AF447 Thread No. 3
‘Human – Technical’ thoughts
Some ‘Human – Technical’ thoughts … … and questions, … mainly questions.
It is widely assumed that the event was triggered by ice particle blockage of the pitot tubes. With all three blocked it is also likely that the TAT probes suffered blockage, probably simultaneously; ref previous events and see VH-EBA (not simultaneous), #692 graphic and report link.
If the ‘sudden failure’ of airspeed (an erroneous low value) was accompanied by an erroneous low altitude (as with VH-EBA), then the indicated altitude deviation could also attract the crew’s attention. Would the altitude alerter sound?
Thus with apparent simultaneous failure of all air data and logical auto-flight disengagement, the crew’s initial thoughts might have been of a system failure – a technical issue.
The PF took the controls, but with the perception above, may have concentrated on altitude. In every day operation there is significant focus on altitude – ‘beware altitude bust’, TCAS, interview, loss of license, etc; these could easily be hidden (subconscious) operational and social pressures which might have influence the initial action; pitch up to recover the assigned altitude, back stick and follow-up trim.
I assume that the trim datum is still attitude (flt path?) and zero control input. Would the control law be attitude or flight-path stable with these system errors. If flight-path, how is flight-path computed, inertial or with air-data mix, and if air-data and the air-data is erroneous … … ?
Would the FD still be active, if so, in which modes and what command might be shown?
We should recall the many discussions of how modern crews depend on the FD; a general low experience level of basic instrument flying. Was an erroneous FD command followed resulting in an inappropriate manoeuvre?
Would an abnormal TAT (rising towards zero, based on previous events, A/B and other aircraft types) affect any other instrument display systems of flight guidance computation?
If airspeed was erroneous, and the altitude is also affected by this and also possibly from ‘abnormal’ temperature, what about Mach, and any residual protections (MMO pitch-up bias), i.e difference between computed Mach (and rate) vs computed MMO.
From what has been describes so far it may be impossible to take any of the quoted air-data values as representing the true condition of the aircraft; except perhaps the vane derived AOA. What source of information does the FDR record, is this always the same as what the crew will see?
Roll deviation might be considered a separate but confusing issue. If the YD ‘froze’ (fail-safe) with air data ‘failure’ (as designed?), then an offset rudder would induce roll and a permanent roll bias, but its effect varying with actual airspeed – need to use rudder. But who uses rudder in normal (symmetric power) flight?
With these distractions and concerns of system failure, then any more speculation about crew thought and behavior would be unjustified without further data.
However, the trim position is a dominant issue, and without awareness of this offset, basic flying could be difficult and probably added further confusion.
At a late stage, nose down control action did effect a pitch change and an apparent speed increase (a semblance of stall recover but with the confusion and aerodynamic complications of mis-set trim). The speed increase was sufficient to reinstate the previously inhibited AOA driven stall warning; probably further confusing the mental picture – nose down pitch apparently caused a ‘stall’ ??? … what next; undo what you have just done, back stick? Perhaps this is normal (to be expected) human behavior.
Also, think about how the industry discusses stall – some thread discussion, education, and training. Virtually every aspect is speed related – stall ‘speed’, warning ‘speed’, stall displayed on a ‘speed’ tape, the importance of gaining / maintaining ‘speed’; even if a crew had been taught that the stall warning was AOA driven and that the aerodynamics of a stall involve AOA, is this information recallable, or would ‘speed’ dominate our thoughts.
There is much more to come, much to be learnt, and probably all intermixed in technical, human, and social system (operational/organizational) interactions.
At this early stage, we might consider how crews are educated / trained, and what is currently expected of them. How might everyday operation influence encounters with a very rare and unusual events? Are modern norms based on a remarkably benign (safe) operation so that it is now unrealistic to expect extremes of experienced-based innovative behavior? Has the industry overlooked ‘change’, and retains old, inappropriate assumptions?
To reassure any PAX Ppruners (and the media), the industry remains very safe. Incidents such as this are very rare and the assumed precursor (high confidence) – ice particle blocking is generally understood. Safety action is in place – changing the type of pitot tube, manufacturers have refreshed crew dills for loss of airspeed and stall recovery, and operators are applying these training aspects. The industry is not (must not be) complacent; a thorough and tenacious investigation is underway from which we expect ‘lessons to be learnt’; but it’s still up to us to learn.
