a zero roll rate command, but we could manually trim to about 150 deg/sec roll rate command ( or something like that).

I still don't agree with PJ that I couldn't recover in 10,000 or 15,000 feet. Give me a chance in the sim and I'll do my best. The jet had plenty of dynamic pressure, just a bit below normal approach speed. With gear up and some added power, reducing the AoA to zero should have resulted in a significant increase in speed and then a careful, gentle pull back up to level. Screw the mach/overspeed stuff. The jet is tough, and seems very stable.

[Mr Optimistic]

Yes i have heard of lean 6 sigma. Toyota inspired no doubt. management texts are often scientifically illiterate so dont hang your hat on a $10 text.

[Old Carthusian]

Machinebird
Sorry for the delay in responding but I would submit that what you were writing about is the perception not the actuality. Some commentators are adopting a them and us approach - pilots and airline companies. It more suggests to me a misunderstanding of just what it takes to run an airline than an actual perception of reality. It also ties in with a common human perception that things were better in the past whether it is true or not. It may indeed be that some things were better in the past but it is also true that some things weren't.
When you actually study some of the older accidents (and you don't even need to go back to the 70s to see this) you realise that there have always been pilots who have lacked the necessary skills. What automation has done is allowed a massive increase in the number of flights with a concomitant increase in the level of safety needed to sustain this. Of course you will find some negatives in the introduction of automation and the way it is used but this is due to the humans involved NOT the machines. It is clear that the skills of this particular crew played the significant role in the accident but this is not a general malaise as this accident is unique and other crews have dealt successfully with similar incidents. This alone should be sufficient to demonstrate that the accident is not indicative of a general problem.

[jcjeant]

Hi,

It is clear that the skills of this particular crew played the significant role in the accident but this is not a general malaise as this accident is unique and other crews have dealt successfully with similar incidents. This alone should be sufficient to demonstrate that the accident is not indicative of a general problem.

Maybe not a general problem ..
Not related with this particular accident .... but the fact is that for decades .. automatism or not .. the human part of the plane does not seem to increase in security ... it remains stagnant in the statistics .. even with all the progress
Accident statistics (http://www.planecrashinfo.com/cause.htm)
1001 Crash - Airplane accidents statistics - Crashes causes and aeronautical terrorism (http://www.1001crash.com/index-page-statistique-lg-2-numpage-4.html)

[Lyman]

I think it is a reach to even call this tragedy an accident.

If all standards, precautions and minima had been met, this wreck does not happen. End of.

The population of UAS incidents was low, 32 I think, 6 involving AF. Including them in 2 Sigma is a breach of courtesy. The very availability of Back up Speed Scale on Airbus suggests an awareness of the problem, plus the AD on Thales, and the mandated r/r. The threatened RTF by the pilots shows they thought along those lines.

This crash doesn't belong in the gross numbers, it skews the reality like a bookie having a bad and "middled" day.

Flying is far safer than this c/f suggests.

[Mr Optimistic]

PJ2, thanks for your time - it's a pleasure.

[Old Carthusian]

Jcjeant
But as a function of the number of flights - human error is decreasing. The number of HE accidents remains fairly consistent but the number of flights have grown significantly. Partially automation but also partially better standards and culture.

[bubbers44]

Pilots need to know how to hand fly when the autopilot disengages. Usually just holding present attitude and power will handle it while figuring out the details. We have known for decades pulling back at 35,000 ft will stall any airplane so why do it? If you can't remember what your attitude and power settings were just set attitude about 2.5 degrees up and power to where it normally is until you figure out what you want to do.

I have no idea why these guys did what they did but no competent pilot would do it.

[PJ2]

Turbine D;

No worries, I understood what you were explaining and am somewhat familiar with Six Sigma - it was actually quite a good explanation, thank you.

In response to your important comment:Now pertaining the A-330, there is a memory list for pilots to memorize regarding the occurrence of unreliable airspeed. It defines what a pilot should do if UAS is encountered and as I recall, it emphasizes low altitude critical situations, takeoffs and landings while also mentioning altitudes above 20k feet. It does not mention the words "at cruise" or "high mach". Now in cases of UAS at cruise and high mach, most pilots have figured it out, what to do to prevent LOC, but more recently, not all have done so. Would an improvement to the memory list to include what to do "at cruise" help future situations that might occur? Yes, I think so - we're thinking along the same lines and in fact I worked on a design change of this drill last summer, (July, 2011). I have always considered the Unreliable Airspeed memorized drill poorly designed and confusing primarily for those circumstances in which I considered the safety of the flight "not impacted". Here is why:

The UAS memorized drill has a primary decision point, "If the Safe Conduct of the Flight is Impacted", which requires that the crew decide "are we safe or unsafe?" In my view, this primary decision point which is intended to lead the crew into one drill or another, is too imprecise. The "one way or the other" is either to pitch up to certain degrees and set power to TOGA or CLB, or to not execute these memory items and instead "Level Off > Troubleshoot > QRH Pitch and Power tables". In other words, don't pitch up. To my knowledge this crew never experienced this kind or level of training in their recent sim sessions, (and I can categorically say that I have never even encountered the UAS drill in all my Airbus A320/A330/A340 initial or recurrent ground or sim training.)

In a recent and much more detailed description of this drill than I have seen before, a note in the PPT indicates that the meaning of safe conduct will be defined in training. Then it goes on to explain how the drill is done. The presentation is dated 2006.

In other words, "when" the safety of the flight "is/is not" impacted has a definite meaning but this is not defined in any SOP/QRH/FCTM with which I am familiar.

But this isn't the important point. The important point is, no matter what phase of flight, the primary "bifurcation" point...the point where the memorized drill launches one way or the other, applies to all altitudes and all phases of flight - climb, cruise, descent...and, as you have suggested...there is no defining of "cruise", or "above FL200", etc etc.

So I thought why not base the primary "bifurcation" or decision point on phase of flight rather than the very subjective and individual assessment, which defines the "safe conduct of the flight"?

Clearly, the safe conduct of the flight is impacted at/during the takeoff phase...the closer to the ground, the higher the risk. This is a direct result of the two loss of airspeed & altitude information accidents we are now familiar with, the Birgenair and Aeroperu B757 accidents, (which, for others, please see).

So immediate pitch and power numbers need to be memorized and instantly applied until "at/above MSA or Circuit Altitude" and then the aircraft is to be leveled off for troubleshooting. Leveling off includeds selecting the GPS Altitude on the FMC/MCDU, and getting out the QRH pitch and power tables.

In the memorized drill, there are three sub-decision points depending upon where the airplane is in the takeoff/early climbout phase.

For information purposes, the Airbus thrust reduction altitude in force at the time was 1500ft AGL. I believe it has since been changed to 1000ft AGL, but no matter.

