Pushers, shakers and limiters
 

I have to go with PJ, TD and Cland on the crew procedures and following both untuitive actions and those in the "books". Still a mystery to me why the PF did what he did and why the PNF did not take a more pro-active role.

I only flew one jet with the pusher, and that sucker was a 28 pound forward force. If close to the "pitch-up", and handling the jet carefully, it literally took the stick outta your hand! Flew another with the rudder shaker and anybody that didn't think they were getting close to a stall by then was a sorry pilot.

So I don't think a pusher would have helped. But there's more.

The Cl versus AoA curve we see in many 'bus discussions is not accurate. The jet does not lose a lot of lift when the curve reaches the max. and more degrees AoA are required before the thing completely stops flying. Further, the jet seems to have very benign stall characteristics, so entry could be disguised by WX turbulence, mach buffet, spoiler deployment, etc. In short, lowering the nose with a bit of power seems to be a viable recovery technique ( assuming the crew realizes their predicament). No point on the pitch coefficient plot to show a "deep stall" such as we had in the Viper. The THS and elevators could have enabled recovery.

My last point on the pusher is what good is it if the FBW system ignores AoA?

Some really good points lately, and I am happy to have been given a seat at the table.

Clandestino;The bolding is mine. The > 10 second disagreement is the trigger for ALTN 2, and it seems that the NAV ADR DISAGREE would remain for rest of the flight.

I know the quote is from A333, but I don't believe this is one of the minor differences between A332 and A333.

mm43;
Yes, and for ALTN 1 as well. This change was made around 1997. Prior to this change, a momentary loss of valid air data would latch the EFCS in ALTN law for the balance of the flight. As far as I can determine, the only conditions which would cause the loss of AP 1 + 2 for the balance of the flight are the NAV ADR FAULT 1(2) + 2(3):

http://www.smugmug.com/photos/i-Nvvn...NvvnsPr-XL.jpg,


http://www.smugmug.com/photos/i-Z37F...Z37FZHq-XL.jpg

The only other possibility (only because it isn't clear if the AP is lost until landing) is the Abnormal Attitude Law, which if set pitch and/or bank angle values are exceeded, the laws revert to, in roll - yaw alternate, and in pitch - an adapted Nz law, without autotrim.

The autopilot will not engage without autotrim, (Use Manual Trim). After aircraft recovery, and until landing, the laws stay the same for roll but revert to Nz law with a recovered autotrim. I would expect the autopilot could be engaged depending upon other factors, (both elevators controllable, yaw damper functioning).

@PJ2... "The only other possibility (only because it isn't clear if the AP is lost until landing) is the Abnormal Attitude Law, which if set pitch and/or bank angle values are exceeded, the laws revert to, in roll - yaw alternate, and in pitch - an adapted Nz law, without autotrim."

I probably sound impatient or worse... I have read here that AP is lost until ground maintenance. Are you saying it is not yet established? From the report, with the new addition of the Captain's comment re: "Wait, the autopilot.....(sound of selector switch, CAM)" which he spoke very late in the descent, we do not know if the a/p was merely left on, or if the PF or PNF selected it, do we?

If the a/p is locked out after AL, it is probably not important, but if the a/p was reselected, the AD definitely says that the autopilot can come back on line with rectified NAV ADR disagree? Of course this would interest the investigation....or it's discussion....

PJ2;The FCTM is a CX one dated June 2005, and it seems strange that it made reference to the >10 sec ADR disagreement resulting in ALTN2 only.

The notable difference between ALTN1 and ALTN2 is the Roll Direct, and it is possible that the A/P cannot be re-engaged in that condition, i.e. PRIM requires a hydraulic reset on the ground. The RTLU can be reset with slat/flap extension below FL200.

Hopefully someone can identify exactly when and why the A/P re-engagement is inhibited.

EDIT :: My crib sheet has the following on Law reversions:-

Alternate 1:Alternate 1A:Alternate 2:Alternate 2A:Alternate 2B:

mm43;

Doh!...my expanded abnormals indicated the following:

NAV ADR DISAGREE
F/CTL ALTN LAW (PROT LOST)
Note: Following an ADR DISAGREE, detected by the PRIMs, ALTN law is latched. Resetting the PRIMs by using the pushbutton does not allow normal law recovery.