It is widely assumed that the event was triggered by ice particle blockage of the pitot tubes. With all three blocked it is also likely that the TAT probes suffered blockage, probably simultaneously; ref previous events and see VH-EBA (not simultaneous), #692 graphic and report link.
If the ‘sudden failure’ of airspeed (an erroneous low value) was accompanied by an erroneous low altitude (as with VH-EBA), then the indicated altitude deviation could also attract the crew’s attention. Would the altitude alerter sound?
Thus with apparent simultaneous failure of all air data and logical auto-flight disengagement, the crew’s initial thoughts might have been of a system failure – a technical issue.
The PF took the controls, but with the perception above, may have concentrated on altitude. In every day operation there is significant focus on altitude – ‘beware altitude bust’, TCAS, interview, loss of license, etc; these could easily be hidden (subconscious) operational and social pressures which might have influence the initial action; pitch up to recover the assigned altitude, back stick and follow-up trim.
I assume that the trim datum is still attitude (flt path?) and zero control input. Would the control law be attitude or flight-path stable with these system errors. If flight-path, how is flight-path computed, inertial or with air-data mix, and if air-data and the air-data is erroneous … … ?
Would the FD still be active, if so, in which modes and what command might be shown?
We should recall the many discussions of how modern crews depend on the FD; a general low experience level of basic instrument flying. Was an erroneous FD command followed resulting in an inappropriate manoeuvre?
Would an abnormal TAT (rising towards zero, based on previous events, A/B and other aircraft types) affect any other instrument display systems of flight guidance computation?
If airspeed was erroneous, and the altitude is also affected by this and also possibly from ‘abnormal’ temperature, what about Mach, and any residual protections (MMO pitch-up bias), i.e difference between computed Mach (and rate) vs computed MMO.
From what has been describes so far it may be impossible to take any of the quoted air-data values as representing the true condition of the aircraft; except perhaps the vane derived AOA. What source of information does the FDR record, is this always the same as what the crew will see?
Roll deviation might be considered a separate but confusing issue. If the YD ‘froze’ (fail-safe) with air data ‘failure’ (as designed?), then an offset rudder would induce roll and a permanent roll bias, but its effect varying with actual airspeed – need to use rudder. But who uses rudder in normal (symmetric power) flight?
With these distractions and concerns of system failure, then any more speculation about crew thought and behavior would be unjustified without further data.
However, the trim position is a dominant issue, and without awareness of this offset, basic flying could be difficult and probably added further confusion.
At a late stage, nose down control action did effect a pitch change and an apparent speed increase (a semblance of stall recover but with the confusion and aerodynamic complications of mis-set trim). The speed increase was sufficient to reinstate the previously inhibited AOA driven stall warning; probably further confusing the mental picture – nose down pitch apparently caused a ‘stall’ ??? … what next; undo what you have just done, back stick? Perhaps this is normal (to be expected) human behavior.
Also, think about how the industry discusses stall – some thread discussion, education, and training. Virtually every aspect is speed related – stall ‘speed’, warning ‘speed’, stall displayed on a ‘speed’ tape, the importance of gaining / maintaining ‘speed’; even if a crew had been taught that the stall warning was AOA driven and that the aerodynamics of a stall involve AOA, is this information recallable, or would ‘speed’ dominate our thoughts.
There is much more to come, much to be learnt, and probably all intermixed in technical, human, and social system (operational/organizational) interactions.
At this early stage, we might consider how crews are educated / trained, and what is currently expected of them. How might everyday operation influence encounters with a very rare and unusual events? Are modern norms based on a remarkably benign (safe) operation so that it is now unrealistic to expect extremes of experienced-based innovative behavior? Has the industry overlooked ‘change’, and retains old, inappropriate assumptions?