The three conditions of the memorized portion of the drill (and after the decision is made that the "safe conduct of the flight is impacted") state:

Pitch/Thrust:
- Below THRUST REDuction ALTitude..................15°/TOGA
- Above THRUST RED ALT and Below FL100........10°/CLB
- Above THRUST RED ALT and Above FL100.........5°/CLB

The FCTMs I've seen further confuse the issue by saying first of all, if you've chosen 5deg of pitch because you're above FL100, (thinking the "safe conduct of the flight is impacted"), then very quickly get out the tables to ensure speed stability, (specifically, the overspeed sitution is mentioned...nothing is said of the opposite problem...loss of speed if pitched up too much). But then the FCTMs state that if the safe conduct of the flight is NOT impacted, the crew will not apply the memorized items but get out the QRH pitch and power tables while maintaining stable flight.

So the potential for applying the wrong procedure is there, but the potential is largely molified by the actions described in the QRH and as such reduce risk providing the crew gets out the tables quickly and establishes stable, level flight.

The difficulty comes when the subjective assessment of the "safe conduct" of flight is made...some would consider the loss of all speed indications a clear and present danger to the flight and launch into the drill. Others would light up a pipe, (metaphorically speaking...to indicate a calm, measured approach! :bored: ) and maintain level flight while the other pilot got out the numbers. Why not base the decision for subsequent actions on the phase of flight, where there is an emergency, and where there probably isn't, (as in cruise flight)?

It made sense to divide the phases simply...into "Takeoff or Below MSA" and "Above MSA in Climb/Cruise/Descent.

For the former, to the usual memorized items. For the latter, level off for troubleshooting, set the GPS Altitude and groundspeed on the MCDUs and get out the QRH pitch and thrust tables.

[Machinbird]

Machinebird
Sorry for the delay in responding but I would submit that what you were writing about is the perception not the actuality. Some commentators are adopting a them and us approach - pilots and airline companies. It more suggests to me a misunderstanding of just what it takes to run an airline than an actual perception of reality. It also ties in with a common human perception that things were better in the past whether it is true or not. It may indeed be that some things were better in the past but it is also true that some things weren't.
When you actually study some of the older accidents (and you don't even need to go back to the 70s to see this) you realise that there have always been pilots who have lacked the necessary skills. What automation has done is allowed a massive increase in the number of flights with a concomitant increase in the level of safety needed to sustain this. Of course you will find some negatives in the introduction of automation and the way it is used but this is due to the humans involved NOT the machines. It is clear that the skills of this particular crew played the significant role in the accident but this is not a general malaise as this accident is unique and other crews have dealt successfully with similar incidents. This alone should be sufficient to demonstrate that the accident is not indicative of a general problem.
OC, the delay was expected due to time zones.
There has been a major de-skilling of the airline pilot community by virtue of automation. The new guys coming up are very good in handling the automation and there are apparently an increasing number of them who when asked to hand fly an aircraft break out into a cold sweat. The environment makes it difficult to acquire and maintain essential hand flying skills. The periodic simulator training sessions are too infrequent to really maintain hand flying skills. Many of the formerly accomplished hand flyers have commented on their personal loss of the touch. There is no doubt that automation has permitted a high level of safety despite this apparent loss of skills, but when an aircraft loses critical systems and the automation is crippled, are these new pilots ready to take over and fly?

When I was actively flying, such a loss of control as AF447 experienced for the reason it lost control would be unthinkable. The weakest pilot in my squadron could fly solid instruments by hand (Where we sometimes saw problems was in headwork.)
With 32 UAS events and 1 loss of aircraft, the statistics for that condition are terrible. Is it a statistical fluke? I don't think so.Statistics doesn't work like that. UAS is clearly much more hazardous than ordinary flight.

IMHO AF447 could well represent the "canary in the coal mine" warning us that the hand flying deterioration has begun to cripple not only the third world airlines but also the legacy carriers.

Any airline pilot should be able to fly cruise by hand, cold without a warmup. If he cannot do that simple task, then he really doesn't belong in the cockpit.

[Old Carthusian]

Machinbird
I am still not convinced by this deskilling you write about. So far what seems to have been discussed is a perception that this is the case which may be widely shared but is just that - a perception. A lot of people saying something is so does not necessarily make it so and given that accidents due to pilot error have remained fairly constant with the massive increase in flights across the globe the statistics would not seem to back this up. I have no doubt that there are pilots who react as you say to the challenge of hand flying but whether this is a general case I am less sure.
I suppose what I am saying is that one must be careful not to over-generalise and to fall into the human error of thinking that things were better in our younger days. Military pilots were notorious (and probably still are) for their high crash rates but this is the price you pay for pushing the envelope as it were. The type of accident represented by AF447 may well have been impossible in your squadron but possibly your colleagues found different ways to have accidents? Error does not manifest itself in a consistent pattern. Where I do think you have a point is that 32 UAS incidents do indicate an issue that does need attention.

[gums]

'bird's last comment is also what scares me.

Any airline pilot should be able to fly cruise by hand, cold without a warmup. If he cannot do that simple task, then he really doesn't belong in the cockpit.

I fully realize that many of the heavy pilots here have not flown to the edge of the "envelope", or exceeded it. I have no problem with that.

What scares me is a basic lack of airmanship I would expect of a teenager that I was helping learn to fly in a Cessna.

Lost a friend at Cali back in 95 or 96 or...... Stoopid flight management system turned the jet the wrong way and they noticed the error but kept descending whle turing back to the approach fix. Not good.

The increased use of automation seems to be a significant factor in recent incidents.

I would prefer a crew that flies the plane using a combination of "auto" aids and manual kills. IMHO, not enough manual flying these days. I am not a dinosaur. Flew with the latest and greatest avionics and FBW and such since the 70's. I would simply hope that the folks up front in the cockpit can actually fly the plane when all the automation and computer stuff goes away. In short, I want a "pilot" up front.

[HazelNuts39]

I still don't agree with PJ that I couldn't recover in 10,000 or 15,000 feet. Give me a chance in the sim and I'll do my best. PJ2's sim pitched down with "full forward stick applied and held (...) from an AoA of 40deg to 10deg (in) 24 seconds", i.e. at an agonizingly slow rate of 1.25 deg/second. How are you going to improve on that? Using manual trim might help, but see Dozy's post #1096. Pulling the thrust levers back to idle would help to increase the ND pitching moment, but that's about all I can think of.

In this initial phase of recovery the sim is performing outside the envelope of known aerodynamic data from flight test or windtunnel. It is possible that actual airplane performs better. On page 111, at 02:12:45 in TOGA and with the THS at -13.8 deg, the airplane pitches down in response to a short push on the stick at 2.7 deg/second, and again at 02:13:30 at CLB thrust.

[rudderrudderrat]

IMHO, not enough manual flying these days.
I agree.
I would simply hope that the folks up front in the cockpit can actually fly the plane when all the automation and computer stuff goes away.
They can in Normal Law - no problem.

On a non FBW aircraft, in the cruise we'd take the autopilot out occasionally, enjoy the feel of the aircraft's response, re-trim it then re-engage the AP.
On AI FBW, there is no requirement to re-trim - it's done automatically. Even if you take the AP out - it's still in Normal Law and handles differently to ALT Law.

There is never an opportunity to practice Alt Law flying skills.
I think the lack of opportunity to practice manual flying skills in Alt Law needs to be addressed in recurrent sim training opportunities..

