Wake Turbulence / Structural Damage Studies - AA 587

[strobes_on]

With so much rumour about this terrible accident, are there any studies that actually support a theory that wake turbulence events can lead to catastrophic structural failure.

What could be the effect of the subject aircraft's involvement earlier in an overstress condition. How deep would xray techniques have looked to accurately determine structural damage?

Any engineers / specialists care to take this on ?

[Zeke]

Due to the short time period witnesses have said it took for the aircraft to begin to self-destruct in flight, I am inclined to think that the aircraft broke up as a result of some flutter mode, possibly involving the engines.

Passing through the wake of a 747 in itself would not cause the failure of the primary control surfaces, however a aeroelastic effects induced as a result of exciting the airframe at a particular frequency may.

If the aircraft was found to have passed through the wake of a 747 it may have induced an unknown flutter mode which could lead to a rapid self destruction.

IMHO it will be a very long drawn out investigation.

Genghis do you have any thoughts on this theory ?

Z

[Backoffice]

Just to speculate on what may have weakened the fin on the past has anyone considered the affects of high winds on the ground.
I seem to recall, a 747 was moved, sideways, some distance by the 100 mph gusts
at London Heathrow during our famous October “hurricane”.
This A300 may have spent a lot of its life on the Carribean sectors, perhaps more exposed, than most, to buffet from high winds whilst parked.
I doubt if the airline would keep any records of exposure to such conditions, unless actual physical damage was identified at the time.

[lame]

I find it impossible to believe that the amount of wake turbelence in this case, or any amount of pilot input (which some have blamed), could rip the vertical stabiliser off any kind of modern Airliner in even reasonably serviceable condition.......

About two years ago, ironically at NY JFK, a similar Aircraft while being towed out of one of the hangars, collided with the steel girders of the hangar. Before the Aircraft came to rest the steel girders had ripped almost 75% of the way through the vertical stabiliser from fore to aft, it did NOT however even then dislodge the fin.

Obviously the fin and mounts were subsequently changed by a team from Airbus, but I find it hard to believe that an Aircraft that can survive this, can lose a fin due to any amount of wake turbulence...... :confused:

[jugofpropwash]

A side note regarding composites. My understanding is that the tail is made from carbon fiber or a simular material?

Recent day Indy racing cars are also made of a carbon fiber material. A few years ago, racing politics resulted in a number of year old/2 yr old/3 yr old chassis that normally would have been retired after a single season being called back into service. What is interesting is that the manufacturers put out warnings that the chassis were being used (and repaired) well past their design limits, and that there could be problems including delamination.

At the time many of us felt this warning was political in nature, but recent events make me wonder if they did suspect a possible problem.

[john_tullamarine]

Continuing airworthiness of structural composites is a major source of head scratching... particularly as a failure is sudden, unlike with more traditional materials which usually give some indication of their state.

I suspect that Zeke may be proved correct in part ? There have been reports regarding prior damage to the stucture .. a change in structural stiffness could well have a disastrous effect on flutter onset. As to the particular sequence of events .. the investigation likely will pinpoint the problem ...

[ 18 November 2001: Message edited by: john_tullamarine ]

[Deaf]

Further thoughts on this, a key feature of composites is the different properties on different axes and this would have been modified in one area only by the repair. Add in the wake turbulence ie varying aerodynamic stress from odd angles and we have a possibility (I'm no expert on aeroelasticity) of a flutter mode unique to THAT a/c subjected to THAT combination of airflow on different parts of the fin.

The problem with this senario is that it is unlikely to be possible to reproduce/model it and hence to definitely include/exclude it.

On a related topic while I believe composites to be the way to the future the current state of the art appears to be just that, an art. Design in metals appears fairly straightforward (I have a degree in Metallurgy. Reading various conference reports on composites I come away with the suspicion that it is still basically cut and try with QC a real problem.

[Flight Safety]

Correct me if I'm wrong, but doesn't material stiffness play a direct role in flutter? I understand that natural resonance frequencies of a structure are determined in part by material stiffness (the stiffer the material the higher the frequency). Wouldn't the natural resonance frequency of a composite structure change with a partial delamination of the composite material in that structure?

Regarding the accident aircraft, I'm curious as to what caused the delamination of the tail fin mounting point while the aircraft was still on the factory floor. Was it caused by an accidental collision with some other object in the factory during assembly, or was it a resin mixture problem, or maybe an incorrect heating profile in the autoclave?

An impact delamination would not have the same long term consequences that a materials processing error would. A materials processing error, that might cause more widespread delamination over time, would certainly alter the natural frequency and therefore the flutter characteristics of an aircraft structure. Any loss in stiffness in any of the fin attachment points would seem to have a direct bearing on the natural frequency of the fin/fuselage join.