To reassure any PAX Ppruners (and the media), the industry remains very safe. Incidents such as this are very rare and the assumed precursor (high confidence) – ice particle blocking is generally understood. Safety action is in place – changing the type of pitot tube, manufacturers have refreshed crew dills for loss of airspeed and stall recovery, and operators are applying these training aspects. The industry is not (must not be) complacent; a thorough and tenacious investigation is underway from which we expect ‘lessons to be learnt’; but it’s still up to us to learn.
Join Date: Jun 2009
Location: Paris
Posts: 691
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by sensor_validation
Highlights lots of gaps we haven't been told about - but plotting ACARS messages at receipt time only shows when aircraft attitude able to communicate with satellite - not when they were generated.
ACARS sequence/order/time is due to the processing time logic, being queued, and to protocole limitations as nothing (so far) is indicating that she could have reached an attitude that would broke the satellite link: up to +20 deg pitch and +/-40 deg bank angle combined should make any ACARS sending possible every time from the begining to the end.
Now, it is also possible that the satellite could be, at one point, busy with other aircraft sendings (as there is not so many satellites above this place and the total bandwith was not illimited). That is also how I interpreted the BEA quote "that the satellite link could be broken" (due either to aircraft attitude or satellite not available for login).
Join Date: Jan 2002
Location: USA
Posts: 286
Likes: 0
Received 0 Likes
on
0 Posts
Wondering what the proceedure syllabus AF uses for application of TOGA. Would assume that at some point, the airplane would wish to climb, and that the THS would move to nose up.
Begs the question also, is there any system , irrespective of whatever Law may be active, which effects the THS if and when TOGA is enabled?
Begs the question also, is there any system , irrespective of whatever Law may be active, which effects the THS if and when TOGA is enabled?
Join Date: Jul 2002
Location: UK
Posts: 3,093
Likes: 0
Received 0 Likes
on
0 Posts
Join Date: Jun 2009
Location: LSZG
Age: 52
Posts: 107
Likes: 0
Received 0 Likes
on
0 Posts
Actually there are two time stamps. The first on is the sending time stamp. The second one, within the message, is the time of the generation.
Join Date: Jun 2009
Location: I am where I am and that's all where I am.
Posts: 660
Likes: 0
Received 0 Likes
on
0 Posts
HN39 #847: "First three sentences agreed. However at FL375, 215 kCAS, AoA=4° the airplane is not stalled: the normal acceleration is less than 1 (approx. 0.8) hence v/s is decreasing."
That brings us back to spurious stall warnings, doesn't it?
That brings us back to spurious stall warnings, doesn't it?
Join Date: Jun 2009
Location: Paris
Posts: 691
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by MartinM
Actually there are two time stamps. The first on is the sending time stamp. The second one, within the message, is the time of the generation.
Now, the BEA confirms that she crashed in the same second at 0214:28.
Join Date: Jun 2009
Location: Paris
Posts: 691
Likes: 0
Received 0 Likes
on
0 Posts
Hi JD_EE,
Right. This first stall warning seems to be due to a change in stall function (it is mach based above Mach 0.75) to AoA based function. Unreliable airspeed seems to affect everything like static pressure, which is mach sensitive, AoA indication which is mach sensitive... Then when you lose the mach function for a while, this become a real issue. Beside, the stall warning is kicking above VSw1g with some margin for taking into account any possible overshooting.
Originally Posted by JD_EE
That brings us back to spurious stall warnings, doesn't it?
Takata, re the A340 AIRPROX, you wrote:
Hmm, the flight controls were in normal mode and no sidestick controller input was made until 28 sec. after AP diconnection when they were already a thousand feet above their assigned level.
I thought that in normal law, compensation for thrust variation was provided.
Anyway, except for human factor aspects, this incident appears unrelated to AF447, this one being in ALT law.
In this case, the climb (and pitch increase) is related to this sharp thrust increase (4 engines at 100% N1), being not countered by nose down imput from the PF.
I thought that in normal law, compensation for thrust variation was provided.
Anyway, except for human factor aspects, this incident appears unrelated to AF447, this one being in ALT law.
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by JD-EE
That brings us back to spurious stall warnings, doesn't it?