[PJ2]

HN39;

Re, "Pulling the thrust levers back to idle would help to increase the ND pitching moment, but that's about all I can think of."

The TLs were in the CLB detent in the exercise cited, so yes, that would have helped in the speed with which the AoA reduced.

And yes, recovery from AoA's of around 16° (just before the apogee but well into the stall warning) was a lot quicker - between 10k and 15k with more altitude required for the pull-out once unstalled, (secondary stall occurred a few times when we pulled too hard). The exercise described was a recovery from a fully developed stall, (AoA > 40°).

[RR_NDB]

Hi,

gums and OK465

Stoopid flight management system turned the jet the wrong way... (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-61.html#post7114157)

AA flt 965 almost cleared the hill by crew action (based on GPWS). Crew errors, (http://www.sozogaku.com/fkd/en/cfen/CA1000293.html) data base deficiencies, etc. played a role.

Clever (http://en.wikipedia.org/wiki/TAWS) System was developed and introduced "in the aftermath" of this accident.


PS

The combination of errors was reflected in the percentages:

In June 2000, the jury found that Jeppesen was 30 percent at fault for the crash, Honeywell was 10 percent at fault, and American Airlines was 60 percent at fault. (http://en.wikipedia.org/wiki/American_Airlines_Flight_965)

[PJ2]

Hi gums;

Re, "Stoopid flight management system turned the jet the wrong way "

IIRC, towards the first selection in the NDB list, "R" which was way east, and not the approach "R" (Rozo) NDB to which they had been cleared. In the B767/B757 one has to execute the selection while in the Airbus FMC once selected the airplane is going to go there. I always thought that was not a good design precisely because of this accident but it's still that way.

Painful I know - I recall the Dec 20 '95 accident well and know you lost a friend. You can take some solace in the many positive outcomes regarding CRM, SOPs, automation behaviours/procedures but especially Don Bateman's/Honeywell's work on the improved EGPWS system.

On the recovery altitudes, I think familiarity with the boundaries of controlled flight, which comes with an understanding of aerodynamics of your airplane would help but of course that's not what we do nor are we taught a high level of aerodynamics regarding our designs, and, we have no business being there at the boundaries! Not even test pilots actually stall the aircraft anymore as you know.

The exercise was a worst-case - fully developed stall, late recovery attempt, thrust was not idle. As I mention to HN39, a smart recovery, (as in brisk forward stick at the first stall warning blip, held fully forward without variation, thrust at idle), can be made which reduces the altitude required. It's still going to take a lot of altitude to a) regain the AoA and b) regain the energy, (due low availability of excess thrust, so its height for energy, initially).

[Lyman]

"In the B767/B757 one has to execute the selection while in the Airbus FMC once selected the airplane is going to go there."

Not always, though I'm sure statistically reliable. I have heard (anecdotally) of the bus abandoning the Flight Path in favor of a turn, caught by pilots who then fly manually to destination. Since I am unwilling to be more specific, note I use the word "anecdotally". These occurrences, upon follow up, were not satisfactorally explained.

Have you heard of this?

On pp9-10 of BEA #3, there is a flight path narrative. On pp29-30-31, graphs indexed to CVR. The latter is more explicit, and one assumes, more reliable.

On the last of these three pages, at the top "Thrust levers to TOGA" (DFDR), there is an interesting flow of data, conversation. With TOGA, the RHS inputs 3/4 stop NoseDown. Strange, eh? The LHS, just prior to RHS "I no longer have control of the a/c" states: "We have the engines....what's happening?" If he was concerned about AoA/STALL, why would he take note of the Thrust at FULL as somehow comforting, instead of a dire threat to LIFT?

So it is strange to entertain that the pilots (including Captain) would be concerned with STALL (high altitude), and be OK with the engines at 100%+ N1?

At the very least, why did BEA choose to truncate the CVR just as the controls went LEFT, the PF had lost control of the ship, and the Captain enters "What are you doing?"

Is it a tease? Obviously there is a reason. What is in there that cannot be seen by the public?

[PJ2]

Lyman;

Re, "I have heard (anecdotally) of the bus abandoning the Flight Path in favor of a turn, caught by pilots who then fly manually to destination."

"Flight Path" is defined as the vertical element while "turn" is obviously the lateral element. VNAV requires LNAV because the descent is defined in terms of altitude and possibly speed constraints which are either part of the STAR arrival or merely slow-down or speed control points such as found at Heathrow or Frankfurt. Put another way, the programmed track between waypoints is necessary before the vertical element (VNAV) can manage the descent to achieve the required constraints.

If radar-vectored off the course programmed, the lateral/vertical elements are no longer managed and become instead selected, (short-term operation), and HDG is selected, the lateral element is removed and the vertical element is logically no longer capable of managing VNAV calculations due absence of waypoint constraints, therefore the vertical reverts to Vertical Speed, or more rarely, FPA, neither of which are 100% suitable but for unimportant reasons. The PF can then choose a more suitable mode such as Open Descent - the Alt Sel on the MCP is reset to the cleared altitude if it was previously set to the lowest altitude on the STAR...some airlines allow this, some don't, "just in case."

The Airbus doesn't abandon anything even if in VNAV it is going to miss a constraint - in such a case it will signal on the PFD the need for speedbrakes.

In short, the scenario you're describing can't logically occur and there are no Tech Bulletins extant describing such odd behaviour so I doubt it occurred at all. Leaving the flight path does leave the FMC-programmed LNAV course and the autoflight lateral mode drops to HDG but that is completely normal behaviour for all these autoflight systems, with minor variations on the theme. If what is meant by "fly manually to destination" is the use of HDG on the MCP and Open Descent then this is the same thing as saying one had to use the steering wheel in one's car to get to one's destination and we all know how risky that is. If one selects a downstream waypoint to which one wishes to "go direct", one selects the waypoint, puts it into the requisite line and the airplane will turn towards the waypoint using the shortest route, left or right. In the Boeing, after selecting and placing the waypoint into the requisite line, one then must execute the selection, leaving the crew one more step to make sure that that is what they want. There is no "maybe" about this.

[RR_NDB]

I would simply hope that the folks up front in the cockpit can actually fly the plane when all the automation and computer stuff goes away. In short, I want a "pilot" up front. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-61.html#post7113208)

I would add:

And assure the man-machine interface will always be able to HELP effectively the crew even when facing (all possible) extreme conditions. Allowing a FAST ("immediate", if possible) understanding of the problem(s) or threat(s).

In order, at least, to allow the very basic: Aviate and Navigate safely.

In a "graceful degradation" environment in order to increase chances ("giving" time) to succeed.

(*) The "effective aircraft" (System + crew) must always have (good) chances to "survive". Redundancy is the Key. "Sully" case may be is a good example on the need of a "pilot" up front. The System suffered a major failure. Compare "tiny ice crystals" with "flock of canadian geese"

Problem seems:

You need a System specialist and a Pilot. "Better" would be, first a Pilot and (last but not least) a System specialist.

