NTSB Recommendation re Airbus Rudder Travel Limits
Join Date: Aug 2005
Location: fairly close to the colonial capitol
Age: 55
Posts: 1,693
Likes: 0
Received 0 Likes
on
0 Posts
The following is a discussion about AA's well received AAMP - American Airlines Advanced Aircraft Maneuvering Program.
kbrockman - The only time AA's training program suggested the use of the rudder during unusual attitude recovery was when the ailerons/spoilers were not responsive.
From AA's response:
Use of Rudder
Let me say this one more time, we do not advocate the introduction of large sideslip angles when flying at a high angle of attack. You seem to be predisposed to the belief that we are using rudder first or rudder only. The workbook is not a stand-alone document and nothing should be inferred without listening carefully to the presentation.
In four different sections of the AAMP, emphasis is focused on the fact that when the airplane is not responding to aileron and spoiler control, you should use smooth application of coordinated rudder to obtain the desired roll response. Additionally, let me re-emphasize that AAMP stresses keeping the airplane inside of the flight envelope at all times regardless of attitude. Our pilots are taught to always "respect" the stick shaker.
"I am very concerned that one aspect of the course is inaccurate and potentially hazardous," Tribout wrote. His concern: Pilots were being taught that the rudder should be used to control a plane's rolling motion
From AA's response:
Use of Rudder
Let me say this one more time, we do not advocate the introduction of large sideslip angles when flying at a high angle of attack. You seem to be predisposed to the belief that we are using rudder first or rudder only. The workbook is not a stand-alone document and nothing should be inferred without listening carefully to the presentation.
In four different sections of the AAMP, emphasis is focused on the fact that when the airplane is not responding to aileron and spoiler control, you should use smooth application of coordinated rudder to obtain the desired roll response. Additionally, let me re-emphasize that AAMP stresses keeping the airplane inside of the flight envelope at all times regardless of attitude. Our pilots are taught to always "respect" the stick shaker.
Join Date: Sep 2000
Location: Bielefeld, Germany
Posts: 955
Likes: 0
Received 0 Likes
on
0 Posts
vapilot2004 (and others),
the AA reply to the FAA/Boeing/Airbus letter did not address the matter of losing structural integrity through use of rudder.
It commented on the possible results of one application of rudder hard-over (referring to Colorado Springs and Pittsburgh 737 upsets), and as far as structure goes that behavior was necessarily covered in certification throughout the flight envelope.
PBL
the AA reply to the FAA/Boeing/Airbus letter did not address the matter of losing structural integrity through use of rudder.
It commented on the possible results of one application of rudder hard-over (referring to Colorado Springs and Pittsburgh 737 upsets), and as far as structure goes that behavior was necessarily covered in certification throughout the flight envelope.
PBL
Join Date: Jun 2009
Location: US
Posts: 497
Likes: 0
Received 0 Likes
on
0 Posts
glueball, you remind me of someone who would go down the turnpike in cruise control with their feet crossed because if nothing goes wrong the drive will be uneventfull. An in flight thrust reversal event might catch you totally off guard but hey, it isn't supposed to happen like that. Hope you have an uneventfull future because you don't seem to be prepared to have something out of the ordinary happen.
Join Date: Jun 2009
Location: US
Posts: 497
Likes: 0
Received 0 Likes
on
0 Posts
hetfield, so with your feet on the floor and the departure requires a 30 degree right bank at 500 feet and at 600 ft you lose the right engine, what do you do? Are you allowed to use the left rudder? Remember your sop says feet on floor. Inquiring minds want to know.
I think we have to distinguish between V1 cuts, where large and timely rudder deflections may be required to adequately contain the yaw and failures at higher speeds.
If you had the critical engine run down during a close-in turn after takeoff, flying manually, how would you a) recognise the problem and b) deal with the short-term effects? I put forward that unless it was a sim training detail where you were expecting something like this to happen, the first clue would be an increasing tendency to hold off bank, followed by reducing airspeed as the pitch attitude would be too high for the remaining thrust. The initial response of most pilots to a roll deviation is to apply opposite roll control, as 99.999% of the time it is caused by turbulence, wake or otherwise. The first response to a trend of airspeed loss is to lower the nose to a more appropriate attitude.
After a short period of
the amount and duration of the aileron input required would lead most people to feed in rudder to assist, especially if the roll input was a significant proportion of the total available. At that point, there would be enough clues, including engine instruments now showing a problem, that would trigger recognition of engine failure and the appropriate responses and drills.
Point is, it doesn't require violent rudder inputs a split second after the event to stop the aeroplane flipping onto its back. Applying a measured amount of corrective yaw in response to what you have recognised as an unwelcome trend I'd say was a better solution...
