Interesting note about AA Airbus crash in NYC

[Few Cloudy]

Carbon, also a Technique recap from John Farley

As we are now on the tech site - carbon has a different ageing characteristic from that of metals, which actually start to age (in a more or less predictable way) from day one. Carbon stays "as good as new" until its breaking point is reached, when it just goes all at once.
The trick is to design a high enough breaking point. To illustrate this in everyday life, racing bike frames of carbon last much longer than aluminium ones - but if you crash one it won´t bend - it will snap.
On the matter of technique, it might be appropriate to recall John Farley´s words from another thread:
I am concerned that several pilots seem genuinely baffled that rudder use could have the fin off a fully serviceable aircraft that was properly designed and certificated. Unhappily it can, as I will try to explain in a moment.
Before I go on please may I emphasise I am not suggesting this caused the accident this thread is about, but offer these comments to the general debate about rudder use that is going on.
The mechanism I refer to is the same one employed by adults when helping youngsters to enjoy a swing. The adult applies a very small force at just the right moment in the cycle and by so doing builds up the oscillation until in the end the displacement of the swing can become very large indeed (with much yelping from the occupant)
A similar relatively small side force generated by rudder deflection, repeatedly applied, can cause a yaw oscillation to build rather than damp. There are two possible end points in this case, either the fin will stall due to the size of its AoA (in which case the aircraft will depart from controlled flight) or the fin will break before it stalls due to aerodynamic overload.
We all enjoy the powerful damping effects of yaw autostabilisers when they produce a very small force and apply it at the right moment to reduce the yaw oscillations. Now imagine a yaw autostabiliser that is working in reverse. Or a pilot that is out of phase with his feet. Or a pilot that is pushing the rudder pedals correctly but control actuation lags result in those correct inputs being delayed to an incorrect time.
Quote from John Farley,
Post from FC.

[AirRabbit]

Originally Posted by misd-agin

The F/E did not remember the incident that the Captain testified about. The Captain said he discussed it with the FO and it never happened again.
Don't take my word for it, read the testimony.

While I don't have any additonal comments on this particular subject, I DID want to acknowledge the link and the reference to the Flight Engineer's comments you provided. My information was obviously bogus and you are, indeed, correct. I stand corrected about what I thought were the F/E comments.

[Rainboe]

Having just watched the documentary on Sky about the accident, the AA pilot replicating the event in the simulator was applying not 1.2" rudder, but a good 6 inches or so, which is what I would expect at low speed. Who threw in this blinder about 1.2"? It is incorrect! Large inputs of rudder were made. It would appear almost full scale deflection, not 1.2" movement. To have another AA pilot testifying that it was not known that you should not use rudder under such circumstances I found slightly bizarre, and unconvincing.

[RatherBeFlying]

Who threw in this blinder about 1.2"? It is incorrect! Large inputs of rudder were made. It would appear almost full scale deflection, not 1.2" movement.

The AA587 NTSB report is available at: http://www.ntsb.gov/publictn/2004/AAR0404.pdf
The 1.2" of rudder pedal travel appears on page "27" (41 of the document) in:

Table 4. A300-600 Rudder Control System Design Characteristics Compared With Those of Other Airplanes
A300-600 22.0 65.0 4.0 30.0 32.0 1.2 9.3

The columns are:

Airplane
Breakout force
135kt. Pedal force
135kt. Pedal travel
135kt. Rudder deflection

250kt. Pedal force
250kt. Pedal travel
250kt. Rudder deflection

[Rainboe]

Well the guy doing the simulator replication used large amounts, not 1.2". I'm inclined to believe what the simulator did, not what the FAA report stated, as all jets I know of work pretty similar. I do not believe the A300 uses 1.2" rudder pedal deflection to produce large rudder response at low climb speed. If the report insinuates this, it is written wrongly. That AA pilot on film was using large foot movements. Small or large anyway, there is no argument- the fin came off through overstressing because of large rudder deflections, not through any defect and not because it collected water. The only productive discussion is why was this technique used and are the implications of it understood by those who professed at the time to not understand the hazards involved. If anybody is still in doubt about the point, they should stop passenger flying forthwith until they are trained up.