Takata,
the stall warning threshold has to maintain a certain margin below alpha-max which defines the 1g stall speed. For all Mach numbers greater than 0.275 alpha-max reduces with increasing Mach, and therefore also the stall warning threshold AoA. Just take a look at the FCOM page presenting Vs1g.
Last edited by HazelNuts39; 1st Jun 2011 at 06:15.
Amazing amount of material posted in the last few days. I was out of town, and so did not FOD the thread.
A thought on AoA probes and 60 knots.
Aye, but here is a thought: if the mission designed approach and takeoff speeds are typically 150 kts and more, maybe the design criterion can accept them not being as swift as the nautical brand at less than half of the operating speeds?
Two thoughts I see bubbling up, the answers to which are not clear to me.
Is there an unambiguous answer to those nested questions?
What I don't see from the latest BEA narrative is how many attempts to move the stick down, or move the nose down, were in play. Two are mentioned. Was that all?
This related to an earlier observation: a continuing attempt to move the stick down (though they mention some instances of stick movement toward nose down, the event flow is hard to discern) to unstall also runs into temporal distortion and its effects on pilots handling upsets.
Many pages ago, one of our posters pointed out that he was suprised to learn how long it takes for his inputs to unstall one of the big jets. With that in mind, was the crew in AF 447 (if and when they finally figured out "we are stalled!") not aware of how long it takes to get the AoA to start increasing again for the position they were in when "we are stalled and falling" became evident?
We may never know. CVR might tell.
Aside/musing. (Not sure how applicable this is to the mishap underdiscussion).
FWIW, from a pilot, but not an Airbuss 330 pilot.
If I command the nose down or up with the stick, and the nose does not go up or down, there is something horribly wrong, and I am by definition, in a condition known as out of controlled flight.
I am using a very simple definition of out of controlled flight.
You are in the aircraft, you make inputs with the stick and rudder and throttle, and the plane does not respond as you command it to. (Or, as you expect it to). Recovering from out of controlled flight is a skill set that relies on the flight controls working as they are known (or designed) to work in response to your inputs, so you have to get to that state. As an emergency or recovery procedure, out of controlled flight recovery means taking a set of steps that gets you to the point where the flight controls respond to your inputs to go up, down, left, right, etecetera. It often takes time and altitude to get through those steps.
From a pilot and human factors point of view, and the flight regime the cockpit crew found temselves in ...
Given the flight regime they were in, and considering the likely concerns about not overspeeding or going to fast in turbulent air present due to envirnonemt, the crew was in the condition for a surprise upset:
You don't expect to be stalled, which means you have to spend "x" seconds (or fractions thereof) overcoming the denial stage, and getting to the action stage. (Responding to your training and knowledge of your equipment and situation).
That some of the cues were either ambiguous or at odds with previous training looks to have led to counterintuitive responses ... but we don't know what PF was seeing.
Further thought on aerodynamics and degree of stall.
If the answer to the two nested questions is "no visual alert" and "yes, you need to take over manually to get THS to work the nose attitude to your advantage" then there is a possible training improvement to be harvested.
To the question of ineffectiveness of elevators, as aircraft speed decreases, due to V^2 ... the slower you are in the airstream, the more pronounced the effect of THS overwhelming the elevators in pitch control. (I hadn't considered the masking due to high pitch/AoA). Does this make sense, in terms of me understanding that characteristic of the flight controls?
The deeper into stall you go, particularly if THS running most of the way up was a factor, the harder it is to unstall with your stick control.
Maybe my math on that is off, and it washes out and is a linear relationship.
Yes, pitch and power brigade, the basics are important.
A thought on AoA probes and 60 knots.
Originally Posted by a former flying squid
My personal experience with the US Navy peg, cone and vane type AOA sensors is that they would come alive in a definite breeze, perhaps 10 knots.
A vane type that needed 60 knots to be reliable would be a sticky one in my book.
A vane type that needed 60 knots to be reliable would be a sticky one in my book.
Two thoughts I see bubbling up, the answers to which are not clear to me.
Once they were in Abnormal Attitude Law, the Stab (THS) Trim remained where it was last. (In this case full nose up)
From the "flight & laws" link above, if in Direct Law the pilots will see "USE MAN PITCH TRIM" on the PFD. Is that also the case in Abnormal Law?
