AA 587. Systems problems? (thread #4)

[Cyclic Hotline]

NTSB Reveals Flight 587 Failures

By JONATHAN D. SALANT, Associated Press Writer

WASHINGTON (AP) - Two components that help pilots control an airplane didn't work during a preflight check of American Airlines Flight 587, the National Transportation Safety Board reported Tuesday. The plane crashed last month soon after taking off from Kennedy Airport.

The maintenance log reported problems with the pitch trim, which helps keep the nose in an up or down position, and the yaw damper, which uses the rudder to keep an airplane from swaying.

The problem was corrected when a mechanic reset the computers that control the components, according to the log.

Aviation consultant Jim McKenna said safety investigators will focus on those components of the Airbus A300-600. "They'll take a very close look at it,'' McKenna said.

The NTSB said that the vertical stabilizer - or tail fin - and the attached rudder fell off the plane, as did both engines. The Nov. 12 crash killed all 260 people on board and five on the ground.

Following the accident, the Federal Aviation Administration ordered inspections of the tail, which is made from lighter-weight, nonmetallic composites.

The NTSB is conducting ultrasound and other inspections of the vertical stabilizer and rudder, and is developing a plan for additional tests of the composites.

Maintenance records showed that the vertical stabilizer and rudder last were inspected visually in December 1999, and no problems were reported, the NTSB said.

Board investigators reported that they found no evidence that the engines broke apart, sending shrapnel into nearby control systems, nor any evidence of a collision with a bird. There also was no evidence that there was a fire or a malfunction.

NTSB investigators again reported that they have found no evidence of a terrorist attack. All the evidence continues to indicate that the crash of Flight 587 was an accident, the board said.

[Volume]

After lots of speculations, most based statements of so called ´experts´ we returned to a more technical point of view.

Nice to speculate on thrust reverser deployment, eventually caused by sabotage, but the very first NTSB report stated ´No evidence was found of an uncontained engine failure, loss of blades, bird strike or in-flight fire. The thrust reversers were in the stowed position.´ so let´s forget about this theory.

What we have are the pictures of the fin, the rudder and the fuselage attachment brackets on
<a href="http://www.ntsb.gov/Events/2001/AA587/tailcomp.htm" target="_blank">NTSB-Report page</a>and <a href="http://www.aviationnow.com/content/publication/awst/20011203/avi_air.htm" target="_blank">Aviation Now</a>and our knowledge about loads, structures and aerodynamics.
This tells us, that the 6 main attachment points failed in at least 3 different ways (maybe 4). The two front one failed in net tension of the eye, the mid right and aft left failed in the unidirectional carbon fibre skin of the fin itself, the mid left failed in tension at the uppermost rivet row of the repair patch. We don´t know about the rigt aft in detail, but it is for shure an eye failure, maybe in shear because no remainders are left on the fuselage bracket.
If there is a design flaw in the composite attachment points, why did they fail in different ways ? This points very strong to an overload above ultimate strength of an otherwise good fin.

The rudder and fin are seperated, they were found in a certain distance in Jamaica Bay. This tells us, they were already seperated when hitting the water. The (much heavier) fin is nearly undamaged, the (much lighter and therefor much slower impacting) rudder is broken. It is very unlikely that the rudder broke on impact, so it must have broken while connected to the fin or while seperating.
If the complete fin and rudder assemby broke of the plane, would they seperate in free fall ? Very unlikely, the forces that could act on such a free falling part are small compared to the attached condition because they are compensated by inertia forces (Sir Isaac Newton tells us so) and not taken by the attach points. So the rudder must have seperated or at least became strongly damaged while the fin was still attached to the fuselage.

The rudder is broken in two pieces, a large area of debonding is visible on the sidefaces, the trailing edge failed in an area of about one rudder chord, the lower piece is also damaged at its lower end, maybe due to contact with the fuselage. The failure in the rudder´s front spar is clearly visible right below the uppermost lightening hole, thats between the upper and the mid rudder actuator.

Some members already discussed about previous incidents and accidents where tails ripped of, what about cases where the rudder ripped of ?
There are 4 very interesting incidents on the Concorde. (remember, the Concorde was in some respect the predecessor of the Airbus, a lot of engineers were the same in both projects) for example <a href="http://www.aaib.dtlr.gov.uk/bulletin/dec00/gboac2.htm" target="_blank">AAIB-Report 1</a>or <a href="http://www.aaib.dtlr.gov.uk/bulletin/dec00/gboac1.htm" target="_blank">AAIB-Report 2</a>.
The pictures don´t look to different from the failure of AA587 rudder, the failure started at the trailing edge and resulted in delamination of the outer skins. There are no ribs at the actuator attachment points in an A300 rudder, this means all the loads are introduced into the sidepannels (in tension and compression) and distributed to the upper and lower part in shear. If the outer skin is debonded or the trailing edge fails in the area of the actator attachment, the sidepannels are no more able to carry compression loads, resulting in much softer attachment chracteristics and reduced ultimate loads.