Coming back to the central theme of this discussion, I agree with many of the posters that there is no need for significant rudder inputs in jet airliners, outside of low-speed engine failures and crosswinds. The only possible exception that I can think of is if you had reached the limits of control and were heading outside of them, then you could (gently) try the rudder *as a last resort*. At any more than moderate speeds, the primary and secondary effects of rudder input are so powerful that you stand to lose much more than you could possibly gain.
If you had the critical engine run down during a close-in turn after takeoff, flying manually, how would you a) recognise the problem and b) deal with the short-term effects? I put forward that unless it was a sim training detail where you were expecting something like this to happen, the first clue would be an increasing tendency to hold off bank, followed by reducing airspeed as the pitch attitude would be too high for the remaining thrust. The initial response of most pilots to a roll deviation is to apply opposite roll control, as 99.999% of the time it is caused by turbulence, wake or otherwise. The first response to a trend of airspeed loss is to lower the nose to a more appropriate attitude.
After a short period of
the amount and duration of the aileron input required would lead most people to feed in rudder to assist, especially if the roll input was a significant proportion of the total available. At that point, there would be enough clues, including engine instruments now showing a problem, that would trigger recognition of engine failure and the appropriate responses and drills.Point is, it doesn't require violent rudder inputs a split second after the event to stop the aeroplane flipping onto its back. Applying a measured amount of corrective yaw in response to what you have recognised as an unwelcome trend I'd say was a better solution...
Coming back to the central theme of this discussion, I agree with many of the posters that there is no need for significant rudder inputs in jet airliners, outside of low-speed engine failures and crosswinds. The only possible exception that I can think of is if you had reached the limits of control and were heading outside of them, then you could (gently) try the rudder *as a last resort*. At any more than moderate speeds, the primary and secondary effects of rudder input are so powerful that you stand to lose much more than you could possibly gain.
Per Ardua ad Astraeus
Join Date: Mar 2000
Location: UK
Posts: 18,579
Likes: 0
Received 0 Likes
on
0 Posts
Once again all this handbag swinging
Why not post clearly? My policy is cover until clean.
Next?
Now then, back to #83 - do I take it in view of the lack of clarification here that is is only PROBABLE that the accident occurred due to mishandling of the rudder, so a pre-disposing structural failure remains a possibility?
Why not post clearly? My policy is cover until clean.
Next?
Now then, back to #83 - do I take it in view of the lack of clarification here that is is only PROBABLE that the accident occurred due to mishandling of the rudder, so a pre-disposing structural failure remains a possibility?
Takeoff: "From the application of takeoff power, through rotation and to an altitude of 35 feet above runway elevation."
ICAO link.
So I suppose if you levelled off at 30' and flew a circuit, you'd never leave the takeoff phase!
ICAO link.
So I suppose if you levelled off at 30' and flew a circuit, you'd never leave the takeoff phase!
do I take it in view of the lack of clarification here that is is only PROBABLE that the accident occurred due to mishandling of the rudder, so a pre-disposing structural failure remains a possibility?
- presumably your 'etc' includes premature fin failure BEFORE the rudder input (not by aliens), then?
I don't see much in the report (not that I've read every single word) that would lend itself to a 'premature' scenario. To do that you'd have to regard the data and conclusions as false/incorrect, which takes you into conspiracy theory territory.
If there's anything simple to be gleaned from this accident it must be to leave the rudder alone at high speed (240kts/275mph in this case). An unchecked PIO often ends in disaster.
Join Date: Dec 2007
Location: Cornwall-on-Hudson, New York
Posts: 875
Likes: 0
Received 0 Likes
on
0 Posts
"@p51guy
The term "takeoff" in our SOP means until slats/flaps are retracted, so something around 2.000' AGL.
Hope that helps"
Incredible as it may seem, I think we all know what Hetfield means.
The term "takeoff" in our SOP means until slats/flaps are retracted, so something around 2.000' AGL.
Hope that helps"
Incredible as it may seem, I think we all know what Hetfield means.
Per Ardua ad Astraeus
Join Date: Mar 2000
Location: UK
Posts: 18,579
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by FW
an audible "bang" which correlated with a change in lateral loadings.
Thread Starter
BOAC;
Re #83,
Animations are not strictly investigative tools in the sense that they should be relied upon to determine what happened when. Animations are used to provide situational awareness from which an examination of the actual data can be launched.
The "draw" of animations in a visual, (non-textual) culture such as ours is psychologically very powerful. The format does not invite examination. Rather, for non-specialists especially, the "agenda" of such animations "convinces" rather than "raises questions".
