New Theory and Speculation On AA A300 Crash In New York
 

Moral of this story: If your plane is flying out of control, don't try and recover, you'll upset things: <img src="rolleyes.gif" border="0">

From the New York Times
[quote]Investigators looking into the crash of an American Airlines flight in Queens in November have so far found no pre-existing flaw in the jet's tail section and are now focusing on the performance of the pilots, who they believe triggered the airplane's wild rolling and yawing in the seconds before it went down.
The data recovered from American Airlines Flight 587 showed that the plane hit turbulence from a plane in front of it and seconds later, began to swing violently and break up before it fell 2,900 feet to the ground, killing 265 people. The vertical tail of the plane, and the rudder attached to it, were the first parts to break off, and investigators began to look early on at whether that caused the crash, possibly because of some undetected flaw.
But now, after extensive testing of the tail, they have found no pre-existing problem. And so they are intensely exploring whether the pilots, in trying to correct and control the plane after the turbulence, might have put more stress on the tail than it was designed to handle.
"A brand-new tail would have broken," said one investigator, underlining his belief that the effort by the pilots to control the plane set in motion the fatal series of events. Another investigator involved in the National Transportation Safety Board's inquiry pointed out that it is possible to take an airplane in perfect condition and maneuver it into a breakup, just as a driver could take a sport-utility vehicle in perfect condition and make a radical maneuver at high speed that results in a rollover or other accident.
The plane that crashed, an Airbus A300, is a long airplane — 177.5 feet — and with the fuselage acting like a long lever, sudden movements from side to side produce powerful pressures at the end, where the vertical tail sits. By international regulation, the tail is supposed to be able to withstand a force 50 percent stronger than the largest it is likely to ever encounter, and Airbus officials said that the A300 tail exceeded even that standard. But investigators now believe that the tail was overstressed.
The latest developments in the investigation come eight weeks after the American Airlines plane bound for the Dominican Republic went down in Belle Harbor, on the Rockaway Peninsula in Queens, shortly after takeoff from Kennedy International Airport on Nov. 12.
And although investigators are interested in the latest theory, they emphasize that they are far from declaring a definitive cause. Indeed, some are still trying to determine if the rudder moved differently from the way the pilots intended. The investigation is being led by the N.T.S.B., with the participation of the F.A.A., Airbus, American Airlines, the union that represents American Airlines' pilots and other aviation experts.
Investigators had focused originally on the rudder and the vertical portion of the tail to which it attaches, both of which fell off the plane. The tail is made of carbon-reinforced plastic, a composite material that has come into common use in airplane structures only in the last 15 years and investigators have relatively little experience with it.
The Safety Board had the tail trucked to a NASA laboratory in Hampton, Va., for analysis. But the lab has turned up no sign of fabrication error or damage to the tail before the accident, according to people involved in the investigation. Now, investigators think it tore off because of the increased strain placed on it by the pilots' maneuverings — rolling and skidding the plane in the air.
The Airbus had hit the wake of a Boeing 747 that was about five miles ahead of it, which is considered a safe distance. That preceding plane created what investigators say was a minor bump, but the encounter may have prompted the Airbus crew to try to compensate.
"They thought they had something from which they thought they needed to recover quickly," said one investigator, reflecting the current hypothesis. At the controls of the jet was the first officer, Sten Molin, 34. Mr. Molin was an experienced pilot, with 4,400 hours of flying time, 1,835 of them as co-pilot of an A300.
After using the flight controls to steady the airplane, the objective normally would have been to bring it back to its previous orientation — in this case climbing and banking slightly to the left in its first turn out of Kennedy.
"Before they could do that, something else happened," the investigator said.
Several investigators said the training of pilots would be carefully examined. Crews at American were trained in the mid-1990's to use the rudder to recover from "flight upsets," but Airbus, Boeing and the F.A.A. later warned against this practice, saying it could produce dangerous stresses. American said it changed its training in 1999 to de- emphasize use of the rudder. Evidence recovered from the plane's data recorder indicates that the pilots were using the rudder to try and stabilize the plane.
The investigators are finding their work slowed by limitations on their tools. One tool is a computer-driven simulator owned by Airbus, which can predict what happens to the plane with each change in rudder or other control surfaces.
But the computer does not make good predictions at dangerous angles, because it is difficult to conduct test flights to gather such data. "You don't do extremely weird things to airplanes" to gather such data, one investigator said. As a result, he said, "When you get in a situation way outside the envelope of the airplane, you don't know."<hr></blockquote>

