AF 447 report out
Join Date: Nov 2011
Location: Kentucky
Age: 77
Posts: 22
Likes: 0
Received 0 Likes
on
0 Posts
Flight Computers Logic?
Appreciation to all for their comments on smart cows and sticks.
One theory of control computing is to use the inertial unit as the main source of decisons. This because GPS and other sensor inputs are subject to blocking, loss, failure.
Inertial is subject to drift errors over time but essentially allows you to 'coast' through the data loss period with minimal accumulating error rather than no data to fly with.
First integral of acceleration is velocity and second integral is distance.
The inertial units shoulda/woulda been calculating three axis velocity.
So the priority of incoming bad multiple pitot data was given priority above the inertial calcualtions from multiple inertial units (3 if I remember the Elect bay layout correctly).
Three bad blocked sesnors sending poor data overroad nine good accelerometers reporting good data.
I'm having programmers nightmares...
Thanks to all for taking the front seat workload.
One theory of control computing is to use the inertial unit as the main source of decisons. This because GPS and other sensor inputs are subject to blocking, loss, failure.
Inertial is subject to drift errors over time but essentially allows you to 'coast' through the data loss period with minimal accumulating error rather than no data to fly with.
First integral of acceleration is velocity and second integral is distance.
The inertial units shoulda/woulda been calculating three axis velocity.
So the priority of incoming bad multiple pitot data was given priority above the inertial calcualtions from multiple inertial units (3 if I remember the Elect bay layout correctly).
Three bad blocked sesnors sending poor data overroad nine good accelerometers reporting good data.
I'm having programmers nightmares...
Thanks to all for taking the front seat workload.
Last edited by gleaf; 8th Aug 2012 at 11:54.
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by gleaf
Inertial (...) allows you to 'coast' through the data loss period
Per Ardua ad Astraeus
Join Date: Mar 2000
Location: UK
Posts: 18,579
Likes: 0
Received 0 Likes
on
0 Posts
gleaf - I understand the nightmares.
As HN39 says, the inertial and GPS values are of great use for a short-term 'fix', but become far less reliable at the lower speed end where wind changes are more significant. There is, however, really no need to consider them, as there is a perfectly adequate procedure for coping with loss of pressure-driven information - it was just not followed.
As HN39 says, the inertial and GPS values are of great use for a short-term 'fix', but become far less reliable at the lower speed end where wind changes are more significant. There is, however, really no need to consider them, as there is a perfectly adequate procedure for coping with loss of pressure-driven information - it was just not followed.
Join Date: Mar 2009
Location: Malaysia
Posts: 112
Likes: 0
Received 0 Likes
on
0 Posts
@gleaf
Why was the automated system not programmed to adequately inform the pilots of their exact situation, ie:, no reliable airspeed indication resulting in AP disconnect?
If the pilots had clearly understood their situation they could surely have flown through it?
I really do not want to believe that experienced pilots could not know how to fly their aircraft manually.
The issue seems very clear to me: The pilots did not truly understand what was happening and so were unable to properly address the situation.
The question is how and why this situation occurred, was it over automation or inadequate automation?
Three bad blocked sesnors sending poor data overroad nine good accelerometers reporting good data.
If the pilots had clearly understood their situation they could surely have flown through it?
I really do not want to believe that experienced pilots could not know how to fly their aircraft manually.
The issue seems very clear to me: The pilots did not truly understand what was happening and so were unable to properly address the situation.
The question is how and why this situation occurred, was it over automation or inadequate automation?
Join Date: Jul 2009
Location: DFW
Age: 61
Posts: 221
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by carjockey
The issue seems very clear to me: The pilots did not truly understand what was happening and so were unable to properly address the situation.
The question is how and why this situation occurred, was it over automation or inadequate automation?
The question is how and why this situation occurred, was it over automation or inadequate automation?
Join Date: Jun 2011
Location: Devonshire
Age: 96
Posts: 297
Likes: 0
Received 0 Likes
on
0 Posts
If PF has had limited recent experience of hand flying at altitude ( and I do not know whether this was the case), then the chances of HAVING to do so, and SUDDENLY...
It OUGHT not to have been a problem.
Could he, should he, have asked PNF for assistance ? What is the etiquette ? Would this be a "black mark" on his record ? Or should he be congratulated for being honest about his own current limitations ?
WE know (now) he has a couple of minutes or less in which to decide.
LT
It OUGHT not to have been a problem.
Could he, should he, have asked PNF for assistance ? What is the etiquette ? Would this be a "black mark" on his record ? Or should he be congratulated for being honest about his own current limitations ?
WE know (now) he has a couple of minutes or less in which to decide.
LT
Last edited by Linktrained; 8th Aug 2012 at 14:24.
Join Date: May 2010
Location: -
Posts: 66
Likes: 0
Received 0 Likes
on
0 Posts
Why was the automated system not programmed to adequately inform the pilots of their exact situation, ie:, no reliable airspeed indication resulting in AP disconnect?
