AMR 587 Airbus Crash (merged)

[Wino]

worse than that is what happens if the airplane on its own starts swinging the rudder back and forth. AA Airbus's have had several incidents where the rudder started swinging back and forth on its own hard enough to cause injuries.

If it decides to do it on its own you may be screwed. The Yawdampener has up to 15 degrees of authority depending on the regime of flight. The rudder load limiter limits the rudder down to 3.5 degrees of travel around 315 knots. So basically the Yaw dampener has full authority over the rudder. I am now to believe that should the nose yaw wildly (say for example in a wake turbulence encounter) that when the yaw dampener applies opposit rudder to counteract the yaw it may snap the tail off the aircraft. What a lovely design.


Plus has anyone else noticed how badly airbus is behaving during the hearings?

Apparently there has NEVER been a design defect in an airbus aircraft.

Cheers
Wino

[NigelOnDraft]

cwatters:

<<So what should you do if the initial yaw is caused by sever turbulence?>>
How on earth does turbulence lead to significant yaw that is not almost immediately corrected by normal aerodynamics?

<<Wouldn't you be tempted to apply full oposite rudder to correct it?>>
Absolutely not... and I would be horrified to know you would even consider touching the rudders in this case?

In any jet aircraft I have flown (except the one that hovers, which is different), from light jet trainer to FBW widebody, to Boeing, you only use the rudder pedals:
1. To taxi
2. To kick off drift on a xwind landing
3. To conteract (and no more) the effects of an engine failure

NoD

[BlueEagle]

Apart from the three occasions mentioned above the only other time I have been required to use rudder on a jet is in the SIM recovering from unusual attitudes, in this case a massive overbanking to almost 90 degrees, then just enough rudder opposite to the direction of bank to prevent the nose dropping, roll off bank and gently pull out. That was taught on the B744.

[Edited to get it right!:o ].

[Anthony Carn]

Anyone know the airspeed at the time of this incident ?

[NigelOnDraft]

260ish?

Blue Eagle - interesting about the rudder. At 90 AoB, I would have thought the nose needed very little persuading to drop!

NoD

[N380UA]

As the printed NTSB report stated, the number two pitch trim and the yaw damper would not engage. The corresponding CB was cycled and all was back on line and cleared by line maint..
The cause may have never really been fixed. The yaw damper as well as the pitch trims both serve the FAC, at least on the 310, and there may have been the problem.
Now, what if the vortices from the JAL flight hit the Airbus in such an unfortunate angel, that the parameters given through the FAC were exceeded, the alpha-floor trying to correct a too high angle of attack, re-triggers the pitch and yaw damper problem causing the accident.
Such a scenario should of course have produced an ECAM warning, and yet, if there was a system problem in the first place it may never even have been displaced.

A bit of maybe but just a thought.

Cheers

[patrickal]

In my 20+ years experience in computer technology, I have never been confident in recyling circuit breakers to fix a problem. At best, this process usually masks an intermittent problem that almost always returns. I do not understand why the airline industry, especially in todays electronics dependant aircraft, continue to endorse this process as valid for fixing problems.

[Belgique]

Reliable Redundancy

Patrickal

Quite right. And the "cleared" problem need not have been electrical. It might have been, in the case of AA587, mechanical and related to the 3rd servo-actuator (the one that serves up the output for both yaw dampers). Yaw damper actuators spend 99.999% of their time moving through a very limited travel and so probably pick up quite distinctive wear patterns. Perhaps it's only at "those other times" that travel outside their normal comfort zone detects a bit of stiction (or dirty hyd fluid or corrosion/erosion) and causes the FAC to hiccup. Those other times? Pre-start BITE checks and the dynamism of wake encounters or CAT.

Fascinating design arrangement that one. A soft-drive (#2YD) and a hard-drive (#1YD) act through a singular servo-actuator valve. Hardly fail-safe or fail-operational if a momentary intermittency in one causes the system NOT to trip, but to cycle between the two yaw damper outputs. I wonder if that might induce rudder oscillations?

During the NTSB's Public Docket (just finished) they very vaguely referred to a failure mode of those three rudder hyd actuators as "force-fighting" and reassured everyone that because of that FEDEX hangar failure of an actuator they'd implemented a 1300 hour synchronization check. Now that's about as reassuring as Firestone introducing a 10,000 mile safety check on their quality radial tyres. It doesn't make that failure mode evaporate. Until force-fighting was mentioned, I thought that that possibility might just have been a figment of my imagination.

Also interesting that the Airbus chappie gave, as a reason for having three rudder actuators connected to a single rudder panel, the logic that "how else would you provide for each of the 3 hyd systems to be available and cover a dual failure (in the uncontained engine failure case)". But no-one seemed to think it relevant to query whether it constituted "reasonable redundancy". That term was introduced by Dr Loeb in the NTSB Public Docket on the 737 rudder hard-overs. Eventually Boeing has had to introduce a 737 fix. I personally find much similarity between the 737 rudder valve's shenanigans and the misconduct of the A300-600's rudder over the many incidents culminating in AA587. The similarity of the 737's actuator and the dual acting A300 yaw damper servo actuator is very striking. I'm suggesting that it may be the quintessential Achille's Heel.