The latching only after 10 second was the feature introduced after 1997. I drew the conclusion from this and from another document which discussed the 1997 change, that both ALTN 1 and 2 latched. I've checked with my FCTMs as well and they agree with your comment that only ALTN 2 is latched. Thanks for the correction.

Interestingly, the following comment is offered in the CP FCTM:

Handling Characteristics

In pitch, handling remains similar to normal law.

In roll, depending on the failure level, control is either normal (ALTN 1) or direct (ALTN 2). In roll direct, the aircraft appears to be very sensitive and bank stability is no longer active.

Thanks mm43, seems that idea that AP was permanently lost stems from interpreting: 1. "This condition is latched until a PRIM reset is performed on ground without any hydraulic pressure." to be applicable to AP inhibit and not just to flight controls law 2. "the AP and A/THR can be re-engaged when the NAV ADR DISAGREE message has disappeared." to mean that AP is inhibited rather than "can" indicates it is possible yet inadvisable to engage the AP before ADR DISAGREE has cleared.

Unfortunately, comprehensive list of A330/40 UAS incidents, such as posted by Takata is absent from final report. It is obvious that all of the incidents involved blockade of at least two pitot probes, so not much difference from AF447 there. 1.16.2 of the report further refers to automation behaviour. ALTN law mostly remained latched for the duration of the flight yet AP and ATHR were recovered, there were even cases of engaging the autopilot while UAS was underway and following the wrong but agreeing air data.

Gums, pusher and shaker are only really meant for the aeroplanes that don't have acceptable natural stalling tendencies. If you don't need pusher - you don't install it as there's always risk of untimely activation. To mitigate it, pusher must push positively but not with such a force that it cant be manually overridden. That's the difference between alpha protection and pusher - alpha prot can not be overridden, that's why it is designed to shut itself down when there is doubt about the integrity of the air data feeding it and that's what we, very technically, call "alternate law". Perhaps tihs is the reason why it is apparent to you that FBW ignored the AoA - it got ignored because FBW is based on unintelligent computers, it detected that there is discrepancy between AoA and speed but it cannot see the whole aeroplane energy picture and has no memory to remember what was the aeroplane doing for last couple of seconds to make a judgment what is right and what is false. The decision must be made by the system both possessing and properly using intelligence.

Regarding the "flying near the edge" and "absence of G trim", I have to return to the teleological discussion regarding the passenger transport aeroplanes. G trim is useful on fighters, to set it to 1 when cruising to decrease the workload and set it to zero while maneuvering to achieve prompt unload with stick neutral. For aeroplanes designed to spend most of their lifetime in straight and level flight, it would be completely superfluous. Flightpath stability makes much more sense. As for the edge:

40 knots indicated for the aeroplane that makes only gentle heading corrections while cruising is actually quite enough.

That a shaker had on occasion been ignored, which resulted in stall, loss of control and subsequent fatal injuries and damage to the property is a solid fact.

That a stall warning was heard and heeded in some cases of A330/340 UAS is a solid fact.

That there were A330/340 crews who believed that stall warning is false but they didn't pull, kept AoA under control and lived to tell the story is a solid fact.

AP and ALT2B:
 

If ALT is caused by double ADR FAIL the AP is not available (see PJ2 #356) but after UAS the AP can be available.

One of the conditions to engage AP is CAS > VLS, For AF447 this means that after 02:10:46 (2 consistent and good airspeeds available) but CAS < VLS the AP could not have been reengaged.

In the 'air caraibes memo' AP2 could be reengaged after UAS while in ALT2x, although VLS was not displayed here, it must have been calculated by FMGEC and CAS>VLS.

@MM43:
 

A33ZAB

Turning that into english, both the AP and FD are inhibited unless at
least 2 IRU and 2 ADR are valid, amongst other things. If either pair
are invalid, engagement or reengagement of the AP would fail..

A/THR needs only the ADR condition valid of the above pair...

@syseng68k:
 

Agreed,

Invalid ADR is not equal to a failed ADR.