The "microprocessor fired" the FE's. Is the automation (Super Systems) threatening the "pilot side" of the guys up front? :confused:

[CONF iture]

the AB Chief Test Pilot also seems to talk about "deterrent buffet"
His words :
"The buffeting is a fairly low frequency vibration of the whole airplane, it shakes. It is described per the English term : it is deterrent, which means it is frightening. A sign that all pilots must identify to let them know : Do not go further.
it is impossible that they did not perceived it. You cannot miss it. It shakes badly."

At FL245 the stall warning stopped 40 seconds after it began, the AoA was 10degND, M0.658, VSI 7000fpm down, CAS 278kts.
ref this post (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-post7109943.html)
You most probably didn't mean ND for the AoA.
Isn't Stall Warning supposed to be lost with all ADRs selected off ... ?
Pitch slowly reduced to about 10degND still with full forward stick. As it was held the THS unwound and returned to normal settings.
In my experiment (http://www.pprune.org/tech-log/460625-af-447-thread-no-6-a-81.html#post6786691) THS never did ... ?

In the approach to the stall, the SOP is to apply TOGA power, lower the nose and minimize altitude loss. This still applies!
Not any more.
At the initial aural stall warning, the airplane is still flying and considering as approaching the stall. The procedure is now clearly to lower the nose first and even to have to reduce the thrust in case of lack of pitch down authority.
The initial TOGA action has been banned. Minimizing altitude lost is not the priority anymore, reducing the AoA is.

[PJ2]

Quote:
Originally Posted by PJ2
Thirty seconds after the first Stall Warning passing through FL270 the AoA was 10degND
ref this post (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-post7109943.html)
You most probably didn't mean ND for the AoA.
Isn't Stall Warning supposed to be lost with all ADRs selected off ... ?

Wow..."10degND"! Shoulda caught it, thanks.
Quote:
Pitch slowly reduced to about 10degND still with full forward stick. As it was held the THS unwound and returned to normal settings.
In my experiment (http://www.pprune.org/tech-log/460625-af-447-thread-no-6-a-81.html#post6786691) THS never did ... ?

Hm, I can't explain that without more information. Ours unwound normally and one could feel it in the response of the airplane.
Quote:
In the approach to the stall, the SOP is to apply TOGA power, lower the nose and minimize altitude loss. This still applies!
Not any more.
At the initial aural stall warning, the airplane is still flying and considering as approaching the stall. The procedure is now clearly to lower the nose first and even to have to reduce the thrust in case of lack of pitch down authority.

The initial TOGA action has been banned. Minimizing altitude lost is not the priority anymore, reducing the AoA is.

Yes, I do recall the change now - what I was trying to do was separate two phases here - the "approach to the stall", and the "full stall" obviously and had referred to publications before the change. Kind thanks once again for the correction.
Have you any thoughts on the altitude required to recover from the full stall? What was your experience in your own experiment?

[HazelNuts39]

CONF iture;

Strange, my version of PJ2's post reads: At FL245 the stall warning stopped 40 seconds after it began, the AoA was 10degND, M0.658, VSI 7000fpm down, CAS 278kts.The other strange thing about it is that per S/W schedule (the ADR's were turned back on after the stall) the warning should stop at 6.1 deg AoA for M.658. OTOH the flight path angle for the airspeed and VSI is 9.8 deg down.

[Lyman]

I don't know about the S/W system, but I do know, per BEA, that the A/S bug for S/W is absent from the speed tape when all ADRs are AWOL.

"..this marker disappears..."

HazelNuts39...."Another strange thing..." Keep them coming, I relish strange things, especially if they are not crew actions.... :ok:

[PJ2]

Thanks HN39...the FPV symbol was 10deg below the line dividing blue and brown on the PFD, which to me indicated an AoA of roughy +10deg while the aircraft was still descending rapidly, so it was my clumsy writing even though I think the meaning came through for most others. On the stall warning, in fact the first exercise there was no such warning because we had all the ADRs off, then on for the FPV later - we did it again using a different failure method to get the sim into Alt 2...these are details behind the larger exercise but technically correct - my posts are almost always too long as it is - PJ2

[HazelNuts39]

PJ2;
Thanks for contributing your valuable exercise to the discussion. I was wondering: Do you have to rely on your notes and memory, or does the sim have a facility for recording the flight, e.g. for a post-flight debrief between instructor and pupil?

[PJ2]

HN39;

First, I debated long and hard on a decision to contribute the sim experience at all, and not for the obvious reasons of potential misunderstanding etc,...it was a personal exercise with colleagues who in one case are equally interested and who are in a good position to teach others who fly this aircraft and in another case those who for other reasons are interested in the outcomes. But the more I read from the current discussion the greater some contribution seemed to be of value especially since at least two other experiences have been noted. I wanted to give a sense of the event from initial loss of airspeed through the pitch-up/apogee and holding the stick back so the stall was truly "set", and then the recovery characteristics. It is by no means definitive but I think it is sufficient.

The sim has the ability to record its traces. However, arrangements like these are almost always ad-hoc and time for preparation is at a premium with training schedule demands and hours of availability and coordinating with the technicians to obtain materials after is for various reasons an uncertain process.

There isn't time for note-taking in the moment but other recording methods were used. Still photos from an earlier session were helpful but are not complete nor do they give a sense of situation. Videoing is more helpful but doing it well requires close coordination and ideal circumstances; - covering all instruments and in a timely/relevant way is a challenge!

In my non-engineering view, the sim "behaved" as I expected the airplane might. I don't have experience with the Cooper-Harper scale although I am familiar with it and my own characterizations of the airplane were probably generous. The warning and cautions which accompanied the CAS and ADR failures were less distracting than I had anticipated. The loss of two hydraulic systems or the Emergency Electrical configuration is vastly more complex, demanding and challenging for a two-pilot crew. I don't know how to post audio files, (except through links to YouTube - SoundCloud doesn't provide links) just to demonstrate what the Stall warning sounds like but it was hard to ignore. Also, it was not possible to set aside hindsight so it was not possible to be "confused" by the various behaviours.

[CONF iture]

Strange, my version of PJ2's post reads:
You are correct, my paste was erroneous, I will modify it, thanks.

Have you any thoughts on the altitude required to recover from the full stall? What was your experience in your own experiment?
I can't remember precisely but your 20000 feet figure would be my number too.
I was certainly impressed by the amplitude of negative pitch required to silent the stall warning. Not much blue sky left on the PFD. We, airline pilots, are simply not used to command such attitudes.

[PJ2]

CONF iture, Re, "I was certainly impressed by the amplitude of negative pitch required to silent the stall warning. Not much blue sky left on the PFD. We, airline pilots, are simply not used to command such attitudes. "

No kidding...not much blue left at all. And the AoA crept up - it didn't just change right away. The last bit from around 20deg to 10 seemed to go a bit faster.

[HazelNuts39]

PJ2;

In AF447 pressure errors developed in the pitot/static system as the AoA exceeded approx. 25 deg, causing the IAS to drop rapidly, to below 60 kt at about 41 deg AoA. I gather that AoA in your exercises did not exceed 40 deg enough to silence the stall warning, but did you by any chance observe a rapid drop in IAS between 25 and 40 deg AoA?