Except for, "because they can", I am not certain of why animations have become part of the "investigative" kit at all. They are helpful in the act of imagining either the airplane or the cockpit indications but there are problems with using animations for detailed examinations such as the kind often suggested regarding the rudder and rudder pedal symbols' motion. I have posted my comments numerous times on these problems in hopes of placing the animation tool in appropriate perspective. Animations may be useful to "place context" but should not be relied upon for detailed information or conclusions.
The reason for this lies in the way flight data is recorded. Very simplistically but importantly, there are two problems: 1) within the one-second recording time frame, dozens if not thousands of parameters are recorded sequentially and not all at once, and, 2) each of these parameters is recorded at an instant in time and most of the time, (the rest of the second), is not recorded. There are exceptions to the "once-per-second" recording - Vertical 'g' is usually recorded 8x per second, etc.
But, stating it differently, most flight data is collected at the rate of one "snapshot" per second. Within a one-second time frame, thousands of parameters sequentially take their place in the dataframe until the beginning of the next second.
These one-second digital snapshots are placed in a dataframe one after the other, so to speak. To re-emphasize, the parameter that may be recorded at the precise beginning of a second, let us say, rudder pedal position, will come slightly before the parameter, let us say, rudder position, which is recorded, in its turn, slightly past the precise beginning of that second, and so on, through thousands of parameters.
But because animations are mathematically smoothed to avoid an inherent jerkiness from digital input which has more periods of "no-data", than not, assumptions are made by the animation software designers about "what happened in the spaces where there is no data", and so data is filled in where there is none.
Think of one stroboscope when observing the motion of just one thing in the dark - while the strobe is lit, we can see what is going on with the control or indication under the strobe-light, but once the light is out, nothing can be determined for sure even though we may be tempted to interpolate "between data points".
Most of the time such interpolation is innocent and usually does not lead to mistakes in understanding what happened.
But in swiftly-occurring events such as a hard landing where examination of both 'g' and flight control surfaces would benefit from a "frame-rate closer to video", we are stuck with interpolation of parameters sampled at only 4x and 8x per second. Thus, when the "strobe is off", we are "in the dark" about where the sidestick was, or what the rate of descent was, in-between samples. The same may be said even more so with parameters which are only sampled once-per-second.
The assumption implicit in a casual reading, (an animation) of the data is that "things happened at the 'same time' ", when in fact they may not have, and incorrect conclusions may be drawn when only looking at an animation. Thus the advice to be cautious when watching an animation of an incident or accident. We just do not know what went into the animation to make it look the way it does. Within such animation programs are dozens of corrections to wind, to magnetic variation, to rates of descent and so on, which are available to the analyst and which, when used with intelligent comprehension of the event achieved through close examination of the actual data, can reflect, within the limits of the technology, what happened. But, despite what the Discovery Channel teaches, animations of incidents or accidents are only rough approximations, and are not videos of what actually occurred.
So, for detailed understandings, I would readily dismiss the animation of AA587 except as a general, situational tool and I would avoid drawing the conclusion about rudder pedals and rudder positions until the actual flight data was examined. We don't even know if the rudder pedal position "parameter" was merely derived from the rudder position, (such derivation is not unusual but is in my view a waste of time and calculating capacity), and not an independent parameter. I have been unable to locate any DFDR readings and none is provided in the NTSB report.
That all said BOAC, in response to your question, I think it is reasonable to conclude that fins don't just come off airplanes without cause. No plausible alternate theory as been posited which places the departure of the fin prior to the rudder inputs.
PJ2
Re #83,
The wmv shows rudder movement in line with pedal displacement until the supposed 'separation', but is that supposition or confirmed?
The "draw" of animations in a visual, (non-textual) culture such as ours is psychologically very powerful. The format does not invite examination. Rather, for non-specialists especially, the "agenda" of such animations "convinces" rather than "raises questions".
Except for, "because they can", I am not certain of why animations have become part of the "investigative" kit at all. They are helpful in the act of imagining either the airplane or the cockpit indications but there are problems with using animations for detailed examinations such as the kind often suggested regarding the rudder and rudder pedal symbols' motion. I have posted my comments numerous times on these problems in hopes of placing the animation tool in appropriate perspective. Animations may be useful to "place context" but should not be relied upon for detailed information or conclusions.
The reason for this lies in the way flight data is recorded. Very simplistically but importantly, there are two problems: 1) within the one-second recording time frame, dozens if not thousands of parameters are recorded sequentially and not all at once, and, 2) each of these parameters is recorded at an instant in time and most of the time, (the rest of the second), is not recorded. There are exceptions to the "once-per-second" recording - Vertical 'g' is usually recorded 8x per second, etc.