An ADR FAIL (announced by ECAM msg and local warning) is always an invalid ADR!
This ADR FAIL condition was not present in AF447.

To be exact:

AP needs AP/FD common conditions AND AP specific conditions AND AP operational conditions.
FD needs AP/FD common conditions only.

note:
Not completely clear in this - selfmade logic - is that [AP/FD & A/THR common conditions] are only required in managed modes.

There is a system, meat based, that at no additional charge and with no further engineering complexity, flies the plane. Going further into automation dependence seems poor path to follow.

As I read your translation: so long as the IRUs are up and operating, and the ADRs return to "valid" once the icing in the pitot tubes clear and airspeed input becomes reliable again, the A/P can be reset.

Did I get that right?

Side stick imputs.
 

Just one question; what idiot designed a system where by max nose down input on the left stick was canceled by max nose up input on the right stick?

What possible situation could arise where this feature would have any benefit?

I'm referring to the inputs as AF447 went through 10,000 feet.

Opendoor, I think I see a rationale behind summing / averaging the inputs by two SS: avoid potential over controlling (and possible damage) to the aircraft if the two pilots are both trying to make an input at the same time.

For example, if you have a 2.75 g limit, and the Ground Prox warning goes off, both pilots (even though one is flying at the time) may instinctively grab the stick and pull to the limit. If you allowed both inputs to add, then you have a 5.5 g pull. That may be a few g's more than needed, and potentially damaging, and might also induce an accelerated stall just when you cannot afford one. What needs to be done and trained to deal with the situation you cite is one pilot takes controls (I have the controls!) and the other pilot relenquished (you have the controls!) and ONE flies. It seems to me that the control logic is built to accomodate sound CRM principles.

Granted, not pulling enough could profoundly damage the aircraft via ground impact ... but I think the limit design tries to account for that.
(Aside: There was some very good and informed discussion of this in the Islamabad crash thread where they hit that mountain, over a year ago).

While the situation I cite may not be the best example, I think it illustrates the problem of two FBW inputs being used at once, inadvertently or not, without a filter to process two inputs when, ideally, the system is only ever supposed to have one man flying in the first place.

This follows good CRM practices, usually, except when you need one guy to save us all from the idiot trying to kill us in the other seat ... :eek:).

A33Zab:

This is, as usual, very interesting. A check of the STATUS page at any given point during one of these various UAS scenario sequences might pin down exactly when the A/P becomes available again, if in fact it actually does.

I'm having a problem with understanding that an A/P pb will physically engage in 'latched' ALT2 when it wouldn't engage during the period of time that the condition precipitating the latching was in effect.

If you were able to get the A/P pb to engage in ALT2, what would the FMA displayed lateral steering mode be in roll direct? Same as roll alternate? Blank?

To Lonewolf_50 and OPENDOOR
 

There's always room for improvement, whatever the make of the aircraft.

What about a "smarter" sidestick? Keep the averaging IF the sticks are going in the same direction; if they are going in conflicting directions, supress the non-priority input (null).

The Shaker
 

"That a shaker had on occasion been ignored, which resulted in stall, loss of control and subsequent fatal injuries and damage to the property is a solid fact."

From Clandestino....

You supply evidence (anecdotal) that supports a partial story. Shall I waste our time and bandwidth by supplying reams of data that support the opposite side? That the stick shaker is a life saver, is trained in approach to Stall syllabi, and is found on your very own aircraft?

You fly a turboprop, a T tail, I assume. That is a complex platform, and in challenging conditions, I propose that your aircraft so equipped, is a safe one?

My comments were meant to gain information from others, not meant to be argumentative. I said that in the type of Stall 447 experienced, and flown by a confused pilot, that the outcome may have been better if the stick turned into a buzzing snake in her pilot's grip? At the moment when the a/c exited the envelope? Further, the rest of the way down, he held back stick. Would he have done so if the stick started to actually shake his hand numb? Also, wouldn't the other two have noticed the stick shaking? Could the Captain's side have been included in the warning state? For three minutes, would the PNF have allowed the PF to continue holding the mushed attitude? I say things would have been markedly better, and though you may claim, "one in a million", would that cause the families of 228 dead people to agree with you?

