NTSB Recommendation re Airbus Rudder Travel Limits

[Machinbird]

BOAC

- I'm still intrigued by these thoughts - how do you envisage the motion of 587 itself causing the co-pilots legs to push on the pedals - do you see him sliding around in his seat? Why did the first wt encounter not do the same?

587 to me exhibits a simple, classic PIO. The only query I have is why did it start.

Why did it start? First lets look at what Hess said in his report regarding the subject of Triggering events:

Triggering Events
PIO incidents are almost invariably preceeded by a "triggering" event [3]. A trigger is a
stimulus that can cause a pilot to change his/her control behavior. Triggers have been categorized as (a) Environmental Triggers (such as those that involve changes in the vehicle dynamics that cause a mismatch between pilot control strategy and the aircraft dynamics), and (c) System Failures, (such as the failure of an actuator or hydraulic system).

Triggers can cause a pilot to move from non-tracking or low-gain tracking behavior to high-gain tracking behavior. For example, a sudden and large turbulence encounter can cause a pilot to actively begin high-gain, compensatory attitude tracking when previous to the encounter he/she was only monitoring aircraft trim or making low-gain corrections to vehicle attitude.

Of and by itself, a triggering event may not be a catalyst for a PIO. Typically, some flight control system property conducive to a PIO is revealed when high-gain behavior begins. A good example of this is the PIO that occurred in the Shuttle Orbiter Enterprise in October of 1977 (Alt-5). This flight involved the Enterprise being carried aloft on a Boeing 747, then released to make a landing at specific touch-down point on a concrete runway at Edwards AFB (previous landings took place on a dry lake bed). The triggering event here could be described as an Environmental Trigger associated with the stress of attempting what amounted to a spot landing. High gain pilot tracking activity then began that involved large and rapid pilot inputs in the final segment of the approach. The combination of large time delays in the flight control system coupled with the aforementioned control inputs caused the vehicle's elevon actuators to rate saturate or rate limit. This means that the actuators were moving the elevons as rapidly as their designs permit. The intrinsic time delays constituted the control system property conducive to a PIO. The rate saturation dramatically changed the vehicle dynamics by introducing even larger time delays into the control loop. A PIO in both lateral and longitudinal axes ensued. The PIO was terminated when the pilot-in -command released the control stick, i.e. completely "backed out of the loop".

It has been hypothesized that a true PIO will involve the pilot adopting a "regressive" form of tracking behavior marked by the control of error-rate rather than an error itself. For example, if a PIO in the roll-control axis has begun, the pilot will regress to control of roll-rate rather than roll-attitude. Once this regressive behavior has been adopted by the pilot, it is difficult for the pilot to "back out of the loop". A sustained PIO is likely. Often at this point the pilot believes that something is wrong with the aircraft, i.e., that a failure has occurred. As far as the pilot is concerned, the aircraft is behaving strangely. Some pilots who have survived serious PIO encounters have said that they simply "no longer recognized the aircraft".

As I mentioned before Hess seems to be looking at AA587 from a control rate limiting perspective, There may be another way to look at the APC forcing function.

But first, a short story about my attempts at driving my brother’s 1967 Pontiac GTO.
I attempted to accelerate the vehicle in a normal manner by letting out the clutch and applying a little accelerator only to be embarrassed by a jerky heel toe type acceleration profile. When this happened a second time immediately afterward, I began to realize that there was more at play than just a clumsy operator.
A few experiments revealed that while starting in first gear, the seat back yielded enough to pull me off the clutch and the accelerator together. As I my weight left the clutch pedal the clutch engaged smartly which further increased the acceleration and completed pulling me off the accelerator. But now the deceleration caused by no throttle moved me forward and the throttle was applied again and the clutch disengaged. As long as I attempted to accelerate this jerky charade continued. The fix was simple…Start in second gear.
This is a simple (and I hope easily understood) linear single axis DIO (Driver Induced Oscillation) that resulted from accelerations upon the “crew”. The enabling factor for this oscillation was seat back rigidity incompatible with the accelerations created on the crew member and by a clutch that engaged over a very short pedal travel distance. The initiating event was starting the vehicle in first gear.