But, stating it differently, most flight data is collected at the rate of one "snapshot" per second. Within a one-second time frame, thousands of parameters sequentially take their place in the dataframe until the beginning of the next second.
These one-second digital snapshots are placed in a dataframe one after the other, so to speak. To re-emphasize, the parameter that may be recorded at the precise beginning of a second, let us say, rudder pedal position, will come slightly before the parameter, let us say, rudder position, which is recorded, in its turn, slightly past the precise beginning of that second, and so on, through thousands of parameters.
But because animations are mathematically smoothed to avoid an inherent jerkiness from digital input which has more periods of "no-data", than not, assumptions are made by the animation software designers about "what happened in the spaces where there is no data", and so data is filled in where there is none.
Think of one stroboscope when observing the motion of just one thing in the dark - while the strobe is lit, we can see what is going on with the control or indication under the strobe-light, but once the light is out, nothing can be determined for sure even though we may be tempted to interpolate "between data points".
Most of the time such interpolation is innocent and usually does not lead to mistakes in understanding what happened.
But in swiftly-occurring events such as a hard landing where examination of both 'g' and flight control surfaces would benefit from a "frame-rate closer to video", we are stuck with interpolation of parameters sampled at only 4x and 8x per second. Thus, when the "strobe is off", we are "in the dark" about where the sidestick was, or what the rate of descent was, in-between samples. The same may be said even more so with parameters which are only sampled once-per-second.
The assumption implicit in a casual reading, (an animation) of the data is that "things happened at the 'same time' ", when in fact they may not have, and incorrect conclusions may be drawn when only looking at an animation. Thus the advice to be cautious when watching an animation of an incident or accident. We just do not know what went into the animation to make it look the way it does. Within such animation programs are dozens of corrections to wind, to magnetic variation, to rates of descent and so on, which are available to the analyst and which, when used with intelligent comprehension of the event achieved through close examination of the actual data, can reflect, within the limits of the technology, what happened. But, despite what the Discovery Channel teaches, animations of incidents or accidents are only rough approximations, and are not videos of what actually occurred.
So, for detailed understandings, I would readily dismiss the animation of AA587 except as a general, situational tool and I would avoid drawing the conclusion about rudder pedals and rudder positions until the actual flight data was examined. We don't even know if the rudder pedal position "parameter" was merely derived from the rudder position, (such derivation is not unusual but is in my view a waste of time and calculating capacity), and not an independent parameter. I have been unable to locate any DFDR readings and none is provided in the NTSB report.
That all said BOAC, in response to your question, I think it is reasonable to conclude that fins don't just come off airplanes without cause. No plausible alternate theory as been posited which places the departure of the fin prior to the rudder inputs.
PJ2
Guest
Posts: n/a
PJ2
I can appreciate your description of sampling and animation; in early movie houses, one could hear 24 clicks per second, as each frame "posed" in the lens. The human eye sees 24 "samples" as continuous motion, for electrochemical reasons that are weighty, to say the least. The herky jerky of one second samples is well described in your intimation of "bit rash". At the other end of the camera are high speed cameras in the tens of thousands of eps.
587? The first "bang" was most likely the forward mount of the VS severing itself from the fuselage. I see that happening as the first catastrophic shock load of a Rudder slamming into its stop (30degrees?) puts untenable stress on the joint, a result of the VS lever action forcing the forward lugs "up" and away from the clevis acting in concert with side loads (undesigned for) try to root out the entire assembly.
Is the game up? I think no. The VS is now operating in a sloppy plane as loads cause it to articulate with a broken foremount, and an aft one that has been driven into the fuselage (tail cone), swivelling around a middle joint that is most likely compromised, but still holding. This would express itself in excursions in the lateral, though independently of the Rudder (essentially). Now the cockpit is reacting to a new control: a slow and unpredictable loose VS/Rudder, and response would of course be unpredictable re: pedal inputs. Now might be the time to put foot on the floor, and think about RTB. No pilot would entertain what he is dealing with, it is not trained for. Had it been anticipated by the engineers, a sacrificial hinge may have been installed to prevent the Rudder from overloading the VS.
No. Yaw is atypical, and no one thinks to nurse a wounded bird back to land. The pilot is still trying to "save" the a/c. Instead, he pedals again, against a now fully articulating VS/Rudder assembly, which produces new and surprising results. The Rudder is now ineffective, and who would know the a/c is about to lose the tail? I also believe the Rudder at this point is not the problem. It is what can be described as the "procuring cause". Instead, the VS is flopping around in the 259 knot airstream, until the back and forth has worn out the middle hinge, and the VS flies off to land in the Bay.
