Thwartwise

in answer to a Pprune Forum Post here

1.   The A300-600 is typical of all largish twin jets inasmuch as its tail is sizeable and the rudder powerful. A wake vortex can admittedly be very persistent but at a distance of 5 miles or more behind, its wake-speeds will still be down by about 60%. That is to say that a 747 at 250kts will be laying a wake that at 5 miles astern has rotational velocities approximating 100 knots (and at 10nms astern about 40 kts max only). i.e. the rate of dissipation increases as you drop further astern. As you have discerned, it is the thwartwise component (only) that is of interest in ascertaining what the sideload effect might be upon an A300 fin.

 

2.   When you fly across a wake slightly transversely you are going to hit it (the core) with a significant part of your airplane. I know that full well from station changing in a formation aero team making smoke. If you took that hit on the tip of your vertical fin, well you were unlucky - but that's certainly where it can do the most harm for a number of reasons. What do I mean? Well think about it like this:

 

a.  Flying very obliquely across a preceding aircraft's wake, one that is essentially upon the same track as yourself, will give you two wake "hits". The initial one (from his starboard wing-tip's trailing vortex) will be followed by the inevitable next one about 2 to 3 seconds later (assuming about a 15 degree across-track heading differential). Both hits will be equally hard and they will be essentially upon the same bit of your airplane (here we might assume a worst-case of impact effect being upon the uppers of the A300 fin, such that the long moment arm wrench on the fin attachment points will be greatest).

 

b.  Should that initial wake impact be sufficient to wrench a serviceable (to design spec) attachment lug point apart? Not really. So what could do it? The timing of the aircraft flight control system's response reaction being coincidental with the next wake impact might just do it.

 

3. A Data Review - From my input to the Air Safety Week Article: "When the Average Isn’t Good Enough (and the Median too Mean)"  - refers to the A300's DFDR sampling and filtering - and the frustration of the NTSB investigation, because of it.

 

Timings

Four Complete Reversals Inside Seven Seconds

Travel 11º to right for 0.5 seconds.

Travel 10.5º to left for 0.3 seconds (first reversal).

Travel 10.5º to 11º to right for about 2 seconds (second reversal).

Travel 10º left for about 1 second (third reversal).

Finally, travel 9.5º to right before the data became unreliable (fourth reversal).

 

(Note: these are the last seven seconds during which the tailfin and rudder were well enough attached to give reliable FDR readings. The FDR shows four complete rudder reversals inside seven seconds, but the sum of the intervals shown above only comes to 3.8 seconds, and the travel time of the last rudder movement to the right is not at this time a matter of public record, if known. The last reliable FDR reading shows the accident aircraft in a left yaw of 8º to 10º.) Source: NTSB

 A great deal was happening in about a seven-second period in which four rudder reversals occurred inside a period of seven seconds. The accident aircraft was equipped with an FDR capable of capturing 167 parameters and recording 25 hours worth of information, but investigators say for all that flood of information, key items (i.e. data-points) are missing.

“The issue is not the number of parameters,” said an NTSB official. Rather, the sampling rates, in combination with the use of filtered data, may mean the extreme points in the Flight 587 accident sequence have been lost courtesy of the averaging function by which the data was recorded. In addition, while rudder pedal movement was recorded, the amount of force applied on the pedals was not captured. The data deficiencies have set up a situation where it may not be possible to resolve whether actions of the machine or the man (or a combination of man-machine interaction) caused such extreme aerodynamic loads that the tailfin separated from the airplane.

“It took us some time to discover that filtering [of the raw data] was going on, and how it was being filtered,” said the NTSB official. “Given the filtering, we can never recapture the exact motion of the controls and control surfaces.”

“Averaging will, by definition, tend to produce a value that’s less than the extremes,” the NTSB official said.