I have a few experiments anyone can try in the safety of your own car.
Consider the case of cornering in a vehicle. As long as the turn is in one direction, you settle into whatever support you can find and ride out the turn. If the turn rapidly reverses, you (and your passengers) are now accelerated in the opposite direction. If the forces (speed) is low, friction holds you in place and nothing much happens. With higher forces (speed) your upper body displaces to the opposite side of the seat to find whatever purchase it can and particularly if you are a passenger, you use your legs to help support your body. At still higher turn rates and forces, your posterior slides to the opposite side of the seat until it finds support. The key event as far as your body is concerned is the reversal of force from one direction to another. If this occurs rapidly and unexpectedly it is difficult to steady your body and you use all your appendages to steady yourself.

The following is a generic pilot seat support assessment based on various seats I have ridden in: Your favorite aircraft seat may be different.
Subjective support levels:
1. Good support against positive g events,
2. Fair support against negative g events,
3. Good support against linear acceleration,
4. Fair support against linear deceleration,
5. Fair to poor support on lateral acceleration depending upon whether there are arm rests, location of anchoring straps, and how much the seat width exceeds your actual width.
My expectation is that lateral support in aircrew seats is going to have the worst level of performance.

I suggest you do the car experiment yourself and observe what your turn coping strategies are. The driver in the vehicle has forewarning of the maneuvers and can hold onto the wheel so the best results would be with you as passenger. Observe how you use your legs. Do you use your right leg to brace yourself in a left turn (and vice versa)?
I am going to have to stop at this point. You won’t hurt my feelings if you tell me I am all wet after having run this experiment. I don’t expect 100 % consistent results, but you should be able to see where I am going with this in the context of AA587.

[BOAC]

Machin - I am not disputing your scenario in general - it is the application to 587 I query. Your 'car ' scenario is well known and we have the term 'Kangaroo Petrol' in the UK.

My post " The only query I have is why did it start. " refers to my uncertainty, having looked at the 'conspiracy' theory of 'US Read', as to the actual 'timeline' and what triggered the initial pedal inputs. The FDR trace shows rudder broadly in line with c column, so I do not see your 'physical' motion at play here.

[Mr Optimistic]

Perceptions

If the rudder effect was much more than the pilot expected or knew, the building motion would just have been interpreted as the effects of worsening wake turbulence wouldn't it, ie requiring even more 'corrective action' ? Once the mental picture was set, would there have been any perceptive feedback to alert him that he was a contributor ?

[bearfoil]

Wake turbulence rotates, and for all the "off course" reaction to be in Yaw only is a stretch, something I've not seen addressed, though early on an animation was seen, it required a very specific relationship between the 300's tail (specifically VS) and the rolling vortex from the 747. If the flying pilot bought into wt initially and through the event, wouldn't other controls and thrust settings be in evidence? The author of the USRead piece was more than slightly convinced of his timeline, discounting the wt and drawing attention most emphatically to the call for max power. Something seems awry, just me?

[DozyWannabe]

At an even higher level, it's also worth remembering through this theorising that even the best pilots (and other professions) have off moments - and in the vast majority of cases we end up with little more than bruised pride, and a salient lesson to take more care in future. In this case the PF was an a truly unfortunate nexus of events that yielded tragic results. Whether there was a structural issue or whether it was just a mistake that escalated into a loss of control the results were sadly the same.

The good that has come out of this includes a revision in training practices, along with considerably more attention paid to the ageing properties of composite aircraft materials. It would be a shame to let the negativity and in-fighting overshadow that. This is not the first accident that has split the piloting fraternity in terms of what they believe the fundamental cause is, and it's unlikely to be the last.