I would expect more from the New York Times, to be quite honest. It is somewhat infuriating the way they can twist one of the various theories (without displaying objectivity and comparing it to others) to make a lay-public believe this nonsense.

Any comments?

SFly <img src="eek.gif" border="0">

[ 05 January 2002: Message edited by: SFly ]</p>

Perhaps an argument for the fly-by-wire protections featured on subsequent Airbus airliners?

There are no protections in Yaw, though.

Ofcourse no word about a design-manoeuvring speed........

<img src="rolleyes.gif" border="0">

It would be interesting to hear from someone with a knowledge of the design of rudder limiters and their protection.

Quite why anyone should attempt to use rudder as a primary control in a swept wing aircraft in flight, I cannot imagine - other than for spin recovery or to counter the initial effect of loss of engine thrust at high thrust settings. In the type I fly, the aileron and elevator artificial feel system is TAS related whereas the rudder feel is IAS related; as a result the rudder feel increasesfrom min to max much more quickly than the elvator and aileron feel, reducing the chance of inadvertant fin stress from pilot input. In the event of total feel failure, 2 of the 3 rudder PCUs are isolated to reduce rudder power. Rudder is never used to recover from a divergent Dutch roll (although we no longer practice this in the air) - the only time I make a conscious rudder deflection other than during a cross-wind flare is to apply a small rudder doublet to excite the lateral stability mode during air tests to check yaw damper operation.

[ 05 January 2002: Message edited by: BEagle ]</p>

I find this all a bit difficult to believe.

I thought that if you were below max maneuvering speed you could apply an individual control fully and would not overstress?

What speed were they doing when the initial "upset" occurred?

Surely you have to grossly mishandle to cause a break up during this phase of flight?

There must be some other factor in this accident.

It's the same old story. The pilots are dead so lets blame the whole thing on them. I'm not an A300 pilot but I do know that you should have full athority of pedal movement at any speed without breaking anything due to the rudder limiter. If you lose an engine at full thrust on that thing I guarantee that rudder is on the floor. What a load of crap! <img src="mad.gif" border="0">

I'm amazed.
I read the NY Times article before I came to this thread as a result, it had the ring of truth then and still does.

Maybe someone who knows the chapter and verse on Airbus product can help me, but I understood the design philosophy behind the whole FBW concept was that the computer control laws were such that they prevented inputs that would exceed the aerodynamic and structural envelopes.
That is, short of turning all the protections off it was almost impossible to "kill" the aircraft.
If I understood it correctly, this was supposed to allow higher design and structural limits and expand flight and stability envelopes to promote the overall operating economics.

Or is it that they'll only keep you safe as long as you remain within "normal limits" and in anything other set of conditions you are out there on your own <img src="frown.gif" border="0"> <img src="confused.gif" border="0">

I find the concept of training to one set of training skills for say engine failure, viz use of rudder and a different set for inflight excursiona in the after take off segment, ie. not use of rudder because we'll come to bits, difficult to grasp.

BEagle suggests, properly, that use of rudder "in flight";
"other than for spin recovery or to counter the initial effect of loss of engine thrust at high thrust settings" should be avoided.