[Tommytoyz]

Upset Recovery in Sims | AVIATION WEEK (http://www.aviationweek.com/aw/generic/story_generic.jsp?channel=bca&id=news/bca/2012/04/01/BC_04_01_2012_p34-432588.xml&headline=Upset%20Recovery%20in%20Sims)

I do think this article touches on very interesting observations that may have direct or indirect association with AF447 crew behavior. I will let the article speak for itself.

[gums]

@ PJ and 'nuts

Yeah the long response bothers/surprises me, too.

Knowing what we do now, I guess we could "cheat" and crank that THS to the other end to help the pitch moment. Unlike the Viper, the AF jet had a decent cee gee at the time of the incident.

Seems that Doze also tried a recovery in the sim, that right Doze? We need a refresher of your findings.

@RR

I sure hope the human factors is given attention in the findings and recommendations. I still fault the crew, but I empathize with them to some extent due to the warnings and displays they dealt with.

My civilian job was designing and implementing crew displays, and secondly weapon control algorithms to be implemented in part by the crew and mostly by the confusers. Think JDAM, AMRAAM, Maverick, Hellfire, Harpoon, etc.

I also think a HUD would have helped immensely, as the FPM would have been at the bottom of the display even tho the pitch attitude was nose high. Being a HUD-a-holic since 1971, I simply loved that display. Being far-sighted, I didn't have to squint to read the steam gauges on a serious instrument approach.

[bobdazzle]

Airbus is a great NEW technology in the hands of a OLD time pilots, pilots who were not scared of turning the autopilot off very unlike today’s (now everyone can fly) type of pilots who are ok as system/ cockpit managers but not aviators (sky managers) as used to be the case right up until fifteen years ago. The AF 447 tragedy is one proof of the point i’m making. Airline managements /accountants don’t understand this fact and are slowly but surely turning airline flying into a scary proposition. Command rejects from one Airline are finding Captains jobs in smaller budget airlines and are making flying dangerous. The cruise captain concept is flawed too. The un-stallable airplanes are only as good in the hands of well trained pilots with the aptitude for flying not so in the hands of (now everyone can fly) kind of pilots. Like in the old days if you could not handle the throttle and stick effectively nobody would allow you to even look at the left seat. CRM courses have become boring mumbo jumbo and just eyewash. AF 447 was avoidable. Senior pilots of the major airlines (only god can save the smaller operators) have a responsibility in so much as to warn their managements that flying is becoming safer only in theory. Cockpit managers are good until the bells start ringing in the cockpit when only pilots with the aptitude for throttle /stick may save the day.
Many airline managements are counting on the *hope* that tragedies such as these don't happen to them.
:= As we know the word *hope* does not exist in the flight safety manual.

[RR_NDB]

Hi,

bobdazzle:

Airline managements /accountants don’t understand this fact and are slowly but surely turning airline flying into a scary proposition. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-62.html#post7114977)

The "WHY" will led to "BECAUSE" (http://en.wikipedia.org/wiki/Why–because_analysis).

AF 447 was avoidable

Very early (after recorders read) BEA told plane operated as designed.

The creation of Human Factors study group is emblematic.

IMHO AF447 could well represent the "canary in the coal mine" warning us that the hand flying deterioration has begun to cripple not only the third world airlines but also the legacy carriers. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-61.html#post7113192)

Let's hope findings could identify trends many are concerned.

OC

Your position imo is not conflicting. Will comment ASAP on that.

Safety is increasing. Due many reasons. Automation is an important one.

[RR_NDB]

Hi,

After analyzing a paper from June 2009 (below) on UAS
(http://aviationtroubleshooting.********.com.br/2009/06/af447-unreliable-speed-by-joelle-barthe.html)
I have some questions:

1) Is it dificult to (fast) detect an UAS? Is it important?

Unreliable speed is one of the difficul situations that a pilot has to face. Once the failure has been identified, a procedure, based on pitch angles and thrst settings, will assist the pilot in safely flying the aircraft.

But the main difficulty is to rapidly detect an unreliable speed situation. Reaction time is crucial, since the aircraft may stall and overspeed conditions could cause aircraft damage.2) Is it important to detect? Why?

An unreliable speed situatio may be difficult to identify, due to the multiple scenarios that can lead to it. Therefore, training is a key element: indeed the flight crew's ability to rapid detected the abnormal situation, and to correctely handle it, is cricial.

In case of any doubt, the pilot should apply the pitch/thrust memory items, and then refer to the QRH to safely fly the aircraft, and to positively determine the faulty source(s) before eliminating it (them).
3) AF447 crew (PF or PM) identified the UAS?

4) If no, this (fast identification) would be important? (safety could be at risk?)


The quoted text comes indeed from an Airbus SAS Engineering "source"?

Link is:

1st segment: http://aviationtroubleshooting.com.br

2nd: /2009/06/af447-unreliable-speed-by-joelle-barthe.html

Please copy and paste both segments in your browser (without space, check this) The "editing tool" (that ALWAYS inserts the link), sometimes change it causing error in the resulting URL.

(It seems there is a glitch here in PPRuNe "editing" tool.)

Will check.

[PJ2]

RR, re your post #1234 and your comments, "3) AF447 crew (PF or PM) identified the UAS? , 4) If no, this (fast identification) would be important? (safety could be at risk?) ", if I may, I would like to refer to you my response to Turbine D(Post 1208 (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-61.html#post7113179)[/B], where identification and action of a UAS event is discussed from a human factors point of view. I've read the Barthes paper and others on UAS, (the link doesn't seem to work).

[Linktrained]

259 Avro Yorks were built, an " un-automated" aircraft, hardly dual-controlled. Between 1942 and 1969 a total of 87 " hull losses" were recorded. (I flew 2000+ hours on type. ( No hull losses !))

Aviation HAD to get safer.

Among the airliner stalls, Aero Mexico DC10 -30 lost 11,000ft. on initial climb before continuing. ( The A/P had been set for a rate of climb not possible when heavy and high.)

W. Caribbean MD82 lacked CRM, (the A/P tried to maintain cruising level when de-icing was selected, reducing thrust. The newish F/O was ignored.)

Reading more about China Air B747SP, I see that their Captains are not allowed to let First Officers do T/Os and landings until the Captain has 1000 hour in command on type. To keep the First Officers in practice extra simulator sessions are given to them, monthly. Is this a good idea, and if so is it done in the West, too ?

( My last session in a Link Trainer was arranged so that I did NDB approaches at two different Middle East airfields with similar patterns. Boxes were ticked, even then!)

[CONF iture]

Interesting article Tommytoyz. Thanks.
Flight simulators “are ‘virtual aircraft’ and they should not be used to develop techniques at the edges of the flight envelope,” Cautions Capt. Wainwright. Rather, he advises, “Concentrate everyone’s attention on taking action early enough to prevent the occurrence of loss of control.” That advice is echoed by other studies.