(Edited for a spelling error, and adding an extra comment)

[ 18 November 2001: Message edited by: Flight Safety ]

[McD]

When you look at historical examples of wake turbulence and of delamination (each independently), it's not too difficult to see how the application of wake turbulence forces on a delaminating surface could have disastrous consequences.

A quick disclaimer: I'm not posting this to speculate on the actual cause of the AA587 crash. Just showing how delamination can combine with other forces to result in severe structural damage.

Delamination may remain undetected until a certain amount of higher-than-normal force is applied to the faulty part. That force may well be within design limits, but it might be just enough to trigger the delamination to a destructive level.

Here's an example. It's only a single example, but it clearly illustrates the effects of force on a delaminating structural component:

-----

In the mid 1980s, delamination on a slab (all-moving horizontal tailplane) of a T-38 went undetected for an undetermined amount of time, partly because the aircraft had no history of problems and had flown just fine during its recent flights (low-g flights, such as instrument/nav or very basic formation flights). Then, a solo student took the aircraft up for some aerobatic training, but quickly returned to base, saying that when he was beginning a loop, things didn't feel right approaching the 5-G point (well inside the G-limits of the T-38). There were vibrations and sounds that didn't seem correct to him. Initial maintenance inspections found no problems, but since it was a perceived flight control problem, a Functional Check Flight Pilot was then scheduled to fly it for a check-out.

During the FCF flight, the aircraft performed normally as the pilot took it through some mild 2- to 4-G maneuvers. Then the FCF pilot began a 5-G loop. The aircraft immediately made a very loud "bang" type of noise and became uncontrollable. The FCF pilot (despite being strapped in tightly) was pushed up into the upper part of the cockpit, against the canopy. Fortunately, he was able to regain control and call for a chase ship to inspect him.

When the chase ship joined up, everything looked fine on the left side. But on the right side, the slab was pretty much gone, ripped away. Only a small spar (about a foot long, and a few inches wide) stuck out where the entire right slab should have been.

-----

So, it's entirely possible that IF delamination exists, and IF an aircraft encounters a high enough trigger force (within "normal limits", but higher than "normally encountered" forces) to the delaminating component, then yes, the component may fail and depart the aircraft.

[ 18 November 2001: Message edited by: McD ]

[GlueBall]

Picture of delamination of tail section.

http://faceweb-004.facelink.com/edit/raw/rawimage/11/2332411.jpg

Photo resized for server load.

[ 19 November 2001: Message edited by: Checkboard ]

[GlueBall]

Vertical Stab attach points null http://www.ntsb.gov/Events/2001/AA587/AA587_09.jpg

[Genghis the Engineer]

An interesting discussion.

Overstress through turbulence is certainly entirely possible, and it does happen. If you want proof, look at http://www.aaib.dtlr.gov.uk/bulletin/mar00/gbvna.htm

Aircraft structures and operating limitations are designed to ensure that an aircraft isn't overstressed by turbulence, but once in a while some poor blighter gets to the wrong end of the curve. In the case linked above, I know that one of the following aircraft had it's (6g !) mainspar bent as it passed through the turbulence.

Failures can be in one of three manners: -

- Catasrophic failure, that is it all comes apart. This is what happened with the composite structures.
- Plastic failure, that is bits of metal permanently bend. That's what happened to the aircraft following the Cuby.
- Fatigue cracks are started or enlarged, reducing the safe life of the aircraft. This happens all the time and is taken into account by the Engineers calculating the fatigue life of critical components.

The latter is the only one detectable by NDT, but it certainly can be.

In this case, well frankly I don't think wake turbulence would cause bits to fall off. Why? because in climb-out the aircraft will be slow, and there's not enough energy there to break things in a certified aircraft. What turbulence can do (and has) is cause a loss of control (I got rolled inverted in a Bulldog behind a 747 once).

So, my tentative opinion, having available to me very little evidence, is that wake turbulence did not cause this failure. Not because wake turbulence can't cause severe problems and loss of an aircraft, but because it's the wrong failure mode.

Having said that, this all assumes no pre-existing damage. At that point the picture changes. Hypothetically, if the engine mounts were misrigged, or had fatigue cracks across them, turbulence could have caused a compressor stall. This puts large torque loads on the engine mounts, and it is possible that the combinations could have causes engine separation. This scenario however the purest conjecture, please treat it as such, but it does indicate a possible sequence.

But turbulence alone, in the climb-out, no, I don't think it's possible.

G

[Shore Guy]

Any Airbus guys out there care to post some details on the rudder limiter system and yaw damp system of the A300? Specifically, authority of yaw damp system and mode of rudder limiting the aircraft was in at 220 kts. (?configuration - were flaps fully retracted).

If the vertical stab did begin to move with the failure of one attach point, the yaw damp system would attempt to keep the aircraft straight to it's max authority. There has been some conjecture that the pilots overstressed the tail in attempting to recover (hardly probable), but the yaw damp system could have provided the torsional loads to cause the ultimate failure of an already compromised structure.