In truth, there are two aspects of the data clarity problem. The first is the rate at which the raw data are sampled. The rudder movement on the accident aircraft, for example, is sensed at a rate of twice per second. The movement of the rudder pedals is captured at the same rate. In the interval between sampling, extreme movements could have occurred in the accident sequence. One source advised that the flight control system (FCS) is capable of moving the rudder more than twice in the time that the FDR records one motion, and such rapid oscillatory motion may provide insight into the rattling noise captured on the cockpit voice recorder (CVR).

Thus, the sensing rate of twice per second is especially important in this case. “How good the data are depends on how often you sample,” the NTSB official said. The rudder is capable of moving at 39º per second, which means it could move about 19.5º between sampling intervals – which is a lot. As an A300 pilot explained, “Consider that the rudder limiter restricts the movement of the rudder to just under 10º at 250 knots. That would mean the rudder, at 250 knots, could conceivably go stop-to-stop and never be recorded.” (complete article http://www.iasa-intl.com/store/avionicsFDR.doc here - in Word format). But before we go on, just remember why the DFDR and cockpit data had to be filtered and sampling-rate slowed…… because the FCS system itself was sampling at a chatterbox rate.

 

4.  So para three above tells us what happened (as far as can be ascertained) - but how might we interpret it further? Firstly it is possible that, because of the low data-sampling rate and only 3.8 secs accounted for (out of a clock-run of seven secs), the rudder was actually wagging the tail at a much greater rate (i.e. building up to a max of twice as fast). In fact a freq breakdown of the CVR audio spectrum indicates that to have likely been the case. What does that remind us of, what aviation phenomena? Well, flutter of course. But flutter is a divergent undamped aerodynamic response of a flight-control surface that quickly slips out-of-phase. Aileron flutter for instance can be caused by vertical wing-flexing (and the aileron bouncing perversely around its hinge-line). Why not a PIO (pilot induced oscillation)? Wrong periodicity. In addition it is just not instinctive for a pilot to pedal rudder pedals because he is well used to the forceful damping effect upon turbulence-induced yaw of a large fin. It would be wholly contrary to a pilot's instincts. Besides, PIO's are usually limited to the elevator circuit. So if it was flutter, what sort of flutter - and what evidence is there for it? The answer = System-induced flutter – and the A300 has a revealing anecdotal and recorded history of it.

 

5.  Firstly, what else might be affected by a wake encounter – that might contribute to a system-led wind-up (aka destructive rudder reversals)? There are very sharp-edged pressure discontinuities within a wake vortex and these are not damped or averaged but fed directly into the CADCs. The CADC needs to make rapid inputs to the chatterbox processing of the FCS so it normally works with delta type info (linear trends - vice stabilised data). It’s likely that the correction for the wake-induced yaw was processed and sent and applied at about the 0.5sec elapsed moment that it took for the effect of the first slip to maximise (nose 11 deg right). But there were three things that may have compounded to make that first corrective rudder movement way off-base (and subsequent rudder-throws likewise inappropriate):

 

a. the feedback conflicts engendered by unfiltered input and filtered output. (leading to incorrect CADC interpretation of airspeeds) i.e. apply a filter to the output of a system that works on I/O feedback and you might get away with it if the environment remains non-dynamic. Any dynamism will beget a degree of exponential error (which in non-extreme conditions may still be fairly benign - such as a mild A300 tail-wag).

 

b. the trend-type operation of the CADC combined with the sharp-edged pressure discontinuities.

 

c.  the FCS calculation of a rudder response (to restore balanced flight) that didn’t account for the yaw-state of the airplane. Because the airplane had encountered wake (and hadn’t yet physically yawed) the rate gyros hadn’t sensed that the new relative airflow was only due to the thwartwise wake component - and so didn’t allow for the fin’s considerable damping moment. The resulting first physical yaw therefore overshot balanced flight significantly – and the scene was set for the second wake encounter to happen just as the FCS applied a larger 3rd dose of righting rudder (2nd reversal).