Difficult to argue with that, but given that the wake turbulence from the preceding aircraft has been accepted as the event initiator, (?)
it is possible that the resulting aircraft behaviour that prompted the pilot inputs could have exhibited either of those effects (spin onset or large "apparent" assy).
Who amongst us wouldn't have reacted instinctively with rudder and/or who amongst us would have been aware of the apparent consequences.

Are we that far out on the "pushing the certification limits" bough that a single moments inattention by a pilot will break it off.

Only two things can operate here.
Either the certification process needs a seriously vigourous rethink and overhaul, and a seriously deep audit of the control laws.
or,
A couple of pilots are being hung out to dry for "killing" an aircraft by using control inputs made available to them by the computer controlled flight control system that they believed provided such protection and limits.

Either way?

gaunty--
The A300-600 is not a FBW aeroplane.
Can remember years ago dutch roll recovery training in the B707....better to keep OFF the rudder.

Quite right 411A; it is not so much a question of overstressing the fin by initial large rudder inputs, it is a question of causing a divergent lateral mode to develop which may lead to overcontrol, departure fromcontrolled flight, overstress, structural failure and an accident.

Without prejudice in any way to the deceased, one wonders quite how much in the way of manual flying is practised these days by most airline pilots.

..and yes, the A300 is not a FBW aeroplane.

Assuming the delicacy of the tail assembly to high-speed gross rudder inputs (ref 411A & Beagle); just how gross do these inputs have to be at relatively low speeds (approx 250 IAS during low level departure) to do damage? BTW, was the autopilot possibly in use at the time?

Thanks guys.
Well there you go, but I still feel uncomfortable with being able to "kill" the aircraft in that flight segment.
I'm with BEagle, with the youngsters first 'big one' now being 'fully automatic' and coming in with relatively low hours compared with the past just how well their manual skills can be maintained, particularly long haul.
In our State a unless you have learnt and pass your drivers license test on a manual car your license limits you to automatics. Dodgy parallel perhaps but you know what I mean.

411A mate is dutch roll recovery training currently included in type rating and is it done in the sim, or have they flicked that one over to the Yaw Damper too?
I ask the same question as Fireflybob and raas767.

In a related thread 'Plastic Bug' made a comment that made think
[quote] 587 managed to intersect a number of non related and normally insignifigant events that resulted in the departure of the vertical stabilizer.

The airplane, in my opinion, managed to bullseye the preceding airplanes wake turbulence not once, but twice.

While recovering from wake encounter number one, they hit number two.

With controls set to recover from a diversion in one direction and then encountering a diversion in the OPPOSITE direction, I am believing they found themselves outside of the envelope.

And that is where the tail went.

You may, of course, disagree.

<hr></blockquote>

Whatever, it seems it just wasn't their day and given the evidence and competing politics it would be a pity if they wore it.

Given the fact that the A300-600 has experienced severe rudder deflection at full authority with no control input from the pilot, no less than twice that I am aware of (not including 587), should be a huge heads up to anyone investigating this accident. For the NYT to publish that article and insinuating that our pilots may have caused that crash before thay have even finished the full technical evaluation of the FDR is nothing short of criminal. I urge each of you to listen to the APA hotline at 1800 APA PILOT were the union responds to this nonesense. These men are dead with no way to defend themselves! <img src="mad.gif" border="0">

Does anyone really understand an Airbus?
There are so many viable, decent scenarios posted here, but folks, we are SPECULATING!
Second guessing what happened to these guys is not fair. We weren't there and have no idea what they went through.
That article was way out of line.Lets just wait and see what the FDR has to say.
I've been through some wake from a Tri-star while I was in a Tri-star, and it was nothing short of violent!

Lets give these guys the respect that we would want.

I have been in excess of 500 stalls in commercial aircraft(not flying). Over half at heavy aft CG, flaps and gear down. Half of those being wind ups. I have seen and heard noises coming from airplanes that 98% of pilots will, luckily, never hear or feel during their careers. I have watched rivets popping up out of wing skins, saw flames shooting forward out of engines as they swing back and forth and I have sat in the cockpit watching the pilots violently twisting the controls after one wings breaks early and the aircraft rolls over on its back and nothing but light covers and flaps delaminating(a bit). Nothing ever broke off, even in a 727(if you know stalls then you understand).