[AirRabbit]

Originally Posted by Rainboe

Well the guy doing the simulator replication used large amounts, not 1.2". I'm inclined to believe what the simulator did, not what the FAA report stated, as all jets I know of work pretty similar. I do not believe the A300 uses 1.2" rudder pedal deflection to produce large rudder response at low climb speed. If the report insinuates this, it is written wrongly. That AA pilot on film was using large foot movements. Small or large anyway, there is no argument- the fin came off through overstressing because of large rudder deflections, not through any defect and not because it collected water. The only productive discussion is why was this technique used and are the implications of it understood by those who professed at the time to not understand the hazards involved. If anybody is still in doubt about the point, they should stop passenger flying forthwith until they are trained up.

Well first, it wasn't an "FAA report," it was an NTSB report - second, the simulator is going to do whatever it is programmed to do - nothing more ... nothing less. With proper programming you can make any simulator fly just like the space shuttle. And as you should know, the FAA evaluates simulators to make sure they perform and handle as much like the airplane as can be determined. The data they use for comparison purposes is data gathered during flight testing of the airplane. If you were watching someone in the simulator flailing around on the rudder and the pedal deflections were more than the 1.2 inches described, then I can promise you the airspeed was significantly less than 250 knots and I can promise you that the deflection was more on the order of 4 inches than it was 6 inches. (Sometimes pilots have a problem discerning the difference between those measurements, but that's a debate for another forum!)

No one has said that "the A300 uses 1.2 inches of rudder pedal deflection to produce large rudder response at low climb speed." The accident report specifically describes maximum rudder pedal deflection at 135 KCAS (note-calibrated airspeed) as being 4 inches using 65 pounds of force, and at that maximum pedal deflection, the rudder should be deflected 30 degrees. The same chart (and I'm referring to the chart on page 23 of the report - but, the chart on page 27, cited by RatherbeFlying, is quite eye popping in the relationships it shows!) describes maximum rudder pedal deflection at 250 KCAS as being 1.2 inches using 32 pounds of force (there is a note, stating that this 1.2 inches may change slightly because of the response of the yaw damper), and at that maximum rudder pedal deflection, the rudder should be deflected 9.3 degrees.

So the comparison between the A300-600 and the B-767 (similar sized airplanes) goes like this (same info as supplied by RatherbeFlying, but perhaps a little easier to read:

At 135 knots:

A300-600
breakout force = 22.0 pounds
pedal force at maximum deflection = 65 pounds
pedal travel at maximum deflection = 4.0 inches
rudder deflection at this point = 30.0 degrees

B-767
breakout force = 17.0 pounds
pedal force at maximum deflection = 80 pounds
pedal travel at maximum deflection = 3.6 inches
rudder deflection at this point = 26.0 degrees

At 250 knots

A300-600
pedal force at maximum deflection = 32 pounds
pedal travel at maximum deflection = 1.2 inches
rudder deflection at this point = 9.3 degrees

B-767
pedal force at maximum deflection = 80 pounds
pedal travel at maximum deflection = 3.6 inches
rudder deflection at this point = 8.0 degrees

[RatherBeFlying]

I have not seen any research that contradicts the NTSB investigation factual findings in the report.

Certainly Airbus Industrie would have raised objections had this (or any other) number been in error.

[airsupport]

Originally Posted by RatherBeFlying

I have not seen any research that contradicts the NTSB investigation factual findings in the report.
Certainly Airbus Industrie would have raised objections had this (or any other) number been in error.

Airbus will of course not object to any of the findings, unless they are blamed.

Then they would object big time.

[DozyWannabe]

That's why accident reports as a rule are findings of fact rather than attempts to apportion blame. While you do get the occasional incident where gross negligence on the part of a single party becomes the primary cause, those are few and far between.

As it happens, this case was not one of those. To imply this was a case of corporate and political collusion to blame a pilot, as quite a few on here have, is disingenuous in the extreme. The pilot's actions were down to his interpretation of the training he had been through, the aircraft did not behave as he expected, and he ended up in a dangerous situation probably before he even realised it.

Those were the findings in a nutshell - no blame, no insinuation of negligence or incompetence, just holes in the cheese lining up with tragic results.

[RatherBeFlying]

There has been a large volume of discussion about the various particular holes in the cheese in this accident.

In the case of this particular hole, had the PF managed to apply 27 pounds to move the rudder 0.6 inches instead of the 32+ pounds to move it 1.2 inches, the triggering event may not have happened.