[PJ2]

Yes Tommytoyz, a good discussion on this under-appreciated fact, thank you.

The question has been raised here before and this is a good contribution to the discussion. Airbus' May 1999 edition of "FAST", (Airbus Technical Digest) (http://www.google.ca/url?sa=t&rct=j&q=&esrc=s&frm=1&source=web&cd=1&ved=0CCMQFjAA&url=http%3A%2F%2Fwww.airbus.com%2Fsupport%2Fpublications%2F% 3FeID%3Ddam_frontend_push%26docID%3D17431&ei=fxt7T4a5HqmoiALSjvUu&usg=AFQjCNG9znzDNZilTLePNBc9_0_SR-U9Sg&sig2=qqy7CbCedJR0tPthr9G4zw) discusses this in an article entitled, "Airplane upset recovery: A test pilot's point of view", p.18, elaborating on this caution.

This said, the next question is, what does the simulator do in such circumstances?

Obviously the sim doesn't just quit working so there is the consideration of degree of fidelity and not a question of fidelity itself. The data up to the approach-to-stall is assumed to be accurate.

It has been suggested that while the behaviour is not based upon post-entry-into-the-stall and actual stall data, there remains a certain level of useful replication by extension. The real question is, how much and how useful? The best advice from the AWST article is to train well to recognize and stay out of such extreme circumstances - good advice!, but we have had a number of full-stall accidents over the past seven years or so.

I am (was) a line pilot and am not a test pilot or an engineer so do not wish to tread in these areas without due care, but what seems required for recovery from a full stall in terms of pitch attitude and unstalling the wing from AoA's as seen here does not seem to differ wildly from technical expectations. Discussion?

[OK465]

Quote from JT:

Like PJ2, I hold the view that a bit of sim time for crew exposure to the improbable may (not necessarily will) provide some benefits.
Going with ‘the only game in town’ reasoning,

Being exposed to entry conditions and the resulting ‘maneuver’ simulation which took 20,000+ feet to negotiate and complete would be inspirational to me…..

It would inspire any pilot to diligently make whatever effort was necessary in the aircraft to avoid ever actually learning the level and extent of simulation fidelity involved.

[PJ2]

HN39;

Re, "I gather that AoA in your exercises did not exceed 40 deg enough to silence the stall warning, but did you by any chance observe a rapid drop in IAS between 25 and 40 deg AoA? "

I gathered from the session that it's difficult to simulate the loss of pitot/static information and get the exact same failures/ECAM messages and the method that got closest was failing ADRs and then re-instituting them to some degree. It didn't affect sim behaviour but it does affect available indications. This is a long way of explaining that I can't answer your question with any meaningful information. On one stall we had no warning which was because all ADRs were still failed and on others we had full FPV indications...a bit ad hoc in this area.

[RetiredF4]

CONfiture
I can't remember precisely but your 20000 feet figure would be my number too.
I was certainly impressed by the amplitude of negative pitch required to silent the stall warning. Not much blue sky left on the PFD. We, airline pilots, are simply not used to command such attitudes.

PJ2
No kidding...not much blue left at all. And the AoA crept up - it didn't just change right away. The last bit from around 20deg to 10 seemed to go a bit faster.

gums
@ PJ and 'nuts
Yeah the long response bothers/surprises me, too.



I tried to explain, why that would be the case in my post 386 out of own expierience with stalls in fighter and trainer aircraft.

http://www.pprune.org/tech-log/466259-af447-final-crew-conversation-thread-no-1-a-20.html#post6769341

Thank´s all for the very good discussion points of the last few weeks, very open minded and not much noise.

I stumbled on a video from an F4 in an out of control situation ending in a flat spin, where the aerodynamic forces asociated with high AOA are visualized by contrails (all in the first minutes of the video).

Ejection Decision - A second Too Late! (1981) - YouTube (http://www.youtube.com/watch?NR=1&v=Aa1Ba_NEobs&feature=fvwp)

[Mr Optimistic]

PJ2 - thanks again. Sorry about all this: was the answer 28000 feet ?

[PJ2]

Mr. Optimistic

Re your earlier question about when the captain returned was 20k of altitude sufficient?

The captain returned when the aircraft was in descent passing FL355 just as the stall warning stopped due to "NCD" from the AoA vanes and the VS was about -10,000fpm. The sim exercises showed that recovery could occur with aggressive ND sidestick, held until the airplane began flying again and that took between 18,000ft and 22,000ft depending upon the exercise and when recovery actions began, (SS moved from NU to ND, etc). (that is a 4000ft difference but bear in mind at these descent rates that is also about 12 seconds.

The answer depends more upon the actions taken than it does on the airplane once recovery actions are aggressive and in place. I don't think part ND SS would do it in 20,000ft and it may not do it at all from an AoA of 40deg.

[RR_NDB]

Hi,

Considering what was learned by operators and pilots until now:

Is it probable a similar case (UAS, cruise) where crew "don't understand"?

What kind of man-machine interface enhancement could be useful to allow immediate crew "understanding" of the issue (UAS), in order to go DIRECTLY to the "fix"? (UAS drill, whatever)
I'm assuming crew never identified the UAS. This (lack of understanding) subsequently being aggravated by inadequate PF SS handling and wrong perception of speed, misleading (PF) persistently to try reduce it (350 to apogee). Always increasing (continuously) the "lack of understanding" of PF/PM then Captain, by several consequences of initial errors.

[mm43]

Originally posted by RR_NDB ...
I'm assuming crew never identified the UAS.Well, they did. Or at least the PNF announced the loss of speeds and later the change to ALT Law.
From IR#3 .. At 2 h 10 min 10, the PF's nose-up inputs increased the angle of attack and the stall warning triggered twice transitorily. Probably in reaction to this warning, the PNF exclaimed "what is that?". The PF then said "We haven't got good … We haven't got a good display … of speed" and the PNF "We've lost the speeds". The angle of attack recorded was around 5°, for a theoretical stall warning threshold trigger value of slightly over 4°.
The crew identified the loss of the speed displays but neither of the two copilots called out the associated procedure. The "Unreliable IAS" emergency manoeuvre requires as a first step to disconnect the automatic flight controls and disengage the Flight Directors. The two copilots had only been trained for the emergency manoeuvre at lower levels, in the course of which the pitch attitude to adopt is 10° or 15°.Page 54/55 of BEA Interim Report No.3 provides detail on what should have been done under the heading:-

1.17.4 Air France crew operational instructions

The real question should be why the PF believed he could manoeuvre the A/C at FL350+ the same way he had been trained to do at below FL100. Ultimately what was done right/wrong and by whom boils down to bad CRM associated with as yet undetermined Human Factors. I do find it "a bit of a stretch" to blame the systems/human interface when many similar situations have been successfully handled by other crews. All this crew were required to do is detailed in AF SOPs.

[Machinbird]

The real question should be why the PF believed he could manoeuvre the A/C at FL350+ the same way he had been trained to do at below FL100. Ultimately what was done right/wrong and by whom boils down to bad CRM associated with as yet undetermined Human Factors. I do find it "a bit of a stretch" to blame the systems/human interface when many similar situations have been successfully handled by other crews. All this crew were required to do is detailed in AF SOPs. It is all well and fine to put something down in a written SOP, but if you do not test people to verify their understanding of the SOP, then you are asking for major violations of the SOP.