The 330's Stall was entered in extreme fashion, quickly, and without the nose drop and falsely advertised Buffet from Stall that the Airframer sold the regulator on when begging for what you describe as "dispensation" from Shaker install.

I am amazed that such a seasoned pro would continue to say that the shaker would have been of no use.

"I'm having a problem with understanding that an A/P pb will physically engage in 'latched' ALT2 when it wouldn't engage during the period of time that the condition precipitating the latching was in effect."

From OK465

My friend, so am I. At the very least, even if the a/p did not return, why did the Captain make his comment? Also, if the a/p is latched out, common sense would indicate the logic would remove the selector from annunciating "ON". In a critical cockpit, let's agree that cleaning up all extraneous (and possibly dangerous) info would be a mind set for the designers of the panel.

Sidestick Inputs
 

What would you suggest should happen when one pilot pushes and the other pulls? How would the systems know which input to follow? What happens when one pilot pushes a conventional yoke while the other pulls?

Ignoramus that I am, my understanding is that on yoke-equipped aircraft resistance to their attempts would trigger recognition of a conflict and they'd either agree who has control or continue to counter the other's input, with the stronger pilot "winning". On an Airbus the trigger would be aural and visual warnings with the respective arm strengths not being a factor.

In either case, wouldn't continued wrestling with the the controls suggest a bigger issue than how the logic works?

"In either case, wouldn't continued wrestling with the the controls suggest a bigger issue than how the logic works?"

Yes, an issue that would be readily known by both pilots, in yoke equipped a/c.
Rather than unknown, and problematic, with SS's? This is the fundamental problem with SS, as interpreted in Airbus a/c, that both pilots can happily input commands, that are followed by the a/c, without knowing the other's actions..... Schizophrenegenic.

The corollary for SS would be that they are connected directly, such that each pilot would know what the commands are, and that they in fact emanate from his/her oppo. One stick could not move independently of its second/other. This is immediate and non verbal, which is currently non existent in the platform.

What's he/she doing?

Opendoor, I think I see a rationale behind summing / averaging the inputs by two SS: avoid potential over controlling (and possible damage) to the aircraft if the two pilots are both trying to make an input at the same time. LW50.

A rationale? Made necessary by something that should be impossible in the first place: Dual Control? A third and synthetic solution to a problem created out of mistakes made by two disconnected pilots? It is a palliative, made necessary by shortcomings of the design in the first place? Airbus makes a serious mistake in the very creation of the problem: they dilute and make a nonsense of what for a century has been critical to safe flight, in the interest of doing something merely different. that something? COMMAND. One pilot at a time. In making the pilots role secondary, they have fallen too far into the trap set by their own automation, to wit: an airplane cannot fly two ways at once. Offering a summed and synthetic solution to a confusion is not dangerous? I say it can be, and 447 demonstrates that one confused pilot can frustrate the mere possibility of salvation by two others in pursuing the path Airbus set: two separate control stations, with a computer that tries to resolve both inputs, when one is decidedly wrong.

A correct result cannot come from a computer that bases its own solution on two others, one of which is false.

In spite of the aural and visual "Dual Input" warnings?

On a yoke-equipped aircraft with one pilot pushing and the other pulling, which input does the a/c follow?

SS Take Over
 

If LHS PNF was convinced that he needed control all he needed to do was push SS forward while at the same time pressing and holding the take over button on the SS and saying, "I have control". If RHS PF does not respond in time honoured aviation fashion, "You have control", PNF keeps the TO button pressed.

If I remember correctly all the time the to button is pressed you have control and after 30secs holding it pressed the opposite SS is latched out.

Correct CRM, I have/You have, allows takeover without the need to lose the other SS as the taking over pilot can release the TO button before 30secs has elapsed.

Press and hold, until release is acknowledged, is taught for capts taking over in the event of a botched landing from the other seat!

In a co-operative handover, does a Priority button even need to be pressed? E.g. FO flying, Captain puts his hand on the stick, calls "I have control" or "my aircraft" or whatever, FO releases his stick.

On a yoke-equipped aircraft with one pilot pushing and the other pulling, which input does the a/c follow?