What I don't see from the latest BEA narrative is how many attempts to move the stick down, or move the nose down, were in play. Two are mentioned. Was that all?
This related to an earlier observation: a continuing attempt to move the stick down (though they mention some instances of stick movement toward nose down, the event flow is hard to discern) to unstall also runs into temporal distortion and its effects on pilots handling upsets.
Many pages ago, one of our posters pointed out that he was suprised to learn how long it takes for his inputs to unstall one of the big jets. With that in mind, was the crew in AF 447 (if and when they finally figured out "we are stalled!") not aware of how long it takes to get the AoA to start increasing again for the position they were in when "we are stalled and falling" became evident?
We may never know. CVR might tell.
Aside/musing. (Not sure how applicable this is to the mishap underdiscussion).
FWIW, from a pilot, but not an Airbuss 330 pilot.
If I command the nose down or up with the stick, and the nose does not go up or down, there is something horribly wrong, and I am by definition, in a condition known as out of controlled flight.
I am using a very simple definition of out of controlled flight.
You are in the aircraft, you make inputs with the stick and rudder and throttle, and the plane does not respond as you command it to. (Or, as you expect it to). Recovering from out of controlled flight is a skill set that relies on the flight controls working as they are known (or designed) to work in response to your inputs, so you have to get to that state. As an emergency or recovery procedure, out of controlled flight recovery means taking a set of steps that gets you to the point where the flight controls respond to your inputs to go up, down, left, right, etecetera. It often takes time and altitude to get through those steps.
From a pilot and human factors point of view, and the flight regime the cockpit crew found temselves in ...
Given the flight regime they were in, and considering the likely concerns about not overspeeding or going to fast in turbulent air present due to envirnonemt, the crew was in the condition for a surprise upset:
You don't expect to be stalled, which means you have to spend "x" seconds (or fractions thereof) overcoming the denial stage, and getting to the action stage. (Responding to your training and knowledge of your equipment and situation).
That some of the cues were either ambiguous or at odds with previous training looks to have led to counterintuitive responses ... but we don't know what PF was seeing.
Further thought on aerodynamics and degree of stall.
If the answer to the two nested questions is "no visual alert" and "yes, you need to take over manually to get THS to work the nose attitude to your advantage" then there is a possible training improvement to be harvested.
To the question of ineffectiveness of elevators, as aircraft speed decreases, due to V^2 ... the slower you are in the airstream, the more pronounced the effect of THS overwhelming the elevators in pitch control. (I hadn't considered the masking due to high pitch/AoA). Does this make sense, in terms of me understanding that characteristic of the flight controls?
The deeper into stall you go, particularly if THS running most of the way up was a factor, the harder it is to unstall with your stick control.
Maybe my math on that is off, and it washes out and is a linear relationship.
Yes, pitch and power brigade, the basics are important.
Thread Starter
Join Date: Jan 2005
Location: W of 30W
Posts: 1,916
Likes: 0
Received 0 Likes
on
0 Posts
data access
Thanks for the reference grizzled.
So is it correct to say that for the time being the data belong to the BEA and the BEA has the liberty to give access to the parties that could provide expertise for the good of the investigation ?
I believe, us, as people of what is called a 'free society' should have access to those data. There is a lot of expertise on the net too ...
Anyway, what would be the downside to publish them ?
So is it correct to say that for the time being the data belong to the BEA and the BEA has the liberty to give access to the parties that could provide expertise for the good of the investigation ?
I believe, us, as people of what is called a 'free society' should have access to those data. There is a lot of expertise on the net too ...
Anyway, what would be the downside to publish them ?
Join Date: Jun 2009
Location: I am where I am and that's all where I am.
Posts: 660
Likes: 0
Received 0 Likes
on
0 Posts
HN39 #894
That is why I used "spurious" rather than false. I thought 215kt and 4 degrees would likely calculate into an imminent stall, if stall warning pays attention to speed at all. (Off hand I'd be surprised if it didn't in my non-pilot naivete.)