Since PF & PNF had apparently not had a chance to demonstrate flight in cruise in ALT2 law under UAS conditions, they were open to uncorrected misunderstandings.

If the regular sim sessions were not long enough to exercise this capability, then they were not long enough. Any misunderstanding should have been found and fixed. That falls right on the bean counter's and upper management's heads.

[Organfreak]

What kind of man-machine interface enhancement could be useful to allow immediate crew "understanding" of the issue (UAS), in order to go DIRECTLY to the "fix"? (UAS drill, whatever)

How about another aural warning? "Unreliable airspeed! Go to memory items!" I realize that this crew was already overloaded with warnings, but if the airplane "knows" it's in UAS, there's no excuse for it not to use every means to notify the pilots.

[RR_NDB]

...when many similar situations have been successfully handled by other crews. (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-63.html#post7116955)

Statistically speaking ONE case of COMPLETE lack of understanding after trigger (3 Pitot's then UAS) among other < 40 cases reccommend attention to EVERYTHING. (HF, man machine interface, etc.). We must remember similar conditions (no visual references, turbulence, etc.) could occur again.

My point is:

1) Considering "frequency" of UAS cases traced to the use of now obsolete Thales probes.

2) Considering some cases were related to the use of competitor probes, currently and still in use.

3) Considering they became "rapidly lost" and "increasingly confused"

4) Considering Pitot's data (simultaneous failure) are still today not adequately processed (NO REDUNDANCY AT ALL) by System

5) Considering identification of UAS presently has to be done by scan and feeling (as linked paper put)

6) Consider PF was caught in surprise and acted almost immediately (by "lack of speeds" as you observed)

I ask:

Why not to provide an instantaneous (even non causal, before an eventual Law change) indication to the crews (Airbus SAS, Boeing, Embraer, etc.) (PRECISELY) of UAS?

Complementing: The point is, everything SIMPLE and PRECISE the System (through the man machine interface) provide to the crew, if possible (UAS is) IMO increases the safety of the plane operation.

Surprises (coming from her :} ) better to avoid.

The mentioned paper (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a-62.html#post7115102) (i am considering still valid and "official") led me to this point.

AF447 case was not "triggered just by "tiny ice crystals" affecting "obsolete AS probes" in a System operating without redundancy (Air data).

AF447 case was triggered (it seems to me) by a "lack of understanding" of what was really happening. UAS as everyone know, causes erratic indication in several indicators as mentioned in the linked paper. The Pitot data (considered GARBAGE by the System designers) during a transient phase are still fed to pilots (they receive this "input") that need to rush the scan in order to just identify UAS. As the paper points.

If the System tells you (on UAS) you save time (precious) and could act MUCH more precisely and fast.

IMO, the "interface" should be reviewed. Why? Because in my opinion can be easily improved with good cost benefit relationship.

BUSS could still be an optional. (STD in 380);

An UAS indicator could be Standard. At least before we solve the Pitot's issue.

I do find it "a bit of a stretch" to blame the systemsEvolution (stepwise) is not based in "blaming". Sometimes "a tweak" (as you know) "makes a difference"

Ultimately what was done right/wrong and by whom boils down to bad CRM associated with as yet undetermined Human Factors.

I agree. But remember:

CRM and HF "becomes critical" under extreme difficult conditions. In Portland the crew had time to digest the surprise (gear problem). F-GZCP had few seconds and apparently never understood and realized the "benign" problem: Simply an UAS. That shortly "dissipated". :E

[RR_NDB]

Hi,

Since PF & PNF had apparently not had a chance to demonstrate flight in cruise in ALT2 law under UAS conditions, they were open to uncorrected misunderstandings.
Open to everything specially in EXTREME conditions.

but if the airplane "knows" it's in UAS, there's no excuse for it not to use every means to notify the pilots. The information (Pitot's "garbage out") is available. The System (a subsystem) could easily process and present the results. How? This must be studied. The interface must be always kept simple. K.I.S.S. principle is mandatory to the interface characteristics.

PS

HF study could be extended to upper management, bean counter's and System designers. :} .

In order to understand why they fail. :mad: And to improve man machine interface.

[DozyWannabe]

Seems that Doze also tried a recovery in the sim, that right Doze? We need a refresher of your findings.

They're on the last pages of the previous thread and the first pages of this one.

[Turbine D]

I wanted to make it a little easier to see each of the SIM experiences, so I put them all together in this post. Each experience is followed by answers to questions by other posters. I didn't post the questions (saving some space) but they should be evident from the answers. Hope this helps....

From CONF iture:

As we got little extra time in the sim today, we did experience a full stall from FL350.
Here is what I can report from the experience :
From the STALL warning we kept a light aft pressure on the sidestick
It was not long before we got a negative vertical speed of 15000ft/min
THS went to 12 deg UP under STALL warning
As we decided to exit the stall, full fwd pressure on the sidestick was applied
But we were unable to lower the nose
THS did not move
THS was then manually rolled fwd
Nose came down
Exit was then possible
I can't remember all the details, too much stuff to look at.
Thrust was kept at idle all the time.

Early fwd pressure on the sidestick at initial STALL warning should prevent a stall to develop.

From DozyWannabe:

What we had was an A320 sim rather than an A330, which comes with some key differences - the most obvious of which is the lack of Alternate 2, the nearest equivalent being Alternate without speed stability, and a different underlying architecture past a certain point.

The first experiment involved setting the conditions to night IMC with CBs in the vicinity, having set the autoflight to take us to 35,000ft and hold us there. We had a friend of his who is a TRE sitting in the LHS to provide guidance and monitor what we were doing. He then failed the ADCs, leading to autopilot disconnect and a drop to Alternate (without speed stability) and we tried to follow through and maintain a 15 degree pitch angle. Things we noted:
I'd suspected it would involve considerable effort to hold the sidestick there for a significant amount of time, but I was genuinely surprised at just how much.
The zoom climb occurred exactly the way we expected
The Alternate Law (no speed stability) on the A320 seems to have a hard trim limit of 3 degrees nose up
It was definitely possible to hold the aircraft in the stall with 3 degrees of nose-up trim and full back stick, but it required effort
The aircraft wanted to nose down and recover itself, and with about 10 degrees of nose-down maintained with the sidestick at the moment we passed about 30,000ft, we managed to effect a recovery with the speed coming back up to a point where we could level out safely at about 20-25,000ft judging by the standby altimeter.
The second experiment was the same as the first, but as my pal had noted, the A320 has a hard limit of 3 degrees NU trim available via autotrim in the secondary Alternate Law. We tried again, this time winding in full nose-up trim manually just prior to the point of stall. This time:
The aircraft seemed more willing to hold pitch with the trim at full-up, but to hold it at 15 degrees still required considerable effort
We had to add a touch of rudder (on the TRE's advice) to control the roll.
Despite full nose-up trim, we elected to start a recovery as we came down through about 35,000ft this time, just to see if it was possible using sidestick only
Following the same 10 degree nose-down sidestick demand as before, the trim rolled forward with the sidestick demand, returning to around neutral within about 5-8 seconds, and we came out of the stall as before.
Based on this, as far as the A320 is concerned at least, recovery is possible using autotrim via sidestick only even when the trim has been manually wound fully nose-up. Given more time we'd have liked to see what happened attempting recovery at lower altitudes, but the general take-away seems to be that with sufficient forward sidestick demand it is possible to recover from stall even with trim forced to where it's not supposed to be.
Further details I've just been reminded of - the stall stabilised at approx 180kts IAS on the sim control with the nose-up trim at 3 degrees (the A320 hard limit). With full nose-up trim the stall was similar, but stabilised at approx. 160kts IAS. The Stall Warning was not only clear, but so loud that the TRE had to cancel it with the Emergency Cancel button in order for us to hear each other. On the second (full nose-up trim) experiment, all I had to do was briefly glance down to my left to see the trim roll forward - smoothly and *very* quickly - following recovery via sidestick pitch down.