We already discussed a lot about system failures (actualy its the topic´s title), so could a softer rudder caused by trailing edge damage due to wake encounter loads affect the stability of the control system, especially the yaw damper ?
Systems specialist, please think about this.

One last interesting point the pictures tell us : All 3 attachment brackets on the fin broke of the rudder´s front spar, they all look the same, they are all alligned in the center position. Does anybody know more about Airbus hydraulic control system ? Does the yaw damper act on all of the hydraulic actoators, or only one one ? Are there springs in the system that center the actuators in case of hydraulic pressure lost ? Does the actuator position tell us, that the control system was working normally ? Is it possible that the upper and mid actuator worked against each other (this would explain the location of the rudder front spar failure) are they connected to separate control systems ?

At last a link to three nice pictures that show how a wake vortex of an B747 can look like
<a href="http://www.aviationpics.com/pretty/m-smoke1.jpg" target="_blank">Picture 1</a>
<a href="http://www.aviationpics.com/pretty/m-smoke2.jpg" target="_blank">Picture 2</a>
<a href="http://www.aviationpics.com/pretty/m-smoke3.jpg" target="_blank">Picture 3</a>

have a nice christmas folks

[Plastic Bug]

The rudder did not disbond, it came off. That's important.

The engine did not go into reverse. That was already identified. The nitwit "expert" who claimed that a sabateur could nip the hydraulic line to the reverser should be bound and gagged.

The CF6 does not use hydraulics to activate the reverser.

period

What are the odds of choking to death on your tooth brush in the morning? Multiply that number by about 6 giga zillion and that number reflects the odds of a tail coming apart.

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.

PB

[Cyclic Hotline]

Another related article <a href="http://www.newsday.com/news/yahoo/ny-liplan222519883dec22.story?coll=ny%2Dnewsaol%2Dheadlines" target="_blank">here.</a>

[RatherBeFlying]

NY Times article reports no pre-existing damage found <a href="http://www.nytimes.com/2002/01/05/nyregion/05CRAS.html" target="_blank">here</a> (You may need to register for free to see this article).

According to the article, the investigators are now focusing on pilot inputs (and likely flight control system responses along with how that loads the structures).

It sure looked a lot like pre-existing crack propagation to me from the pictures, but if the fractures are shiny (or whatever you see in a fresh composite fracture) all the way through, it's an overstress.

[cwatters]

&gt; The rudder and fin are seperated, they were
&gt; found in a certain distance in Jamaica Bay.

This sounds interesting. If the rudder departed
first would there be enough fin area remaining to
retain directional stability? I guess probably
might be.

What would you look for to work out if the rudder
departed first?

[arcniz]

I've started a couple of messages on this - and then demurred because the only way to put the logic together is to wander into hypotheticals that muddy public discussion.

So, leaving out the analysis and stating only my long pondered conclusion (aka 'guess') about how this could have happened, I agree with Plastic Bug that bad luck was likely a big factor, and with Volume that progressive mechanical deterioration after a brief violation of the envelope probably was the mechanism.

I think it plausible that the entire failure process may have derived from the second wake encounter alone, and that the basic failure could have happened within a second or two after the wake hit. The combination of a very strong initial sideways push on the tail, plus some resonances in the airframe, plus some out-of sync responses from the yaw-damper and other controls collectively may have caused the tail to rebound through centerline so fast as to cause vtail stall eddies that fluttered the rudder, promoting a prolonged oscillation of the physical and electronic controls that fairly quickly tore up the rudder and in so doing horsed the tail around so much that it succumbed shortly thereafter. Any significant control input from the crew would likely have sought to damp uncommanded rudder excursions, but the rudder might well have been chaotically unstable in the flutter mode and thus inclined to change dramatically in pitch with only minor control stimulus.

If this is what happened, a bit more delay in the control responses to very sudden displacements would be a plausible fix to avoid recurrence. And a bigger / better flutter recovery algorithm.

Whatever the nuts and bolts, it's all terribly sad. May they rest in peace.

[stuartbaker]

Volume: FYI the A300 has a mechanical link from the Yaw Damper which goes to all 3 rudder actuators (each one servo assisted) via a spring rod. The autopilot yaw actuator also has an input. The YD actuator, the AP actuator and rudder pedal inputs are all "upstream" of a variable stop lever which limits mechanical travel input depending on speed.

Makes FBW sound like a doddle!