Below (in a few links) is the story of two 747 uncommanded yaw incidents. Just reflect upon a similar malfunction involving NOT the life-saving split panel rudder arrangement in the 747 incidents, but the single panel scenario - with a number of FAC driven authorities fighting for rudder command and control (and perhaps passing the ball between them). The dismissive comment in the 747SP incident report is very telling when you think of AA587. It says:
"System redundancy had operated as required to limit the effect of the upper yaw damper anomaly" Here they are referring to the split rudder on that a/c.



A few links
one

two

three

four (the final two posts)

[PickyPerkins]

Does anyone who attended the NTSB Public Hearing October 29- November 1, 2002 at Belle Harbor, New York November 12, 2001, on the American Airlines Flight 587 crash have any observations or comments to share?

[wes_wall]

I watched a little via WebCast, and in part, AirBus seemed to be suggesting that the rudder movement was crew induced. In support of this position, they suggested that AA SIM training
was flawed, in that excessive rudder application was being recommended as a primary tool in recovering from uncommanded roll situations. AirBus also indicated that full pedal travel (about
1 inch) at 250K could cause enough movement and subsequent loading for the rudder to fail. They advocate use no rudder in roll recovery. I have never flown the Bus, other than a PAX, but something doesn’t add up. Evidently, the rudder limiters will not restrict movement at speed sufficiently to prevent structural problems. How can you certify an airplane that has a primary control surface which is likely to fail under normal flight operations?

Question for AirBus crew... There was a lot of conversation about unsafe gear proceedures. AirBus advised their recommended proceedure is to yaw the airplane back and forth in an attempt to free the gear, which requires alternating rudder inputs. I am sure this is at slower speeds, but is their a noted speed restriction to prevent rudder loading? AA's proceedure is to roll the airplane, 45 to 45, and refrain from rudder use. Comments?

And Blue Eagle, are you sure you did not apply top rudder to help prevent the nose from dropping while you returned the airplane to level flight.

[RatherBeFlying]

Yaw Damper and Pedal Movements

Graph on last page shows four full opposite yaw damper oscillations and five full travel opposing pedal movements in the last seven seconds the tail was attached.

There has been considerable discussion that the breakout force on the rudder pedals is very close to that required for full allowable pedal motion at the accident airspeed. At this airspeed the pedal motion was limited to approximately one inch.

The yaw damper shows a classic sine wave with a lengthening of period over the seven second interval while the pedal movement shows greater excursions both in and out of phase with the yaw damper except that for one two second period the rudder pedal was held at full right.

The rudder travel limit was exceeded on four occasions, the last one could be the rudder coming off.

Both the yaw damper and PF seem to have perceived and reacted to rapid yaw oscillations.

While it seems that the evidence points to the tail suddenly being wrenched off, it should be noted that the first oscillation was contained within rudder travel limits and subsequent oscillations diverged. If the forward lug took any deformation, divergent aerolasticity would do the rest.

Taken from exhibit
168467 Oct 04, 2002 Systems 9 - A300-600 Yaw Damper System Overview (Exhibit No. 9-B)

[Anthony Carn]

Looking at the previous thread link, if my co-pilot started "booting" the rudder around like that at those speeds, then I'd have big worries about the fin on ANY aircraft type! Just try to visualise the aircraft response alone, to such inputs !

The yaw damper seems to be reacting correctly and in the correct sense and without excessive input.

Even in extreme unusual attitude recovery, the rudder should have little part to play. Warnings abound regarding the possibility of overstressing the fin with "excessive" rudder input.

Recovery ( I realise all pilots reading know this. Alternatively, correct me if I'm wrong. )

(a) -- Thrust +/- (max/min)

(b) -- Roll upright (smoothly but promptly)

(c) -- Pitch +/- (smoothly, within g limits/stall warning/shaker)

....In that order in your thoughts, albeit apparently "blended" together in enaction.

No yaw/rudder, though !

What flying experience did the handling pilot have, anyone, please ?

IMHO

[RatherBeFlying]

I've not had the "pleasure" of flying an a/c with only 1 inch of available rudder pedal movement in a sudden sideslip condition.

Fine modulation is next to impossible when the breakout force is close to that required for full available deflection; so I find a claim of booting the rudder untenable in this particular circumstance.

Missing from the graph is the sideslip condition -- it may well be buried elsewhere in the docket. Even without that the PF and yaw damper seem to have perceived a sideslip oscillation -- or was it the PF and YD triggering counteractions from each other?

Earlier I laid considerable suspicion on the yaw damper, but the pedal inputs are larger.

I would focus on unavailability of fine rudder pedal modulation and inadequacy of rudder travel limitation for these circumstances.

And I would also like to see other yaw damper readouts to see if the acccident graph is an appropriate action or a runaway. Yaw damper oscillations may trigger corrective actions from the PF. So maybe the YD should have been switched off, not that the crew had much time to figure out what was happening.

[Anthony Carn]

I used the term "booting" the rudder to describe abrupt full use of whatever available rudder travel there is, be it one inch or twelve inches or whatever. The readout shows abrupt full use of available rudder.