Whether the first "hole in the cheese" was a result of First Officer Molin making a mistake, or something technical that precipitated those rudder movements, I'm sure - judging from the personal recollections of those that knew him as a conscientious professional - he'd want the advances in safety and knowledge to be the paramount legacy of the crash.

[Mansfield]

DozyWannabe: It can't be said any better than that.

[p51guy]

Hopefully after 9 years Sten Molin, the FO, is on the first step to being not the cause of the accident but a victim to rudder control engineering faults.

[Ct.Yankee]

p51guy;
I totally agree with you!
It's time to stop blaming a dead pilot!

[p51guy]

. AA 587 had also encured inflight turbulence that injured dozens of passengers 6 years prior and no inspections other than a visual, which we know are not valid now, were done. Airbus probably put pressure on the NTSB to blame the copilot until litigation against them could be erased under the statute of limitations law. I always felt Sten was the fall guy from the beginning. Why did it take 9 years to tell Airbus to fix the rudder system?

[p51guy]

My theory is the vertical stab broke off possibly because of previous damage in 95 and the patch made when manufactured where it broke and caused the severe yaw that separated the engines from the aircraft but they haven't gotten that far yet. Give them another 9 years.

[BandAide]

The thing is, though, the A300 had flown and continued to fly uneventfully with its design rudder system for years.

I was corrected a few pages ago for thinking the A300 had the same autotrim as later Airbi, but the basic principles of rudder application apply regardless. You can't employ rapid, aggressive rudder reversals on big jets, no matter the material compostition of the structure or the manufacturer. The structure won't withstand it. That AA 587's flying pilot didn't know that is basic to the cause of the accident.

Further, a wake turbulence encounter at AA 587's altitude did not require the drastic correction taken. Escape is easily accomplished with an slight upwind turn or altitude level off. Rapid full rudder reversal was not called for, and never is.

Remember Dutch roll on B-707s and DC-8s? The counter was a brief full aileron against the oscillation, neutral controls, then a brief full aileron against the return, opposite direction oscillation. Too much aileron, or any rudder at all accelerated the oscillations. The lesson of the modest control inputs applies to wake turbulence as well.

I think blaming Airbus for faulty design is wrong when the real cause was improper pilot technique, for which I do not fault the pilot as his training curricula overlooked the well known problem of excessive rudder application in opposite direction.

[DozyWannabe]

p51guy, Ct. Yankee:

At this point the evidence is exhausted - there is not and never will be definitive proof one way or the other. "Designed differently from some other manufacturers" is not the same thing as a "design fault". In some cases certain aspects of an airliner's design could be considered non-optimal and require special handling, but that's something that is true of pretty much any airliner you care to name.

I don't think anyone with any degree of self-awareness or compassion would go so far as to out-and-out hold First Officer Molin solely to blame for the accident. Regardless of the controversial issues, contributing factors include inadequate training and insufficient understanding throughout the airline of the important differences between the types they flew and techniques relevant to each. Regardless of whether the rudder design or a mistake was the initiating event, the fact remains that he was spectacularly unlucky in the sequence of events he was given to handle, and there but for the grace of any deity you care to name go many of us.

[EDIT : p51guy - the FDR trace, correllated with the CVR and all other data available at the time, proved beyond any measure of doubt that the vertical stab failed at a level of stress well in excess of its ultimate design load - in layman's terms it outperformed the boffins' expectations - and the failure happened *after* the initial severe upset, not before. Also, Airbus Industrie have no clout whatsoever with the NTSB (and neither for that matter do Boeing or any other manufacturer - see the history of the 737 rudder issue) - it is an independent agency. Airbus couldn't apportion blame even if they wanted to.]

[p51guy]

The MD80 works with trim tabs to control the control surface. What if AA587 lost it's rear lugs on the VS and the rudder was giving opposite commands of yaw. Right rudder, left VS, giving the rudder the function of a control tab? Could you control an airplane on a moments notice using right rudder to yaw left? Just a thought, I have always thought the front lugs failed first which would cause the same crash but in a different way. Rudders alone, I don't think so.