[Cigarman]

I haven't seen any discussion about the training that A/A instituted in the mid-90s for unusual attitude recovery where aggressive use of the rudder was recommended during an upset situation. The recent report that the FDR indicated that the rudder jerked five times during the flight's final seconds with two of the movements being full throw, stop to stop is quite suspicious. Any A/A guys out there want to clarify their training and it's possible implication in this terrible accident?

[ 20 November 2001: Message edited by: Cigarman ]

[McD]

Yes, cigarman, I'll be happy to clarify some words/phrases you used: "jerked" and "full throw, stop to stop".

NEVER, (and I'll repeat it again for emphasis), Never does our advanced handling training at AA include "jerk"-y or "full throw, stop to stop" control inputs.

The purpose of the advanced handling training is to show that CORRECT use of flight controls, during different regimes of flight, can provide a much better recovery of an aircraft from an unusual attitude. Since the effects of rudder and aileron vary greatly, depending on airspeed and AOA, the training strictly emphazises the absolute necessity for controlled inputs to the flight controls.

[ 20 November 2001: Message edited by: McD ]

[widgeon]

Glueball the failure mode on the photo you posted is a tensile failure not a delamination , lug failed at weakest point . The next picture looks more like a delamination failure with some of the material form the inner layers still attached to the lugs.

[Zeke]

Information from the cockpit and flight data recorders show the plane ran into two wakes from a Japan Air Lines 747 that took off 105 seconds before the American Airbus A300-600. The second wake occurred 85 seconds after the American plane took off and 18 seconds before it crashed.

http://www.airdisaster.com/news/1101/21/track.ap.jpg

After encountering the second wake, the plane experienced some side to side movements, which gradually got stronger and coincided with movements of the rudder. The board is investigating the rudder movements. The rudder and the tail fin fell off first, followed by the engines.

http://www.airdisaster.com/news/1101/21/tail.gif

The NTSB is looking at whether the tail was weakened in 1994 when the aircraft encountered turbulance while flying to Puerto Rico. The turbulence was so severe that 47 people were injured. American Airlines President Donald Carty has said he does not believe the incident caused damage that could have contributed to the crash.

[TechFly]

Thanks to everybody for the detailed information apported on the subject.

Do you agree if we say that the tail structure has been loosing part of the aeroelastic design characterics (i.e. lower resonance frequencies) and have been disturbed by the wake (acting as the dist. forces)?

I know for examples that BAe146 has big problems on the tail (pitch oscillation)where the left/right elevators became anti-symmetrically oscillating at high frequencies and the only solution is to reduce the speed.

My condolences to all the people that have lost their lives.

Cheers.

[411A]

When you think about it, 'tis not surprising that the engines separated after the fin departed...would think that the dutch roll would have been quite severe.

[ 22 November 2001: Message edited by: 411A ]

[UNCTUOUS]

Reading this previous incident report at http://www.aaib.dtlr.gov.uk/bulletin/feb01/n14065.htm , you get the impression that it's almost as if it's the B737 rudder actuator problem in reverse (i.e. instead of seizing "hard-over", suffering instead from an excess of motility).

How much does the A300-600 rudder limiter system restrict rudder travel when enabled (gear and flaps up)?

The DC-8s had about 7 degrees with gear and flaps up...15 degrees with both down.

Perhaps the rudder limiter system faulted (permitting the greater travel at 250 knots - as the autopilot coped with the wake turbulence upset) ??

It's beginning to sound to me like a hydraulic hammer may be induced by the rudder limiter-valve cycling rapidly (much like audio feedback can cause a superheterodyne squeal). When they hit the 747 wake, the Flight Control System would have made a much larger than normal rudder input to correct the yaw and perhaps set up a hydraulic reverberation in the rudder limiter valve line that caused the rudder's large lateral oscillations, thereby setting up a destructive rudder-induced flutter in the vertical fin. We've all heard the very noisy hydraulic hammer that you can get in household water-pipe plumbing. If you didn't turn the tap off quickly, you'd swear the wall was going to fall down. Due to the corrective input being from the autopilot (and not the pilot's rudder pedals) that may be a factor in the destructive hydraulic hammer being aroused between the limiter valve and rudder actuator. The biggest factor in the reinforcing (or damping) harmonic of hydraulic hammer is the distance between the two "chattering" hydraulic line components and the feedback harmonic that can be set up. Some hydraulic systems necessarily have Quincke valves incorporated - coils that are designed to soak up these types of destructive hydraulic chatters. The initiator of this rudder-induced flutter may need to be an external force (such as wake turbulence) requiring a rapid autopilot input (i.e. to say that normal rudder pedal input and pilot reaction times would not create the conditions for hydraulic chatter).