It is easy for you to make the connection between the two facts. Not so for the system. Producing a relevant, non-cryptic error message for the end user, may very well require a complicated algorithm of its own! Think of error messages that you get from your computer sometimes, how helpful are they to immediately understand the root cause of the problem??
Join Date: Mar 2009
Location: Malaysia
Posts: 112
Likes: 0
Received 0 Likes
on
0 Posts
The automated system is not capable of reasoning. (Otherwise we would be talking about Artificial Intelligence.)
It is easy for you to make the connection between the two facts. Not so for the system. Producing a relevant, non-cryptic error message for the end user, may very well require a complicated algorithm of its own! Think of error messages that you get from your computer sometimes, how helpful are they to immediately understand the root cause of the problem??
It is easy for you to make the connection between the two facts. Not so for the system. Producing a relevant, non-cryptic error message for the end user, may very well require a complicated algorithm of its own! Think of error messages that you get from your computer sometimes, how helpful are they to immediately understand the root cause of the problem??
But are you seriously telling me that when an automated system disconnects in flight, the reason for the disconnect cannot be accurately communicated to the pilots?
Join Date: Aug 2011
Location: Carmel,NY
Age: 82
Posts: 35
Likes: 0
Received 0 Likes
on
0 Posts
Temp use of accelerometer data
True that ground speed is all they know, but absent a remarkable instantaneous change in the wind, they are a good approximation of the status quo. Seems to me that this could be quite useful to the PF in the short term. Also, reactivation of stall warning on lowering the nose would seem a fairly potent stimulus to resume full NU side stick would it not?
Join Date: May 2010
Location: London
Age: 59
Posts: 27
Likes: 0
Received 0 Likes
on
0 Posts
The question is not so much one of over-automation or under-automation, but the sudden transition from full automation to almost none, without adequate communication of information to allow for continuity of situational awareness.
As others have pointed out, there was more than enough alternative data sources for the flight director to be able to estimate airspeed to an accurate enough level to continue flying the aircraft safely. As we have discovered, the reality is that it would have been much safer to have had a gradual loss of automation fidelity over time than to suddenly and without prior warning disconnect the autopilot and suddenly hand manual control to the pilots.
There is both the INS, with 3 axis gyroscopes to detect change in relative speed, and also the GPS, which is not "useless" others have claimed because it measures ground speed. It would be straightforward for an algorithm to estimate airspeed based on the known difference between airspeed and ground speed during a previous interval (say 1 minute) before the loss of air speed data integrity.
At the very least, this data would be accurate enough for the flight director to apply power and pitch within known safe ranges for the altitude (e.g. N= 85% and pitch = +5 degrees), keeping the aircraft in normal law rather than just "giving up".
As others have pointed out, there was more than enough alternative data sources for the flight director to be able to estimate airspeed to an accurate enough level to continue flying the aircraft safely. As we have discovered, the reality is that it would have been much safer to have had a gradual loss of automation fidelity over time than to suddenly and without prior warning disconnect the autopilot and suddenly hand manual control to the pilots.
There is both the INS, with 3 axis gyroscopes to detect change in relative speed, and also the GPS, which is not "useless" others have claimed because it measures ground speed. It would be straightforward for an algorithm to estimate airspeed based on the known difference between airspeed and ground speed during a previous interval (say 1 minute) before the loss of air speed data integrity.
At the very least, this data would be accurate enough for the flight director to apply power and pitch within known safe ranges for the altitude (e.g. N= 85% and pitch = +5 degrees), keeping the aircraft in normal law rather than just "giving up".
Join Date: May 2010
Location: -
Posts: 66
Likes: 0
Received 0 Likes
on
0 Posts
But are you seriously telling me that when an automated system disconnects in flight, the reason for the disconnect cannot be accurately communicated to the pilots?
Briefly, in a relevant and non-cryptic manner - not sure.
I have no experience with automated systems used in flight, and cannot competently answer this question. But in general, complex software tends to be modular, each module doing a rather narrow task, and communicating with others according to a formalized protocol.
As a result, an error happening in a module will be propagated through the system, but the final recipients of the error may be completely unaware of its original cause.
You may get something like "Autopilot disconnect. Reason: error in flight data processing module"... how helpful is that?
Or even worse, a stream of error messages from all internal levels of the system: "Error in flight data processing module: Incorrect airspeed. Autopilot disconnect. Reason: error in flight data processing module"...
Last edited by ap08; 8th Aug 2012 at 16:24.
Join Date: Jun 2011
Location: Devonshire
Age: 96
Posts: 297
Likes: 0
Received 0 Likes
on
0 Posts
" BEA FR 2.1.2.3...... Inputs applied to a sidestick by one pilot cannot be observed easily by the other one... "
What is the usual range of positions for the THS in normal cruising flight ?
This THS appears to have moved to full NU. ( By sidestick ?)
So the output of a sidestick may indicate... ?
Observable to both pilots ? Or outside normal scan ?
Time is limited...
LT
What is the usual range of positions for the THS in normal cruising flight ?
This THS appears to have moved to full NU. ( By sidestick ?)
So the output of a sidestick may indicate... ?