[DozyWannabe]

A separation of that magnitude would have been easily identifiable on the CVR, or at the very least the NTSB would have considered the sequence of events as ambiguous based on what they heard. No such sense of ambiguity made its way into the transcript, so while what youre suggesting is certainly possible, it would be considered extremely unlikely.

[Mansfield]

The Board often behaves in mercurial ways, and, in my opinion, is perfectly capable of going down the wrong rabbit track. That said, they have some of the better metallurgical and structural people in the industry, and, although I have no first hand knowledge of this accident, I would expect their work to be rather solid in this regard.

Regardless of all of this, I will say that based on my intimate experience with Franco-NTSB relations during the Roselawn investigation, the Board is very unlikely to be influenced by French pressure. Indeed, it is far more likely to yield the opposite result.

As far as the certification limits on vertical stab loads, I had no idea prior to this accident that the stab was not structurally designed to handle rapid rudder reversal. Once I read the rule in Part 25, it was painfully obvious...I just never looked at it before, and I certainly never heard it discussed in training with any airline.

This among many other cases has reinforced my belief that a genuine ATP license should require a comprehensive education in both Part 121 and Part 25 rules (and/or the EASA equivalents, depending on your country of issuance), including the means of compliance acceptable to the authorities and the history of accidents and failures that promulgated the rules in the first place.

For example, how many of us realize that the flight director pitch guidance during a go-around is predicated on the engines being spooled to approach power at the initiation of the go-around? The Air Canada crew at Fredericton apparently did not; they followed the pitch guidance promptly, even though the engines had been at flight idle when the go-around was initiated.

We have been fighting this battle with regard to icing certification for years. Many pilots continue, with substantial help from the authorities and the operators, to conflate holdover tables for freezing precipitation with certification to operate in same. Not the case at all. But beyond that, the accident history is laden with cases in which the pilot did not understand the criteria to which either his airplane was certificated, or to which the approach aid was certificated, or to which his ops specs were approved, and so forth.

Perhaps a little less "unencumbered solutioneering" in various training departments, and a bit more thorough exploration of the regulatory criteria and constraints, would go a long way in this regard.

[Slats One]

Disbond of my belief

After reading this thread focus mainly on rudder ratio and related control systems and human inputs, I am now forced to ask:

Are we going to now consider the events of 2005 when an A310 of Air Transat -
flight 961- routing from Cuba to Canada on March 6, 2005, lost its rudder in
flight and experienced Dutch Roll and some flight profile deviations- followed
by a successful landing back at its POD.

I am unaware that anyone blames the pilot flying it, for dubious rudder pedal
inputs - nor is any such suggestion extant. But I am aware that the rudder came
off, and upon subsequent inspection, two of the main fin fuselage- to -fin attachment points had reportedly been damaged and that the fin was left hanging on by less than its normal number of attachment points (3 each side, half alloy, half composite and glue....).

Stress fractures or fluid ingress damage near hinge points were framed as being
the rudder's issues- but few looked at the main empenage's issues....

I believe that this (subsequent to AA 557) incident may relevant in re-evaluating AA587 and Mr Molin’s actions.

Of note:

The Air Transat A310 rudder failed and departed the airframe – leaving only a
few below hinge sections remaining. The reasons for the failure are linked to our lack of
understanding and lack of monitoring knowledge and procedures for the ageing
composite airframe.

‘Tapping’ the fin – as is done on the ramp - and doing a quick ultrasound are not really going to reveal the true state of an ageing composite and the deep layers of its build, cure, bond, resin state and delamination, let alone decide if the current ‘acceptable’ standards for build and resin faults are safe.

There are known existing problems with the manufacture of large scale hot bonded aerospace structures. Several incidents of manufacturing bonding and curing errors took place- leading Airbus to repair/rebuild new build tail fins pre-delivery.