Part of this particular hole was that there was no broadly shared knowledge at the time of the rudder characteristics at 250 kt. so that once the PF applied rudder he was likely surprised by the response.

I suspect that he tried to correct getting a bunch more rudder than he expected, but was unable to modulate or recognise he had fallen into APC/PIO.

An inch.2 of movement followed by an inch.2 back over a couple cycles ... and the fin and rudder were gone.

[aardvark2zz]

Good stuff.
.
Also, note that the rudder mechanism does have the proper speed of the plane. The speed is approx 30 kts less while accelerating.
.
Therefore as the plane was going 250 kts the internal rudder-mechanism speed signal was 220kts due to time filtering.
.
And therefore more deflection was possible than designed for. See reports and slides.
.

Originally Posted by AirRabbit

Well first, it wasn't an "FAA report," it was an NTSB report - second, the simulator is going to do whatever it is programmed to do - nothing more ... nothing less. With proper programming you can make any simulator fly just like the space shuttle. And as you should know, the FAA evaluates simulators to make sure they perform and handle as much like the airplane as can be determined. The data they use for comparison purposes is data gathered during flight testing of the airplane. If you were watching someone in the simulator flailing around on the rudder and the pedal deflections were more than the 1.2 inches described, then I can promise you the airspeed was significantly less than 250 knots and I can promise you that the deflection was more on the order of 4 inches than it was 6 inches. (Sometimes pilots have a problem discerning the difference between those measurements, but that's a debate for another forum!)
No one has said that "the A300 uses 1.2 inches of rudder pedal deflection to produce large rudder response at low climb speed." The accident report specifically describes maximum rudder pedal deflection at 135 KCAS (note-calibrated airspeed) as being 4 inches using 65 pounds of force, and at that maximum pedal deflection, the rudder should be deflected 30 degrees. The same chart (and I'm referring to the chart on page 23 of the report - but, the chart on page 27, cited by RatherbeFlying, is quite eye popping in the relationships it shows!) describes maximum rudder pedal deflection at 250 KCAS as being 1.2 inches using 32 pounds of force (there is a note, stating that this 1.2 inches may change slightly because of the response of the yaw damper), and at that maximum rudder pedal deflection, the rudder should be deflected 9.3 degrees.
So the comparison between the A300-600 and the B-767 (similar sized airplanes) goes like this (same info as supplied by RatherbeFlying, but perhaps a little easier to read:
At 135 knots:
A300-600
breakout force = 22.0 pounds
pedal force at maximum deflection = 65 pounds
pedal travel at maximum deflection = 4.0 inches
rudder deflection at this point = 30.0 degrees
B-767
breakout force = 17.0 pounds
pedal force at maximum deflection = 80 pounds
pedal travel at maximum deflection = 3.6 inches
rudder deflection at this point = 26.0 degrees
At 250 knots
A300-600
pedal force at maximum deflection = 32 pounds
pedal travel at maximum deflection = 1.2 inches
rudder deflection at this point = 9.3 degrees
B-767
pedal force at maximum deflection = 80 pounds
pedal travel at maximum deflection = 3.6 inches
rudder deflection at this point = 8.0 degrees

[AirRabbit]

Originally Posted by RatherBeFlying

There has been a large volume of discussion about the various particular holes in the cheese in this accident.
In the case of this particular hole, had the PF managed to apply 27 pounds to move the rudder 0.6 inches instead of the 32+ pounds to move it 1.2 inches, the triggering event may not have happened.

Perhaps this is true. But, the question remains ... why was it this pilot thought it necessary to use the rudder to counter any rolling tendency upon encountering wake turbulence in the first place? That's not what the AA Advanced Maneuver Training teaches; in fact, I'm not aware of any teaching anywhere that advocates the use of rudder to control roll - in either high or low speed portions of the envelope.

[airsupport]

So many different theories.

The real cause will be discussed at length at the picnic.

[misd-agin]

Originally Posted by Rainboe

Well the guy doing the simulator replication used large amounts, not 1.2". I'm inclined to believe what the simulator did, not what the FAA report stated, as all jets I know of work pretty similar. I do not believe the A300 uses 1.2" rudder pedal deflection to produce large rudder response at low climb speed. If the report insinuates this, it is written wrongly. That AA pilot on film was using large foot movements. Small or large anyway, there is no argument- the fin came off through overstressing because of large rudder deflections, not through any defect and not because it collected water. The only productive discussion is why was this technique used and are the implications of it understood by those who professed at the time to not understand the hazards involved. If anybody is still in doubt about the point, they should stop passenger flying forthwith until they are trained up.