Hi DL-EDI....

Which input? There is only one, the columns act as one. The efforts of two pilots result in one command only. They are mechanically connected.


"If LHS PNF was convinced that he needed control all he needed to do was push SS forward while at the same time pressing and holding the take over button on the SS and saying, "I have control". If RHS PF does not respond in time honoured aviation fashion, "You have control", PNF keeps the TO button pressed."

Hi 1066....You presume that the system will be followed, when it has built in traps that depend on CRM. With or without CRM, a yoke equipped a/c cannot split command, mechanically, it cannot happen, and is independent of the sanity or skill of either pilot.

Well, that was kind of my point. :)

So what's the difference between that situation and the Airbus method of producing one command only, apart from not giving preference to the pilot with the stronger arms?

The "Airbus method" is not of one command only, like most of us know.

If you mean just the algebraic summation it is true, but the critical point are subtle inputs from either side, without knowing of the other pilot. And this happens day by day.

Howdy...

It is perhaps easier to consider that the two systems are offering two ways to input command. Both have two stations, one has split input engineered into it. In an arse about look, consider that the yokes are made smaller and put on either side of each pilot. That is architecture alone. Now, consider the yokes can be manipulated differently, with simultaneous and differing command, but to a computer?

In making everything computer dependent, Airbus has missed something, the need for Bulletproff CRM. The possibility exists that command is in some cases made murky by a design, which should never be the case?

Split input engineered into it? Why yes, else "summed solution" would not exist. Can one say the split input is an artifact of poor design? Without question. For that to be considered wrong, one would have to say split input was inadvertent. I think the designers might take offense....

Airbus scenario - One pilot pushes his stick to 100% nose-down while the other pilot pulls his stick to 100% nose-up.

Boeing scenario - One pilot tries to push the yoke to 100% nose-down while the other tries to pull the yoke to 100% nose-up.

Assuming both Boeing pilots have the same strength, wouldn't the net result in both cases be zero pitch input? Isn't it the case that neither system is deliberately designed to allow two pilots to engage in a prolonged fight for control?

I'm no aviation expert but, based on my IT experience, I'm not sure that algebraically summing two inputs is a particuarly challenging task for computers. :)

When flight path stabilized use the UAS checklist using pitch and thrust.

Level at FL350 seems pretty stabilized to me. The pitch up and climb power were meant for terrain clearance not for high altitude autopilot and airspeed loss. We can't help AF447 but maybe this investigation will help some other pilot that loses airspeed and autopilot in high level flight to not make the same mistake.

DL, that is what the palm of your hand is for. To smack the FO so he lets go of the wheel. Now you have control. Works every time.

Boeing scenario - One pilot commands 100% nose-down, the other 100% nose-up.
 

In the Boeing FBW planes (777, 787) where the yokes are not mechanically connected to the control surfaces, is there a take-over button to resolve the conflict?

Seems that FCS gives up restoring itself back to normal law after it detects long lasting ADR disagreement, however AFCS keeps checking indefinitely and will restore itself as soon as two ADCs agree, not necessarily at correct value. Page 86 of the report refers.

Probably NAV, HDG or TRK. Airbus AFCS is just an AFCS, only special thing about it is output; it doesn't go to autopilot servos but to FCCs. I suppose it doesn't care if its roll commands are obeyed by moving the spoilers and ailerons with constant roll rate or constant deflection. I suspect that AP doesn't work in direct law is because FBW has no control over pitch trim in it.

Yup, but they don't come about just because someone has wished for them and they never come without a price.

So are a parachute and a helmet but you won't see me wearing those when flying passengers.

On the aeroplanes equipped with it!

Because during certification, her natural stall characteristics were found wanting so they had to be supplemented by the artificial means!

Safe aeroplane with totally unsafe natural stall characteristics. How about that?

BEA actually performed the study to determine what it felt like as the AF447 approached the stall:

What are the chances that the pilots that ignored 1G amplitude shake as the sign of impending stall would pay attention to "buzzing snake"? I hope this lays the notion of stick shaker being useful on A330 to rest.

See previous entry.