 

That was likely the starboard fin attachment’s initial breakpoint – quickly followed by separation (8 to 10 deg nose-left side-slip and rudder deflected 9.5 deg to right). After the first RH lug broke, relative motion (lateral rocking) of the loosened fin was possible, so subsequent fractures were inevitable (as the fin’s inertial overtravels at each yaw apex reinforced the effect (upon the fracturing attachment pivot-points) of the miscomputed rudder applications). Whether the rudder limiter was affected by the errors in CADC output or whether the yaw damper was a factor? Hard to say, but they are two ancillary systems that would not have benefited from any error compaction in a CADC output affected by a cycle of unresolved chaotic feedback.

 


Proof or example?

a. From the 12 Apr 02 NTSB Release:

“The Safety Board is interested in another upset event last year involving an Airbus aircraft. On November 25, 2001, a Singapore Airlines A340-300 departed Singapore for a scheduled flight to Dhaka, with 96 persons aboard. Shortly after takeoff, the pilots noticed a problem with airspeed indicators. Among other things, there were overspeed warnings and large rudder movements without pilot input. The aircraft returned to Singapore and made a safe landing; there were no injuries.

 

Inspection subsequently found problems with the pitot and static connections to the air data computers, which may have been introduced during recent maintenance. The Civil Aviation Authority of Singapore is investigating the incident.

“Inspection subsequently found…….” Well that is de rigeur isn’t it? Can’t have an investigation without an explanation can we? It may (or may not) have been a genuine case of ”problems with pitot and static connections” – but either way it completes the nexus between my theory, the CADC, airspeed (i.e. pressure) error and the rudder dancing to the tune of the CADC – does it not?

b. From the 25 Feb 02 - NTSB Rel

 Upon re-examination of the data from a 1997 event, the investigation team has determined that another American Airlines Airbus A300-600 (N90070) likely experienced high vertical stabilizer airloads. On May 12, 1997, American Airlines flight 903 was near West Palm Beach, Florida, when it entered a series of pitch, yaw, and roll maneuvers as the flight controls went through a period of oscillations for about 34 seconds, during which the aircraft dropped from 16,000 to 13,000 feet. The Safety Board determined that the probable causes of this incident were the flight crew's failure to maintain adequate airspeed during level-off, which led to an inadvertent stall, and their subsequent failure to use proper stall recovery techniques.

 

Conclusion

Wake event = pressure discontinuities (and so does a stall induce sharp pressure gradients – albeit not as sharp as in a wake event). The evidence for A300 system-induced flutter exists, it’s only the exact process that has yet to be resolved. The dynamics of stall and wake event airflows and pressure flux within a one-way filtered I/O is a good start-point. My 11 Feb 02 post on pages 2 & 3 (and later) of this Pprune thread complements this post.


Extract: (from prior Belgique Pprune Post)

These fins and rudders on the big twins are quite powerful because they have to accommodate OEI any old time. Courtesy of the system that tempers that power, the FCS is quite capable of breaking the fin off - both due to it being a composite structure (that won't just partially fail/fatigue crack) and because the range of rudder-throw action is large. At 250 kts it shouldn't exceed 9.3 deg, but if you fool it into thinking it's at 165 kts, you can have the whole 30 degrees of throw. Obviously that's going to be enough to break it at the higher speed. I'm not suggesting that it went full-scale deflection - just that induced static errors caused the ADC to continually re-schedule the FCS to permit inappropriate deflections for the actual airspeeds - leading to an aperiodic oscillatory flutter as each deflection proved inappropriate - yet gave rise to the need for a prompt further deflection (and thus becoming oscillatory). That's not overswing (per the FAA release), it's system error - where the rudder's power is pitted against the inherent dampening of the fin. At a certain point it will become exponential and destructive.

 

Quote

We all enjoy the powerful damping effects of yaw auto-stabilisers when they produce a very small force and apply it at the right moment to reduce the yaw oscillations. Now imagine a yaw auto-stabiliser that is working in reverse.”  John Farley

  See also RAINMAN on AA587 Crash Cause Theory

   Pprune Forums