That suggests that things broke down with that stall warning at 37500 feet. The PF appears to have made an inappropriate action, full throttle and pull up - the apparently prescribed action to take before a stall. But it never seemed to help but the PF kept the stick back fatally long even as the thrust was, at some time, reduced to idle. It was at idle at 2 h 12 min 02, and TOGA at 2 h 10 min 51. That's all I can extract from the report.
It would also be nice to know how long and how much "up" was asserted at 2 h 10 min 05. That could tell a lot more of the story just in itself.
That is why I used "spurious" rather than false. I thought 215kt and 4 degrees would likely calculate into an imminent stall, if stall warning pays attention to speed at all. (Off hand I'd be surprised if it didn't in my non-pilot naivete.)
That suggests that things broke down with that stall warning at 37500 feet. The PF appears to have made an inappropriate action, full throttle and pull up - the apparently prescribed action to take before a stall. But it never seemed to help but the PF kept the stick back fatally long even as the thrust was, at some time, reduced to idle. It was at idle at 2 h 12 min 02, and TOGA at 2 h 10 min 51. That's all I can extract from the report.
It would also be nice to know how long and how much "up" was asserted at 2 h 10 min 05. That could tell a lot more of the story just in itself.
Join Date: Jun 2009
Location: NNW of Antipodes
Age: 81
Posts: 1,330
Received 0 Likes
on
0 Posts
Graybeard;
You are right, it didn't.
I was aiming more at the ALT erroneous speed correction component, and its rate of change which may have been too much for the TCAS. The problem right now is we don't know whereabouts in the 0210 minute the NAV TCAS FAULT occurred, and it could be related to a combination of the speed error correction and the zoom climb, i.e. both were heading in different directions. Or the changes occurring between valid and invalid data.
Here's a clearer copy of the VH-EBA FDR print-out for reference.
Jetstar did not get a TCAS Fail, did it?
I was aiming more at the ALT erroneous speed correction component, and its rate of change which may have been too much for the TCAS. The problem right now is we don't know whereabouts in the 0210 minute the NAV TCAS FAULT occurred, and it could be related to a combination of the speed error correction and the zoom climb, i.e. both were heading in different directions. Or the changes occurring between valid and invalid data.
Here's a clearer copy of the VH-EBA FDR print-out for reference.
Last edited by mm43; 31st May 2011 at 21:57.
Join Date: Jun 2009
Location: Paris
Posts: 691
Likes: 0
Received 0 Likes
on
0 Posts
Hi DJ77,
Well, in fact, even in Normal Law, it is not really able to compensate for thrust change as it is described in the A330 FCOM (3.04.27):
Certainly, beside human factor aspects. But maybe it is an issue that those human factors appears to be seriously challenged when several crisis are declared at the same time (TCAS event/turbulence/overspeed for this A340 crew).
Originally Posted by DJ77
Hmm, the flight controls were in normal mode and no sidestick controller input was made until 28 sec. after AP diconnection when they were already a thousand feet above their assigned level.
I thought that in normal law, compensation for thrust variation was provided.
I thought that in normal law, compensation for thrust variation was provided.
Originally Posted by DJ77
Anyway, except for human factor aspects, this incident appears unrelated to AF447, this one being in ALT law.
Join Date: Aug 2007
Location: Clifton, NJ
Age: 50
Posts: 11
Likes: 0
Received 0 Likes
on
0 Posts
That suggests that things broke down with that stall warning at 37500 feet. The PF appears to have made an inappropriate action, full throttle and pull up - the apparently prescribed action to take before a stall. But it never seemed to help but the PF kept the stick back fatally long even as the thrust was, at some time, reduced to idle. It was at idle at 2 h 12 min 02, and TOGA at 2 h 10 min 51. That's all I can extract from the report.
The report is clear that the descent started 3m30 before the end of the recoring - so that would be 2h10m58s. So, my guess is that there was about 7s between the warning and the actual stall.
Now, they reached a max altitude of FL380 - so there was some climb energy left before the actual stall and we know the two immediate actions at the stall warning by the PF were TOGA and nose up sidestick.The engines would have taken time to spool up. I reckon that they were too close to the stall and the nose-up at the warning pushed them into an actual stall.