The "Finding" must be written by an authority that has many cousins. "Inconclusive" is a favorite word when boxed into a corner by Manufacturers, Old Pilots, Airlines, and politicians. Conclusion? Frangible Rudder or moderated deflection, or "keep your feet on the floor?"
I can appreciate your description of sampling and animation; in early movie houses, one could hear 24 clicks per second, as each frame "posed" in the lens. The human eye sees 24 "samples" as continuous motion, for electrochemical reasons that are weighty, to say the least. The herky jerky of one second samples is well described in your intimation of "bit rash". At the other end of the camera are high speed cameras in the tens of thousands of eps.
587? The first "bang" was most likely the forward mount of the VS severing itself from the fuselage. I see that happening as the first catastrophic shock load of a Rudder slamming into its stop (30degrees?) puts untenable stress on the joint, a result of the VS lever action forcing the forward lugs "up" and away from the clevis acting in concert with side loads (undesigned for) try to root out the entire assembly.
Is the game up? I think no. The VS is now operating in a sloppy plane as loads cause it to articulate with a broken foremount, and an aft one that has been driven into the fuselage (tail cone), swivelling around a middle joint that is most likely compromised, but still holding. This would express itself in excursions in the lateral, though independently of the Rudder (essentially). Now the cockpit is reacting to a new control: a slow and unpredictable loose VS/Rudder, and response would of course be unpredictable re: pedal inputs. Now might be the time to put foot on the floor, and think about RTB. No pilot would entertain what he is dealing with, it is not trained for. Had it been anticipated by the engineers, a sacrificial hinge may have been installed to prevent the Rudder from overloading the VS.
No. Yaw is atypical, and no one thinks to nurse a wounded bird back to land. The pilot is still trying to "save" the a/c. Instead, he pedals again, against a now fully articulating VS/Rudder assembly, which produces new and surprising results. The Rudder is now ineffective, and who would know the a/c is about to lose the tail? I also believe the Rudder at this point is not the problem. It is what can be described as the "procuring cause". Instead, the VS is flopping around in the 259 knot airstream, until the back and forth has worn out the middle hinge, and the VS flies off to land in the Bay.
The "Finding" must be written by an authority that has many cousins. "Inconclusive" is a favorite word when boxed into a corner by Manufacturers, Old Pilots, Airlines, and politicians. Conclusion? Frangible Rudder or moderated deflection, or "keep your feet on the floor?"
Join Date: Sep 2002
Location: La Belle Province
Posts: 2,179
Likes: 0
Received 0 Likes
on
0 Posts
We don't even know if the rudder pedal position "parameter" was merely derived from the rudder position, (such derivation is not unusual but is in my view a waste of time and calculating capacity), and not an independent parameter. I have been unable to locate any DFDR readings and none is provided in the NTSB report.
Rev the rudder parameter - footnote 84 on page 49 of the NTSB report indicates that the control positions WERE instrumented, not derived from control surface positions:
During the initial readout of the FDR, the conversion equation for the rudder pedal parameter was found to be incorrect. Thus, the FDR data had to be examined to obtain reference data to establish the proper equation for the rudder pedal position parameter. In addition, revised equations for the control wheel and control column positions were established based on the FDR data. (The airplane was not initially configured with the rudder pedal position, control wheel position, and control column position parameters because they were not mandatory when the airplane was manufactured.) American Airlines added flight control input sensors and the associated hardware to the airplane to comply with 14 CFR 121.344, which required that, no later than August 20, 2001, all transport-category airplanes be equipped with FDRs that record control input positions.
I haven't found any plots of that data yet though...
Thread Starter
Mad (Flt) Scientist;
Thanks...given the age of the aircraft, (fairly young, compared with manufacture dates in the '90's, where pretty basic parameter sets obtain - 14 mandated parameters, anyone?), I suspected it would be recorded but didn't assume.
It's curious that the plots aren't available. This was a very high-profile accident where many strong interests were/are to be found. One would have thought, given that most reports include at least some traces, that the data would have been available. Perhaps it is and we haven't located it yet.
PJ2
Thanks...given the age of the aircraft, (fairly young, compared with manufacture dates in the '90's, where pretty basic parameter sets obtain - 14 mandated parameters, anyone?), I suspected it would be recorded but didn't assume.
It's curious that the plots aren't available. This was a very high-profile accident where many strong interests were/are to be found. One would have thought, given that most reports include at least some traces, that the data would have been available. Perhaps it is and we haven't located it yet.
PJ2