In Daze of Yaw
Not much is being said about AA587 since Airbus imposed its will upon the last hearing. The pilot error theory
is being supported by the NTSB and endorsed by the FAA. The glitch is more likely (IMHO) to be in the original
design (or introduced later by a software patch)._
And I still like my original theory (below):
So, (in part) addressing these selected quotes:
A. "A good look at the DFDR data does not show rudder movement consistent with a mechanical problem." and
B. "Even the AA flight 903 incident occurred in violent weather and an inflight upset that may have been a stall
or may have been something else."
In Daze of Yaw
The AA587 explanation may be as simple as water getting into the pitot-static lines every time an A300-600
aircraft is washed (or is parked in torrential rain) - and introducing a pressure change damping effect that
cannot be accommodated by the CADC’s pneumatic “expectations” (particularly in yaw). Why would that (i.e.
trapped water) make a difference? It’s not as if it might freeze in the lines and cause glaring static errors
(in airspeed, altitude and rate of climb). The errors might be much more subtle than that. In fact, in normal
flight they might be hardly noticeable. But in strongly yawed flight it could be entirely different. Why would
that be?
Explanation
_The input to the CADC’s is (ideally) purely pneumatic pressures being mildly differentiated by height and/or
speed variations. The CADC as a sensor is capable of picking up very minute pressure changes and rates of change
of pressure variation (as well as temperature increments of course). It feeds the resolved heights, rates of
climb and (of particular interest and significance) current airspeeds to a range of systems – including the yaw
damper and rudder limiter. _If _water is inadvertently introduced into the static lines (say) you then have a
damped and laggard hydro-pneumatic response in all the variables that feed pressure changes into the air data
computers (which will in turn misinterpret these “damaged” inputs and thus apply incorrect outputs). Normally
there will be four or more static ports (holes) – with a couple each side (of the airplane) to guard against
blockage (of a singular one). The siting of those holes is designedly where, in normal symmetric flight, the
smallest PEc (pressure error correction) requirement is generated. Some aircraft have the static ports right up
the front on either side of the nose and some aircraft have them way down back, just forward of the empennage.
Being thus located either well forward or well aft of the aircraft’s vertical (i.e. yaw) axis, the effect of
yawing needs to be cancelled out (and that’s why you have the ports uniformly on the port and starboard sides of
the fuselage). Works well, but how will water trapped in the lines affect the sensed pressures?
When the aircraft yaws due to turbulence (or pilot-pedal input) any airflows injected dynamically into the
static lines are self-canceling because the port and starboard static ports are Y-pieced together. Net effect on
the CADC’s sensed pressures? Not a lot, although extreme yaw can cause airspeed indications to fluctuate wildly.
But what if water has collected at a low point in the static lines – perhaps before the port and starboard
pickups are Y-pieced together?? (even if only on one side). In a sharp yaw to port any water in a starboard line
would tend to flow aft (and water on the port side would tend to flow forwards). What will be the effect of that
on the air-pressures sensed by the CADC and, more importantly, its outputs? At this point I should also
acknowledge that some modern systems (but probably not the A300) do incorporate air pressure transducers that
convert what is sensed to a digital signal that then goes to the CADCs. But whether those signals were valid or
not would depend on the location of those transducers. Because the CADC senses both quantum change and trends,
there will be at least a lag and more likely a contrary signal generated by the adverse flow of water affecting
the line pressures. Any correction fed to the rudder actuator by the yaw damper might then be inappropriate in
both magnitude and direction. So where is this error headed? Will any initial wake-induced yaw be damped out or
magnified by the yaw damper in this scenario? Will the erroneous CADC signal cause an initial significant yaw to
overshoot equilibrium and, in fact exacerbate the L/R yaw cycle? (i.e. "set the ball rolling" for some
rudder-supported extreme yaw cycling).
My theory says that YES, if the initial externally-triggered yaw is large enough, it could do that. If an
inappropriate correcting rudder deflection is commanded because of water trapped in the static lines and the
wrong airspeed being sensed - and the yaw correction consequently overshoots significantly, then we are into a
rudder-inspired undamped phugoid around the yaw axis. And beyond that “threshold of significant external yaw”,
because of the great stabilizing influence of the large vertical fin, any out-of-phase rudder control inputs
would be amplified and rapidly approach the point where something has to give (structurally). And that is what
I’m guessing may have happened in AA587. It might explain why much anecdotal evidence of tail-wagging is quite
irregular – possibly because static line water trap drains are cleared out on (just guessing) each C service (or
might be even more regularly)._ Then the aircraft's unalarming (but irritating) tail-wag phenomenon would
disappear overnight without comment. That would explain why particular airframes don't get (and keep) a bad
reputation.
So the theory really says: "You'd need to unluckily combine a waterlogged static system and a wake turbulence event (with
its large amplitude yaw “kick-off”) to end up generating an AA587 event".
In my scenario is there any scope for pilot complication or compounding of the wildly yawing initial scenario
(prior to structural failure)? The easy answer is: “Of course”. No pilot is going to sit there and allow the
airplane to swing wildly from side to side. He will at least attempt to counter the yawing with judicious rudder
pedal inputs (at least I would). But whether he could be successful (or just get out of phase himself and
exacerbate the situation) is the real quandary. Would the other pilot be aware of any such intervention on the
part of the PF (pilot flying)? Not really – and the whole thing would happen so fast that there’d be little
chance for anything further than the gasped expletives heard on the AA587 CVR.
How could you test this theory? The easy answer would be to “water” an A300-605 and go out and see whether
atmospheric disturbances or pilot pedal input could bring about any strange unexpected yawing responses. Another
approach would be to monitor the static-line water-traps across the fleet, measure the amounts of water found
and note whether aircraft washes (or heavy rinses) increased this amount.
Addenda
To specifically address this input by ALPA: (particularly the observation in blue)
<<"As I understand it, the A300 -600 was being hand flown at the time of the upset._ So the autopilot
would not have been engaged._ The A300 is not a fly by wire airplane._ It is conventional hydraulic controls so
the FCS electronics would be very limited._ The yaw damper would be active but the rudder
swings far exceed the authority of the yaw damper (it is mechanically limited)._ So, based on the DFDR, I do not
see any airplane issues based on flight control malfunctions.">>
Apart from the sampling issue with the AA587 DFDR meaning that the number and size of AA587's rudder deflections were not determinable, you
would not need max deflection rudder authority for a yaw-damper-actuated rudder to achieve large yaw amplitudes.
It's the timing of the actuator inputs that would be critical. I used to do a low-level jet aero display that
incorporated extreme yawing cycles on a flyby. Rudder authority was such that the last thing you wanted to do
was to stall the vertical fin - so you needed a keen sense of timing when "walking" the rudders somewhat
gingerly so as to achieve peak yaw in each direction. It is quite possible to achieve extreme yawing angles
(talking about > plus/minus 30 degrees here) by just getting the timing right. In addition to that, do not
disregard my opinion above - that any pilot (PF) would instinctively intervene (and perhaps disastrously) once
an identifiable cyclic yaw was underway.