I just can't see how a pilot fighting a bit of turbulence can bust his airplane. If they broke that easy I wouldn't be here and you guys would be loosing friends on a daily basis. It just makes no sense to me. <img src="confused.gif" border="0">

I feel a big fat political whitewash coming at our expense. This is where the accident investigation committee of the union should and will raise a HUGE stink!

On the stalls you witnessed you never saw the pilots putting in full rudder followed immediately by full opposite rudder.
Thats what breaks aeroplanes!

apfds.
There is no evidence at this point that the pilot made any rudder input at all. As you well know rudder dampening systems should prevent any overstress on the rudder even if the pilot did use full deflection, which is doubtful.
The bottom line is that we should all stop speculating about what caused this accident until all the FACTS are in. To do so is completely unprofessional, an insult to our fallen comrades, and worst of all it plays right in to the worst form of tabloid journalism. Accident investigation is a science. Good science takes time. lets wait and see.

apfds-I have seen them go hard one direction on the rudder while inputing oposite column input and then go the oposite rudder deflection with again oposite column input and back and forth to make the plane fish tail I think the most was about 8-10 times in a row in rapid succesion. Really tosses stuff around. Other than redlining an airframe, I doubt there is any input a pilot could do that should make control surfaces or the tail or engines come off a plane if it satisfies all FAR requirements for structural integrity.

Unless there was a 10 g gust that wasn't registered I just don't believe the pilot did it even on a ScareBus. Look at that Alitalia DC-8 that lost half its wing and one engine going through, and I forget the name of it, that rare wind pattern. Pilot puts in hard rudder one way plane goes the other he pushes harder, then the plane sails through the cross wind and the controls start to respond as they should now the pilot snaps the tail the other way to stop the fast roll and the wing snaped just outboard of the inner engine. He still landed the plane.

On this aircraft does the flight data recorder
log the position of the rudder pedals or just
the position of the rudder itself?

cwatters.
The last I heard, the FDR cannot determine weather the pilot caused full deflection of the rudder or if the rudder deflection caused a feed back to the peddles. Given the history of rudder problems on the A300-600 it gives credibility to the latter scenario.

Hey there, 747FOCAL, you say.....have been thru 500 stalls (not in aircraft)....this must have been in the Sim. Yes?
What makes you think the sim actually demonstrates actual stall conditions? Are you a test pilot or what?
You sound like an amateur to me. Or, to give you the benefit of the doubt, misinformed. <img src="rolleyes.gif" border="0">

....or perhaps he/she's a Flight Test Observer/Engineer and has been involved with testing although not at the controls him/herself? Incidentally, the term 'breaking' in his/her post refers to a +Gz 'break' at the stall, not structural failure, and a 'wind up' is a manoeuvre incolving increasing +Gz and angle of bank in level flight untilthe point of stalling is reached - rather him/her than me in a large aeroplane!

I suspect he/she is far from being an 'amateur' as you have suggested, 411A.

[ 08 January 2002: Message edited by: BEagle ]</p>

747FOCAL, you stated

´I doubt there is any input a pilot could do that should make control surfaces or the tail or engines come off a plane if it satisfies all FAR requirements for structural integrity.´

Just look at FAR 25.361 and you shurely believe a pilot can overstress a plane´s tail :


14 CFR 25 Sec. 25.351 Yawing conditions.

The airplane must be designed for loads resulting from the conditions specified in paragraph (a) of this section. Unbalanced aerodynamic moments about the center of gravity must be reacted in a rational or conservative manner considering the principal masses furnishing the reacting inertia forces:

(a) Maneuvering. At speeds from VMC to VD, the following maneuvers must be considered. In computing the tail loads, the yawing velocity may be assumed to be zero:

(1) With the airplane in unaccelerated flight at zero yaw, it is assumed that the rudder control is suddenly displaced to the maximum deflection, as limited by the control surface stops, or by a 300-pound rudder pedal force, whichever is less.