Some questions answered:

Having conferred, we loaded extra fuel so that the FMGC showed MAX ALT FL379. C of G was 32% MAC. The ROD in our experiments maxed at approx 6,000ft per minute, with the VSI needle turning amber in the PFD. One of the reasons I hope someone will perform a later experiment will be to see how leaving the recovery till later in the sequence will affect the ROD, and hopefully also find out how a 40% MAC CoG will affect things. The caveat here is that the later you leave it, the further outside the tested flight envelope you go, and the more divergent the sim's performance from the real thing will become.
After the initial NU pitch increase (induced with approximately half back-stick, as in the DFDR traces), we triggered a very short "G" induced stall warning as we climbed, then when the real warning sounded continuously (as happened in the AF447 scenario) we applied TOGA and held 10 to 15 degs pitch on the sidestick - during which full deflection was required in order to come close to maintaining it - as I said, the nose wanted to come down naturally if I released pressure for even a split-second.

Initially, autothrust dropped to Thrust Lock. We pulled the thrust levers back to match the thrust, but as we moved them the thrust increased slightly. The TRE then deliberately staggered the TLs slightly to induce a roll to the right which we trimmed out with rudder and the slightest touch of aileron.

From PJ2:

With weights, CG, SAT mirroring AF447 and a bit of turbulence, following loss of airspeed (all 3 ADRs out), the sim was pitched up at FL350 and held in the climb until stalled, (THS reached 13.6deg). Shortly after the stall we returned the ADRs for use during the balance of the exercise, (to see the FPV during the stall).

Post-apogee (approx FL360), full forward stick was applied and held.

At FL330 the pitch was 8deg ND.

At FL310 the AoA (using FPV) was approx 40deg and the VSI was 18000fpm +. Pitch was about 14deg ND which was all the pitch that could be obtained.

Pitch slowly reduced to about 10degND still with full forward stick. As it was held the THS unwound and returned to normal settings.

We could watch the AoA reducing as the FPV slowly climbed "up" the PFD from past the red ND warning arrows below 30deg pitch marks.

Thirty seconds after the first Stall Warning passing through FL270 the AoA had reduced to 30deg, descent rate was 16000fpm.

Ten seconds later at FL255 the AoA was 12deg, CAS was 250kts, VSI was 7400fpm.

At FL245 the stall warning stopped 40 seconds after it began, the AoA was 10degND, M0.658, VSI 7000fpm down, CAS 278kts.

From an AoA of 40deg to 10deg took 24 seconds and about 6000ft. This exercise took about 22000ft; some were less.

Overspeed was never a problem nor was a secondary stall if one was gentle, (took about another 6000ft IIRC)

Some questions answered:
Normal cruise pitch attitude is between 2.3 and 3deg depending mostly upon weight. A pitch up to 5deg pitch attitude (+2.5deg) results in about an 800 to 1500fpm climb and a gradual loss of energy if held long enough. The UAS QRH checklist and the FCTM cautions strongly against holding such pitch attitudes for long and advises to get the QRH out quickly and set pitch and power. The FCTM also states that the Memory Items are not to be done if the immediate safety of the aircraft is not impacted.

For the purposes of the exercise there was no "judging" of how much to pitch up. We pitched up high enough to stall the aircraft. Fifteen degrees would do it, sometimes we were higher.

The overriding impression of these sessions was how quickly things occurred and how fast was the altitude loss.

My comment about "how quickly things unfold" is an observation on the amount of time it took to lose control. Time is always "relative to perception and familiarity". When a pilot is highly trained, highly experienced and very familiar with his/her airplane, the flow of even huge amounts of information and lots of events flow much more slowly, perceptually, because the mind anticipates much more effectively and easily, (depending upon other factors such as distraction, fatigue) than if one is relatively inexperienced in such circumstances.

In terms of this loss of control, while it unfolded literally over a period of about 40 seconds, oddly that is tons of time to do something, but it is not a lot of time for assessment, discussion, action.

In re the sim exercise, power was at various settings (though never TOGA). Also, there were variations in altitude needed for recovery. In one exercise we "did nothing" and the aircraft happily remained in stable flight at FL350 as power remained in "TOGA LOCK" (last setting before disconnect) and the pitch at 2.5deg, approximately.

gums, you're right about the sim not really being able to teach this kind of thing - you can't feel "the mush"...the slight delay in responsiveness in both changes in attitude and flight path even though the subtle indications may be accurately simulated.

The exercise was a worst-case - fully developed stall, late recovery attempt, thrust was not idle. As I mention to HN39, a smart recovery, (as in brisk forward stick at the first stall warning blip, held fully forward without variation, thrust at idle), can be made which reduces the altitude required. It's still going to take a lot of altitude to a) regain the AoA and b) regain the energy, (due low availability of excess thrust, so its height for energy, initially).

Thanks HN39...the FPV symbol was 10deg below the line dividing blue and brown on the PFD, which to me indicated an AoA of roughy +10deg while the aircraft was still descending rapidly, so it was my clumsy writing even though I think the meaning came through for most others. On the stall warning, in fact the first exercise there was no such warning because we had all the ADRs off, then on for the FPV later - we did it again using a different failure method to get the sim into Alt 2...these are details behind the larger exercise but technically correct - my posts are almost always too long as it is.

In my non-engineering view, the sim "behaved" as I expected the airplane might. I don't have experience with the Cooper-Harper scale although I am familiar with it and my own characterizations of the airplane were probably generous. The warning and cautions which accompanied the CAS and ADR failures were less distracting than I had anticipated.

I gathered from the session that it's difficult to simulate the loss of pitot/static information and get the exact same failures/ECAM messages and the method that got closest was failing ADRs and then re-instituting them to some degree. It didn't affect sim behaviour but it does affect available indications. This is a long way of explaining that I can't answer your question with any meaningful information. On one stall we had no warning which was because all ADRs were still failed and on others we had full FPV indications...a bit ad hoc in this area.