The significant issues are these;

(a) It is hard to inspect or quantify failures in curing (autoclave) and bonding during new build.

(b)Such failures are well known and well documented.

(c) Leading composite researchers have argued for years over the technique of creating a 'peel' during manufacture. Such chemical peel of the composite, can create a non suitable surface on the peeled item once removed - any structure that is then bonded to it can suffer from disbonding etc over time and effects thereof.

So, what are we doing to properly inspect the resin states, lamination, gel
coats, molecule attrition from oil based contaminants and the effects of water / fluid ingress and freezing expansion at high altitudes and low temperatures?

(As an aside -can anyone recall the anecdotes of water allegedly pouring out of Airbus rudders on the ramp after arrival from warmer climes via high level, circa -59 degrees cruise?)

Yet the key reveal of the Air Trasnt A310 event, was the alleged damage to the fin attachment points.

And what are we doing now, all these years later, to address the design and
structural issues that may or may not be apparent in the design of composite rudders and vertical stabs and their crucial attachment point design.

Does the Air Transat event, (and subsequent events with the Air Canada Airbus
fleet subject of a recent statement), have any relevance to the previous incidents?

What other large series production airframe has had such a fin attachment design
and build, and what other airframe suffers fin attachment damage after losing
its rudder and suffering Dutch Roll?

IE: Why did Concorde’s main fin not come off each time part of its composite
rudder failed and broke off in flight – with resulting very severe vibrational harmonics affecting the fin structure? And did you know that the first composite rudder (made by Ciba
Geigy) testing (that is rudder only) was on an RAF VC10 over two years in the early 1970s - a
wonderfully trouble free record over thousands of hours.

What structural differences exist between Airbus tail fin design and strengths
relevant to differing thrust options within each relevant per model type? Answer
? Is it: None?

Why are factory repairs to new cure composite structures allowed? AA587 had such
did it not. And how can adding a strut brace to a composite structure devoid of
a load path carrying chassis, not do anything but expectedly fail at its reinforcement point - that is always going to happen...

Why is the Airbus fin attached with an alloy fitting down one side and a direct
composite fitting down the other- whatever the design loads, whatever the fact
that the AA557 fin failed at beyond its design load, this type of design is
always going to concentrate stress in a small area and at differing resistance
rates - is it not?

Even ff we accept that Mr Molin was using the rudders as evidenced, why do some
blame him solely for that supposed single action - why do we not ask why he was
using the rudders as alleged- was it due just to his own construct, due to the
AA training programme, the rudder ratio issue - or did something happen that was not considered at the time but may, repeat may now be focused by subsequent events - contribute to his actions. Was there a rudder issue not cited then, but apparent now from more recent events? Was there a fin attachment issue? Was there an ageing composite
issue ?

I guess we will never know. But we do now know that the fin is not intended to handle a certain type of rudder action.

Let us remember that in the 1960s a BOAC 707 crashed over Mount Fuji after a
turbulence hit- and the fin failed beyond design limits- just as the AA 587 fin
failed beyond design limits.- so, no, I am not having a 'bash' at Airbus. But
what I am asking is this - does the record of the Airbus fin design, have
a hidden factor only told by the advent of the ageing composite airframe - or
our subsequent knowledge of how to design and build large composite fins- and
crucially, how to keep them attached?

How many more pictures of fins floating in the water - as in AA557, AF 447, and
the NZ 'unexpected turn' Perpignan test flight, will we see?

It has taken years, but are we now at a point where anecdote might indicate a
consistent pattern of evidence?

FYI: I am only asking. The opinions stated above are just that- they are not claims of actuality nor are they evidence. They are just obvious questions after a very long wait.

So shoot me down now why don't you? Someone will....

Meanwhile, what are YOU doing to research the issue of the ageing composite and the legacy of early composite design issues?