Observable to both pilots ? Or outside normal scan ?
Time is limited...
LT
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes
on
0 Posts
Linktrained,
From Part 2 of the final report (mainly based on the results of the work of the Human Factors group):
That must be based on examination of the FDR traces of sidestick movements.
Then they say:
The first sentence is a logical consequence of the first quote, but what are they adding in the second?
From Part 2 of the final report (mainly based on the results of the work of the Human Factors group):
2.1.2.3 Control of the flight path
(...) Although the PF’s various roll inputs indicate his intention to keep the wings horizontal, it is not possible to determine what the PF’s target was in the longitudinal axis.
(...) Although the PF’s various roll inputs indicate his intention to keep the wings horizontal, it is not possible to determine what the PF’s target was in the longitudinal axis.
Then they say:
It would also seem unlikely that the PNF could have determined the PF’s flight path stabilisation targets. It is worth noting that the inputs applied to a sidestick by one pilot cannot be observed easily by the other one and that the conditions of a night flight in IMC make it more difficult to monitor aeroplane attitudes (pitch attitude in particular).
Join Date: Mar 2009
Location: Phoenix, AZ USA
Age: 66
Posts: 0
Likes: 0
Received 0 Likes
on
0 Posts
At the very least, this data would be accurate enough for the flight director to apply power and pitch within known safe ranges for the altitude (e.g. N= 85% and pitch = +5 degrees), keeping the aircraft in normal law rather than just "giving up".
The real underlying issue is the failure of the actual pilots to apply this same fundamental logic. At the end of the day the inability of a "trained" flight crew to keep a perfectly functional airplane from crashing due to a known and relatively minor technical malfunction is simply unacceptable. The answer is not in better automatics but in better truly qualified flight crews.
The real underlying issue is the failure of the actual pilots to apply this same fundamental logic. At the end of the day the inability of a "trained" flight crew to keep a perfectly functional airplane from crashing due to a known and relatively minor technical malfunction is simply unacceptable. The answer is not in better automatics but in better truly qualified flight crews.
Join Date: Jun 2011
Location: Devonshire
Age: 96
Posts: 297
Likes: 0
Received 0 Likes
on
0 Posts
The shock or unbelief of finding that you were now at "10,000ft." when just a few moments ago you had been at F/L 35+ and still with TOGA.
Even I ( long retired) knew that AirTrans A330 had taken 20 minutes without engines (fuel leak) to descend as a glider.
LT
Even I ( long retired) knew that AirTrans A330 had taken 20 minutes without engines (fuel leak) to descend as a glider.
LT
Per Ardua ad Astraeus
Join Date: Mar 2000
Location: UK
Posts: 18,579
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by SLFinAZ
The answer is not in better automatics but in better truly qualified flight crews.
Originally Posted by #656
there is a perfectly adequate procedure for coping with loss of pressure-driven information - it was just not followed.
Join Date: Jul 2002
Location: UK
Posts: 3,093
Likes: 0
Received 0 Likes
on
0 Posts
Paramount for me is getting it into the skulls of airline management that this is not a zero-sum game, and that the advances in automation are not and have never been an excuse to cut corners on crew training.
@Linktrained - the crucial factor with the Air Transat incident is that both pilots were fully aware of their situation and planned their glide path thoroughly. If the Air Transat crew had made the same inputs as the unfortunate Air France crew in 447, the aircraft would have dropped like the proverbial brick.
Join Date: Nov 2007
Location: Up front
Posts: 144
Likes: 0
Received 0 Likes
on
0 Posts
Flight Directors Off, Bird ON, The bird shows you where the aircraft is actually going....level, climbing or descending, the ac attitude on the PFD shows where the ac is pointing in pitch and roll in relation to the horizon. if the ac is pointing up but going down?
A very useful tool
A very useful tool
Join Date: May 2010
Location: UK
Age: 79
Posts: 1,086
Likes: 0
Received 0 Likes
on
0 Posts
What is the usual range of positions for the THS in normal cruising flight ?
Less simple but still incomplete :-
[ Airbus drivers please feel free to correct this explanation ]
First you need to forget the concept of a stick and trimwheel as seen on light aircraft where the stick can move the elevators from full up to full down and the trim wheel is a fine control used to unload forces fed back to the stick.
In an airliner with a THS there is no separate elevator or trim tab, the entire horizontal stabiliser is moved by jacks.
In the Airbus the side stick is effectively a fine control which moves the THS within a limited range. The trim wheel is effectively the coarse control with full authority from full up to full down. If sidestick inputs reach the limit of their effective range the autotrim system system acts, think of this as moving the small window of movement available to the sidestick up and down within the full THS range.
If a pilot holds the sidestick in the full up or full down position the autotrim will obey and drive the THS to its limit. Note that if the pilot now releases the sidestick to its neutral position the sidestick now has a small window of effectiveness at the limit of THS travel.
To get the aircraft back from this extreme situation the pilot must either hold the opposite input for a long time or start winding the trim wheel.
Thus in the case of AF447 it would have taken a long time to get the nose back down from where the idiot pilot had driven it.