(2) With the rudder deflected as specified in paragraph (a)(1) of this section, it is assumed that the airplane yaws to the resulting sideslip angle.

(3) With the airplane yawed to the static sideslip angle corresponding to the rudder deflection specified in paragraph (a)(1) of this section, it is assumed that the rudder is returned to neutral.

(b) [Reserved]

[Doc. No. 5066, 29 FR 18291, Dec. 24, 1964, as amended by Amdt. 25-23, 35 FR 5672, Apr. 8, 1970; Amdt. 25-46, 43 FR 50595, Oct. 30, 1978; Amdt. 25-72, 55 FR 29775, July 20, 1990; 55 FR 37608, Sept. 12, 1990; 55 FR 41415, Oct. 11, 1990; Amdt. 25-86, 61 FR 5222, Feb. 9, 1996]

So just apply full rudder (FAR 25.361 (1)), wait for maximum sideslip angle (FAR 25.361 (2)), dont just return rudder to neutral as specified in (FAR 25.361 (3)) but apply full opposite rudder and you shurely overstess your airframe. And in this case the yawing speed is not even accounted for, for a plane wich such a long aft fuselage it produces another amount of effective tail fin angle of attack. Additionally the maximum dynamic sideslip angle is much larger than the static one, so it is easyly possible to produce about twice the angle of attack dynamically and apply rudder in the direction that increases aerodynamic forces. European JAR 25.361 is just the same as the FAR requirement.

As my structural design professor always said : It is impossible to design a plane that can not be destroyed by wrong pilot control inputs at the right speed and flight attitude. (might not be true for fighter airplane where the physical capabilities of the pilot are the limiting factor and the pilot rather kills himself before breaking the airframe)

There are some more paragraphs important for structural design of the fin like lateral gustloads and single engine failure loads, so it is not clear if the 25.351 yawing condition is the most critical for a specific airplane. But it is for shure possible to exceed the loads specified in this paragraph.

Supposing there was an uncommanded rudder 'hard-over' (like in the 737's) - it still does not explain the break-up? Those 737's didn't break up, and at cruising altitude they would have been flying much faster than this A300 was on departure.

First 411A- Why would you think simulator since I refered to feeling and hearing sounds that most pilots will never encounter or whatever I said? I only said I wasn't doing the flying. Of course they were in a real airplane. Been sitting in the cockpit of a 727 when the pilot took it down to 85 kts at 18,000 ft. He said it was probably wrong as we were at such an attitude that the pitot tubes were not getting proper airflow. In a simulator, thats rich. I can barrel roll a 747-400 at 875,000 in the sim, bet I couldn't do it in real life. <img src="eek.gif" border="0">

[ 08 January 2002: Message edited by: 747FOCAL ]</p>

Volume-
"As my structural design professor always said : It is impossible to design a plane that can not be destroyed by wrong pilot control inputs at the right speed and flight attitude. (might not be true for fighter airplane where the physical capabilities of the pilot are the limiting factor and the pilot rather kills himself before breaking the airframe)"

He said "right speed and flight attitude" This airplane was way below it's design maximum speed. And was not at an attitude that was out of cert limits. Of course if you dive to mach 1 and pull back hard at step on one rudder pedal hard the wings and tail will come off. I been in flutter tests as well.

Volume.
Your profile indicates that you are an Aero Engineer so I am hardly in a position to argue with you in this field, but does part 25 take in to account aircraft with rudder limiting systems designed to negate the large inputs that could cause the vertical stab to fail?
If it was so easy to cause a vertical stab to fail then the world would be littered with aircraft aluminum. I just don't buy it.