My experience of rudder limiters is that accurate modulation is still perfectly possible.

However, I say again that ZERO rudder input should be required/used when recovering from unusual attitudes in commercial jets. Normal teaching suggests that a small ammount of rudder in the same sense as any aileron input may be used, but emphasises that OVERSTRESSING THE FIN is a distinct possibility if the rudder is used too enthusiastically.

On the readout the yaw damper smoothly lags (and therefore is reacting to) pilot inputs. The pilot input is leading the yaw damper, as opposed to reacting to the yaw damper.

All IMHO and open to correction. Nothing in my comments is intended as a slur on the pilot; I'm simply stating my interpretation of the readout.

[BOING]

The US airlines with whose training I am familiar have all been teaching inappropriate rudder use in upset recoveries for about five years. It did not start as excessive use of rudder. Initially the teaching was that an "appropriate" amount of rudder would be used to aid recovery. Over the years "appropriate" slowly evolved into "full" use of rudder. It was easy to see this transition take place sitting in the left hand seat during simulator check rides. My objections that excessive rudder was being used were ignored by the training system.

Of course, a simulator check pilot, in my airline at least, does not actually occupy one of the pilots seats during a check, therefore he is unaware of the actual technique being used during an upset recovery. All the check pilot knows is that the recovery was successfully carried out, he cannot judge (or does not care) that the recovery technique could have seriously overstressed the aircraft.

The use of large amounts of rudder for upset recovery was initially introduced by a competition aerobatic pilot who had no swept wing, high speed flight experience outside of his airline time. A technique which is perfectly valid during aerobatics was misapplied to the airline scenario. The procedure was accepted by the FAA, probably out of similar ignorance.

To see the viewpoint of someone who really knows what he is talking about - been there, done that - in the air not just in the simulator, read Davies's " Handling the Big Jets" again.

[Wino]

Having flown with both airmen in question, neither was hamhanded nor unprofessional. Both had the farside of 5000 hours of experience in large jets and well over 1000 hours in type. They were the extremely professional talented aviators.

Anthony Carn,
The sim has since been reprogrammed to more accurately represent rudder forces. During my last sim I knocked the sim off the jacks while trying to apply a miniscule amount of rudder force.

I canfurther tell you that the A300-605r is different from the original A300b4 in several critical ways beyond the noted tin vs composit tail.

First off, THe A300b4 has a ratio type rudder load limiter (as do all boeings past the 727, which though it has a blocker system, also reduces rudder Power to 800 psi when the trailing edge flaps retract) so that full travel is always the same distance on the pedals, corresponding to a smaller movement of the rudder so that rudder forces/reactions stay somewhat constant. THe A300-605r ha a blocker system that is a menace. Airbus had it right in the A300b4 and went wrong from there.

2nd the A300b4 had outboard ailerons which were removed from the A300-605r greatly limiting its roll authority in certain regimes of flight. Furthermore, you cannot raise a wing with spoilers.


You do not seam to grasp how exquisitely sensitive the Airbus rudders become at high speed.

The blocker type rudder load limiter has been used in other aircraft, but only in conjunction with DEPOWERING the rudder. The A300 keeps its 3000 psi and gets very very sensitive. Picture a car that the brakes get more sensitive the faster you go (with no antilocks). Picture a car that going 80 miles an hour the only thing that will happen if you touch the brakes is that the wheels lock and the car spins out of control. That is an A300-605r.

Your experience of rudder load limiters is not the piece of crap that Airbus put on the A300-605r.


Boing, you are somewhat incorrect or just outdated. Advanced simulators can measure the projected Gs on an aircraft maneuver and excedeing the G limit would be recorded and stop the sim (In crash protect mode). Furthermore I have never seen anyone advocate uncoordinated rudder or to simply mash the rudder. The ball would go way out of the center if you did such a thing and then you would be repeating the maneuver. Otherwise you would be comitting a spin entry. Atleast that is how I was trained, by many different people at American and my previous airlines.


Ratherbe flying.
A rudder load limiter that allows the rudder to go beyond the limit doesn't seam to be limiting, does it?

CHeers
Wino

[Anthony Carn]

Wino -- Thanks for the excellent insight into the A300-605r and into the experience levels of the crew; This lays a few of my "theories" to rest.

My statement regarding unusual attitude recovery stands, but it really is up to manufacturers to give us an aeroplane which behaves "properly". One is left wondering, based upon your description !

[RatherBeFlying]

Wino

A rudder load limiter that allows the rudder to go beyond the limit doesn't seem to be limiting, does it?

That's what the Airbus graph shows.

Possible causes:

• The rudder moved faster than the computer could keep up with
• The limiter mechanicals/structure were overpowered by the rudder movement
• Something wrong in the flight control algorithm

The other question is whether the rudder limits protect the a/c in all flight conditions. While the first rudder excursion was right at the limit, there is a possibility that the first structural damage to the forward lug happened at that point -- a small amount of flex might have started the entire train.

[ORAC]

AC, I've sent you a private message.

[wes_wall]

"The other question is whether the rudder limits protect the a/c in all flight conditions"

According to AirBus this past week, they do not!