Believe a simulator "replication" as opposed to the factual report of the NTSB?

You're inclined to believe the simulator because all jets you know of work similar? Hmmm, I think that's the first time I've seen the simulator being more accurate than the aircraft it's trying to simulate.

You said all the jets you know work pretty similar. I think your profile showed 747 and 737 knowledge. If you're talking about the rudder limiting systems, which is an issue in this crash, then they are not similar. The 747 rudder gives you about 80% more rudder travel per pound of rudder travel vs. the 737. 747 vs. A300-600R? Only 21% as sensitive. 737 vs A300-600R? Only 12% as sensitive.

You do not believe the Airbus uses 1.2" of rudder travel at low speed? Well you're either correct or very wrong. Are you talking about "low speed" or are you talking about the accident? Two distinctly different scenarios.

If you see noticeably leg movement the a/c is probably below 165 kts, or close to it.

In summary, the simulator replication showed leg movement , the facts don't matter, the FDR is wrong, the a/c rudder limiter doens't work like the engineers said it does , and all rudder designs you know "work pretty similar".

[Mad (Flt) Scientist]

To add to the above. On at least some of our (non-Airbus) designs which incorporate rudder travel limiters, the mechanics of travel limiting are similar to the Airbus system (i.e. fixed pedal/rudder gain) as opposed to the Boeing (variable gain) system and pedal travel of the order of 1/4 full travel at 250kts is entirely plausible.

If you consider that you'll want full travel at low speed (say 125 kts) and assume that your travel limits are simplistically set for a constant rudder*Q, then at 250kts you'd have 1/4 travel, and at 375kts you'd have 1/9 travel.

I certainly wouldn't place anecdotal and qualitative simulator evidence ahead of properly analysed engineering test data.

[misd-agin]

Originally Posted by AirRabbit

Perhaps this is true. But, the question remains ... why was it this pilot thought it necessary to use the rudder to counter any rolling tendency upon encountering wake turbulence in the first place? That's not what the AA Advanced Maneuver Training teaches; in fact, I'm not aware of any teaching anywhere that advocates the use of rudder to control roll - in either high or low speed portions of the envelope.

Some military fighters use rudder at high AOA manuevering for roll. I'm not sure about the new 'computer' generation but it was the way to get roll at slow speed/high AOA.

Obviously with a HUGE degree of heightened awareness and feel!
Figher pilots talk about 'stalls' and 'departures'. They ain't the same....

[ftrplt]

actually, modern fighter pilots dont talk about stall; they talk about 'rate of falling out of sky'.

Fighters only use rudder for roll at very high AOA's; not really applicable to this situation.

[Rainboe]

This aeroplane was climbing out over Jamaica Bay within about 5 miles of JFK at the encounter. I would think they were going slower than 250kts! Whatever the distance of rudder travel, I am still bemused that any experienced pilot would use rudder as primary control in that circumstance, and how he assessed it. The fatal flaw was human technique and presumably training, not fin attachment. The A300 has been carrying the can quite unfairly.

[airsupport]

The fatal flaw was human technique and presumably training, not fin attachment. The A300 has been carrying the can quite unfairly.

In your opinion, not everyones.

If the fin of a serviceable Airliner can be ripped off purely by Pilot cockpit inputs, then I for one don't think that Airliner should be flying.

However IF you are aware of these problems with the A300 composite tail, a design Airbus no longer use, makes much more sense.

[Mad (Flt) Scientist]

Originally Posted by airsupport

If the fin of a serviceable Airliner can be ripped off purely by Pilot cockpit inputs, then I for one don't think that Airliner should be flying.

You just grounded every single aircraft flying

Once again, severe enough pilot inputs will cause catastrophic damage to every aircraft out there. There's no doubt about that; the only variable is the degree of severity required, but push them hard enough and they all will break.

*caveat: except POSSIBLY FBW-envelope protected types *cough*Airbus*cough* - but even those may not be fully immune. We don't design aircraft to be flown in a carefree manner, generally, and never in a careless manner.