BEagle- Thanks for your support. You are correct I am a flight test engineer among other things. It is fun to do that stuff, but I am glad I don't do it everyday. 4 hours of stalling and your head feels all swollen up. doesn't hurt just feels different from your brain bouncing around in the the brain bucket. :)

Well 747FOCAL, suggest you keep to engineering and leave the "piloting" to....actual pilots (or flight test pilots).
And...stick to KILOS so our European friends understand.

411A- You must be daft. Then again this isn't the first thread I have seen your rantings. I think I hear the Guv calling........ How is that live in Hey Boy thing going anyway? :)

ps. to convert lbs to kilos divide by 2.2 and you should be close enough for that C152 they let you motor around your backyard.

[ 09 January 2002: Message edited by: 747FOCAL ]</p>

747

like you I am a pilot and an engineer. Surely you should have enough experience to know that we have no idea which way a particular aircraft type will react to certain stresses in certain conditions. History has proven time and time again that 'foolproof' systems have a way of failing. Imagine the bridge buiders shock who built indestuctable bridges when they were confronted by 'resonance' and found a moderate wind could knock their bridges down. I am amazed that you cannot accept that a structural problem caused by control input in certain conditions resulted in this accident.

Of course it may not have been, but JFK conspiracy theories do not lie around EVERY corner in life. History has proven that aircraft crash investigators do a pretty good job and are reasonably honest (thank God). How likely is it that they are trying to cover something up or that they are barking up the wrong tree and that a 500 stall man is correct. Unfortunately I have to say my money goes on the investigators. No offense intended but I think you need to accept that history will probably prove you wrong. Of course there is a chance that what you say is correct and I accept that and welcome discussion on it, but I think you are a bit brazen to shoot down the proven experts based on your lower level of experience in crash investigation (I assume you are not an experienced crash investigator because I'm sure you would have added it to your CV)

I don't care what any of you engineers or crash investigators say. Tails are NOT supposed to fall of transport catagory airplanes due to wake, control input, composites, or anything else.. .There is at least 50 years of design experience with aluminum flight control surfaces and all but maybe 10 with composites. I hate to throw this out, but has anyone considered the fact that Airbus may have screwed it up on this one?

I was reading an article (I think from Flight International) that indicated the data from the flight was, in fact, filtered internally.

Would this tend to hide the sudden transients and control extremes that are being talked about here?

If the data extremes can not be determined from the FDR, then what course of action can the investigators take apart from some intelligent guesswork ?

The same article also refers to the testing of a fin assembly identical to the one from the destroyed aircraft.

If my memory is correct there was a 707 that was involved in a midair breakup, in the early days of the jets, which was attributed to overcontrol in turbulence, and that aircraft was built like a brick outhouse.. .regards

Correcto Steamhead............Over Mt. Fuji in Japan. In that area you can get CAT that will shake your teeth out. I have been in 170 knot winds in that area. But that is a whole lot different than being 5 miles behind a B 747.

Can any of you try this out, as a thought experiment?

Put 10 oz of semtex pressed in a somewhat shaped charge ( as someone pressing their shoes against the AC fuselage would shape a charge), perhaps in the rear head or last few seats in the plane, and then discharge this bomb on 587.

Can you come up with the sudden gyrations and loss of control resulting in the tail falling off, along with the AC breakup, to fit the crash of 587 when you consider this charge? The size of the charge is small, so would that keep it off of the sound recorder? Would such a charge actually be shaped, such that if in the rear of the aricraft it would be similar to a small rocket thrust, violently moving the read sidewais similar to a flat spin?

Or can a stinger be fired with out a tail being noticed?