Air Transat Ref:
Report Number A05F0047. Transportation Safety Board of Canada(TSB). 2007-11-22.

[safetypee]

Training ?

Re Machinbird post #161, also Mr O #163, B Aide #171.
It would appear that the accident may not fit classical descriptions of PIOs, but it is probably sufficient for the NTSB’s discussion.
A wider ranging view is that for a deficient system, more often than not, all pilots will suffer a tendency to couple with the system. However, with only a weak system (high geared rudder) the tendency might be ‘trained out’.
There have been few reports of problems with this type of rudder system in normal operations, noting that normal operations may only require occasional and different methods of use; e.g. rudder input with engine failure does not result in a PIO, similar with the landing drift correction.

Thus as an alternative view, if the crew had not experienced the rudder sensitivity during training (experience which provides a datum for future operations), then they could have been surprised by the aircraft’s response when the rudder was used in wake turbulence.
Add to this, the lack of side force (lateral acceleration) in simulators, then the crew could have lacked key elements of the control loop normally provided by training, e.g. damping any tendency to over control during a first flight, or first time driving a new car.

Thus, due to training weaknesses, perhaps accentuated by poor advice of how to react to wake turbulence, the accident crew were actually learning to fly the aircraft in the lateral axis. This may have been more difficult due to the normally tolerable, but relatively high rudder sensitivity.

The view above may reflect some of the EASA’s comments about ‘training’; how much can be assumed about pilot behaviour when operating a system.

In a system involving a man – machine interface, either end, man or machine can be the source of problems (blame) and cures. However, the actual link between the two is very complex; perhaps we should be looking more in this area, and perhaps EASA has made a start.

[DozyWannabe]

How many more pictures of fins floating in the water - as in AA557, AF 447, and the NZ 'unexpected turn' Perpignan test flight, will we see?

I wasn't aware there was a yaw component to the Perpignan test flight - in fact I'd be surprised if there was. We don't know if there was on AF447, but the structural investigation indicates that the vertical stabiliser separated on impact. As I understand it the design of the vertical stabiliser/rudder and it's mountings are completely different across the A300-A320-A33/40 series, so any link there would only be tenuous at best.

It has taken years, but are we now at a point where anecdote might indicate a consistent pattern of evidence?

I'm a cautious enough person to never say never, but my opinion for what it's worth is that it's very unlikely, for the reasons above among others. The famous pictures of the TWA800 wreckage at the surface were not the first parts to separate. The fin in the pictures from the wreckage of the Air India bombing was also not the first to separate. On the other hand, the venerable 737 had been in the air for 24 years before the "perfect storm" of a PCU valve worn in a specific manner combined with extreme low temperatures exposed a design flaw, and it took a non-fatal incident in the wake of two, sadly, fatal accidents before that design flaw was exposed.

The Air Transat incident caused a lot of focus to be put on potential delamination issues, and that focus was compounded by AA587. As a result new inspection procedures are in place and a much more rigorous fatigue monitoring programme has been implemented, as well as all A30/10 training programmes making a point to explain the differences in the rudder system. None of this would have been the case if the accepted version of events simply pinned it on First Officer Molin and left it at that.

I understand the desire to "clear the name" of one of your own, but I don't think there's any name to clear - as I said, it's understood that many pilots put in that position may have acted in the same manner had they had the same training. First Officer Molin was sadly in the wrong place at the wrong time.

[BOAC]

Originally Posted by DW

I understand the desire to "clear the name" of one of your own, but I don't think there's any name to clear - as I said, it's understood that many pilots put in that position may have acted in the same manner had they had the same training. First Officer Molin was sadly in the wrong place at the wrong time.

- my feelings exactly, but unfortunately there are, as you can see here, those experts who KNOW it was him wot dun it. It is not so much 'clearing the name' as preventing the tarnishing in the first place, I feel?

[DozyWannabe]

I think that says more about the qualities - or lack thereof - of those doing the tarnishing than it does about the one being tarnished.