With a little help from the red warning at the bottom of this page, your posts amaze me no more.

Depends on controls architecture. On some aeroplanes you may end up with split controls and each side flying just one elevator, completely independently of the other. Sounds fun, eh?

No.

It can split. It can happen. It is not independent, especially in the wrestling over controls cases. Priority button is very neat way of taking the control away from the captain Gonenuts.

That's just what they do but it is very, very wrong way to fly an Airbus. Only one pilot can fly her at the time, be it left, right or auto, this feat is just there to ease the hairy control handovers, before PiC clearly designates who will fly.

bubbers44:

Though in either type, that might favour a married FO, with a large wedding ring on his/her inboard hand... at least until they're back in the office... and it would only happen once. :)

kwateow:

They have inter-connected yokes.

Clandestino:

Thank you - I didn't think transferring control was as complicated as some think (no-one in particular).

Indeed but some seem to think there's something dangerous about the Airbus dual-input logic when, functionally, it's not that different. As I said, if you get continued conflicting dual-input then you've got bigger issues to worry about.

OK465:
 

AP was disconnected due to (dual) ADR loss, AS gains could not be established.
Once the (dual) AS returnes, the gains can be established again.

The AP has it own gains, rate- and amplitude limitations build in, it don't need the NORMAL LAW or whatever remaining protections in ALTERNATE.

AP commands a separate pitch deflection order for elevator (DQ) and THS(DQT) to FCPC in control.
For roll command there is a separate roll order(P) for ailerons and spoilers.
Yaw control command is separated in (Ytrim) and (Ystab).
All are deflection! orders.

After 02:10:46 it would return in HDG or TRACK because that are the only SELECTED lateral modes available.
AFAIK with current HDG or TRACK as default.

A33Zab: Thank you, sir.

Should the STATUS page eventually reflect availability of the A/P (by exclusion) even though ALT2 initially latched with AP 1+2 displayed as INOP on the STATUS page?

Is the displayed A/P STATUS and availability updated automatically or does it have to be cleared?

(edit: thanks clandestino also, and is roll direct an FCSC function in this case? additional edit: It occurred to me that doesn't really matter if the commands are deflection orders. Thanks again.)

DL, you have to survive so fighting control over the aircraft is not an option. The PIC normally takes control in a conflict because he is in command. Sometimes the FO has to take control if the PIC is going to kill everybody. I only had to do this once as an FO on a 4 engine corporate jet and forced a go around.

I assumed I was fired but the chief pilot thanked me for saving a totally botched approach on a short wet runway. We were too high to not have an overrun so I went max power and went around on his leg.

A33Zab re Post #362

Many thanks for the logic schematic which certainly is an improvement over that described in FCOM / FCTMs.

With the logic shown, the conditions for A/P engagement are clear, and I assume that the A/P Auto OFF conditions, i.e.will take precedence once the A/P has been engaged?

@OK465:
 

As a general rule ECAM messages and INOP status are cleared automatically when the fault or inop conditions are not active anymore.

If you allow me,

No, roll direct is part of ALTERNATE (ALT2x) control of the FCPC.

FCSC ALTERNATE YAW/DIRECT control becomes active if all 3 FCPCs failed to control in DIRECT.
To set that in perspective:

Priority:
1 = FCPC #1 in NORMAL
2 = FCPC #2 in NORMAL
3 = FCPC #3 in NORMAL
4 = FCPC #1 in ALTERNATE (active @ AF447)
--
7 = FCPC #1 in DIRECT
--
10 = FCSC(#1 & #2) ALTERNATE YAW/DIRECT

Do pilots follow unknown FD commands and if so for what reason? Isn't the checklist saying 5 degrees nose up and climb power until at a stabilized altitude which they were at at FL350 and use UAS and thrust charts to continue. Seems they were already stabilized.

@mm43:
 

Correct, there are small deviations in the values to engage and the values to automatically disengage.

Yours are indeed the disengage values.

to use your format:

Not exactly true on advanced airliners - remember the EgyptAir 990 case highlighted an unforeseen problem with the 767's interconnection, namely that when one pilot pushed and the other pulled, the result was split elevators and immediate departure from controlled flight.