My thoughts initially raised ire and condescending platitudes from most of you, but it is near humorous (but for the fact that this is a most terroble tragedy) now in observing the contortions most of you are going through trying to figure out why a perfectly good AC dropped out fo the sky when piloted by a senior pilot during optimal flying conditions. It seems most are starting to come to an Airbus conspiracy coverup assisted by the NTSB, or soemthing similar. (A NTSB coverup of a bomb to avoid total implosion and panic of the airline industry is more likely.)

In times of war, on a clear beautiful day with little or no normally occuring exogenous forces or events identified, a large AC will usually fall from the sky from an act of war. Makes sense to me. Why do you refuse to take the Occum's Razor approach?

I am very impressed with the majority of "ostrich" like views here that refuse to try on the thought model above, despite all the terrible things that have happened since 9/11 and most importantly since Reid.

Why dont you guys wake up? We are at war.

Come on, as a public service, use your professional skills and engineering ability to proof out how such a bomb charge model can fit the event. Try hard. Then , and only after that effort, should you discard an act of war as the reason. Most of you have it totally ass backwards and will be forced to consider the bomb after losing much personal credibility by trying out incredible contortions first.

Rescue your professional consderations and reputations.

Try the thought experiment.

<a href="http://pull.xmr3.com/p/25356-E59F/30254611/rudder.html" target="_blank">http://pull.xmr3.com/p/25356-E59F/30254611/rudder.html</a>

Dozens of Pilots Want American to Ground A300. .By John Crawley

WASHINGTON (Reuters) - Dozens of pilots at American Airlines want the carrier to ground its fleet of Airbus A300s, saying too many unresolved safety questions have been raised since one of the airliners crashed in New York in November.

"Are we completely comfortable putting our friends and family on an A300? If the answer to that question is not a resounding yes, then logic would lead a well-trained pilot to conclude that no one should be flying on them either,'' according to a recent open letter initially signed by a dozen American A300 pilots in New York, Boston and Miami.

About 60 pilots have signed the letter, which was also circulated to other crew members.

American, a unit of AMR Corp. has about 400 pilots who fly the carrier's 34 wide-body A300s, the company said.

The airline and Airbus SAS said on Thursday there were no plans to ground the fleet, stressing that safety officials investigating the crash of Flight 587 have found no reason to do so.

That plane, An A300-600 crashed shortly after takeoff from New York's John F. Kennedy International Airport, killing all 260 aboard and five people on the ground.

Investigators are looking at a possible mechanical problem, a structural defect with the composite materials that make up the aircraft's tail fin, or actions by the crew to explain the disaster.

No other U.S. commercial carrier other than American offers A300 service, but cargo airlines Fedex Corp. and United Parcel Service operate them.

The letter was not linked to the pilots union at American, the Allied Pilots Association. In fact, the union said it did not back the demand.

"We are not advocating a grounding of the A300 fleet at this point,'' spokesman Gregg Overman said.

The pilots who signed the letter are unhappy that investigators, the manufacturer and the airline have yet to develop a clear idea of why the jetliner's tail fin, or vertical stabilizer, and rudder fell off before the A300 crashed into a residential neighborhood.

"At this point, safety experts and our own safety and fleet people have agreed that no test exists to definitively check the structural integrity of the vertical stabilizers on our remaining 34 aircraft,'' the letter said.

After the crash, the Federal Aviation Administration (news - web sites) ordered American and the cargo carriers to conduct visual checks of the A300 tail section. All reported no safety problems.

Aside from the focus on the carbon-reinforced plastic composites that make up the thick tail fin, investigators are looking closely at three sharp, unexplained rudder deflections just before the plane crashed.

It has not been determined whether these rudder movements were commanded by the crew, or if there was a glitch in the Airbus systems associated with the flight control components.

The safety board is looking into a handful of flight control mishaps over the past several years involving the A300. One involved the ill-fated American jet in 1994.

Investigators confirmed on Thursday they were examining a report that an uncommanded rudder movement prompted the pilot of an American A300 to return to Caracas, Venezuela, shortly after takeoff on Jan. 17. That plane was bound for Miami.