It would also be interesting learn how much lag there is in updating the PFD display in the A330.
Machin(e)bird:
Here's the info from AC-25-11A, Electronic Flight Deck Displays:
from page 47,
Quote:
Data update rates for information elements used in direct airplane or powerplant manual control tasks (such as attitude, engine parameters, etc.) equal to or greater than 15 Hertz have been found to be acceptable. Any lag introduced by the display system should be consistent with the airplane control task associated with that parameter. In particular, display system lag (including the sensor) for attitude which does not exceed a first order equivalent time constant of 100 milliseconds for airplanes with conventional control system response is generally acceptable.
In practicality there is not even a remotely discernible lag in relation to actual aircraft response on modern PFD's or HUD's.
I flew the old 'Vital 4' visual system on the A-7 simulator in the mid 70's, probably equivalent to your A-4 sim visual. Those were so bad, many people would develop a persistent dizziness bordering on nausea for hours at a time after the session. Our rule was you could not fly the actual aircraft on the same day after completing a simulator training period.
(I think if you grabbed hold of an Airbus SS you'd be somewhat underwhelmed by the roll responsiveness in roll direct, even at altitude. But indeed no doubt this can be a relative thing depending on background and training.)
In particular, display system lag (including the sensor) for attitude which does not exceed a first order equivalent time constant of 100 milliseconds for airplanes with conventional control system response is generally acceptable.
Thank you OK465, that is useful data. By itself, the display lag should not enable an oscillation. About what I was expecting.
Quote:
Originally Posted by OK465
(I think if you grabbed hold of an Airbus SS you'd be somewhat underwhelmed by the roll responsiveness in roll direct, even at altitude. But indeed no doubt this can be a relative thing depending on background and training.)
I would personally expect somewhat ponderous response compared with the tactical aircraft and smaller transports that I've flown.
In Bonin's case, he had insufficient time to appreciate he was no longer in Normal law when he made his initial roll correction. I can see how an assumption of being in Normal law could set him up for a roll PIO in Alt 2 law. (His initial input would be 'open loop'.)
Just because an aircraft is quite large is no reason to discount PIO. They PIO quite nicely judging from the record, particularly along the roll axis.
A330, if hand "fistively" flown on approach has a tendency to pio. It is even worse at high altitude. You have to fly it gently & accurately. Then it is quite nice to fly in either normal,alternate & direct law.
IcePack, how come you know but Bonin (and maybe Robert) didn’t?
I’m not sniping at you or them, just pursuing a line I have before – should Bonin have had more training in hand flying, how could such training include high mach and altitude, did you explore this somehow on your own or has training changed, how much latitude is there for hand flying in line flying, etc.
Location: Quite near 'An aerodrome somewhere in England'
Posts: 20,178
Quote:
Bonin's "stirring the mayonnaise" technique...
Isn't that phrase actually a rude French euphemism for....well, something else which is totally unconnected with flying?
AF are anxious to prove that the accident was caused by something other than their poorly-trained pilots who clearly didn't understand the basics of flight. But try as they might, unlike their Concorde accident, this time they won't be able to.
...as I understand, that stick deflection in the 'Bus is essentially commanding a rate, not a control deflection. Zero control deflection equals zero rate.
Thought I'd say hi.
Yes, you're correct. No input = no output in terms of pitch and roll. However, the controls themselves will be moving, albeit in tiny deflections, to maintain the last ordered bank angle and pitch attitude.
Quote:
As such you would be pulsing the control rather than applying a pressure.
Not sure what is meant by "pulsing", but what is required to fly the Airbus series well and smoothly is a gentle, steady pressure...a "squeezing" of the stick in the desired direction, keeping in mind that ailerons and elevators are also moving to maintain the last ordered attitudes even while that attitude is changing as per stick orders...this is especially the case in turbulence*.
In Normal Law the sidestick is a roll-rate request in roll and a 'g'-loading request in pitch. In Alt II Law the autoflight system is in Roll Direct Law, or WYSIWYG Law... ;-) and the control deflection is proportional to the stick deflection, (and therefore one must be very gentle with the airplane especially at cruise altitudes). Pitch remains in a modified 'g' (load factor demand) law but,
"...with limited pitch rate feedback and gains, depending on speed and configuration."
The roll is sensitive in Direct Law and, with mis-handling, (or more bluntly, ham-fistedness), prone to PIO but I hasten to add that it is readily controllable, (positive, not neutral stability). Letting go the stick would be a good first response to any PIO in roll, (I understand the slight right roll of 8deg, but it is inexperience that induces the PIO, not the airplane. In fact, (and it has been discussed ad nauseum), there was no need to pitch the airplane up at all and roll control was/is straightforward).
The pilot who videoed himself flying the A320 should be embarrassed in demonstrating his atrocious "technique" to the world. The key is tiny movements, always, even on approach. Pushing the stick around like that would get a reviewed-standard on any sim ride as that means one doesn't understand fbw or the Airbus autoflight system very well.
On response, OK465's post highlights what I would have to say about displays and airplane response...neither are significant issues.
Quote:
Since he was now in Alt 2 law, neutralizing the stick did not stop the roll rate (It was now a 'conventional' aircraft in roll response with somewhat higher responsiveness to stick input.) He had to then make a corresponding opposing control motion to stop the roll.
Yes. And over-controlling (leading to PIO) is easy to do but equally easy to stop. To mm43's comment, ". . . one has to wonder whether it is the visual search for reaction clues, rather than real SS feedback that leads to the "stirring" technique. ", it could be, if one is using outside (VMC) references, but not in my experience either in the aircraft or the sim in terms of the response of the PFD.
The fairly rapid response of the A330 in Roll Direct doesn't (or shouldn't!) lead to a loss of control but with inexperience and/or lack of training in actual hand-flying it can be momentarily disconcerting. Icepack is essentially saying the same things.
Someone asked about hand-flying experience... Typically, international crews get around 3 to 5 hours of hand-flying experience and perhaps 25 landings on average per year**. Domestic work will get substantially more...about 65 to 90 landings a month and much more hand-flying...difficult to estimate how much.
*When/if one has the opportunity to take a look at some flight data traces for an approach on a windy day, one can see how "calmly" the Airbus autoflight system moves the controls..."just enough" to maintain pitch and roll attitudes. The traces are very interesting once the autopilot is disconnected. The amplitude and frequency of the aileron (and even the spoiler) traces increase, the elevator less so, but higher than with the autopilot engaged. There, a lag in response (because the airplane is such a large mass), tends to make one believe that the airplane isn't responding to one's SS orders, so one 'gives more'...(done it...doesn't work!), and then over-controlling does become a bit of an issue if one isn't careful.
**Long-haul flying schedules yields about six flight legs per month, possibly eight or nine at the most. Split between two, possibly three pilots (depending upon whether the third is another F/O or is an RP who only sits up front during cruise), that gives about 3 takeoffs and landings per month per pilot.
Each takeoff is hand-flown but the autoflight is engaged pretty quickly after takeoff especially where departure routes are complex. Climb, cruise, descent and approach phases are typically on autoflight with the autopilot being disconnected around 400' AGL, sometimes a bit sooner, though usually not much later.
The takeoff and approach/landing phases are hand-flown for around 2 to 5 minutes at takeoff and about 30" to 1 minute on the approach and landing. At the very most, airline pilots flying long-haul may get 6 to 8 minutes of low-level hand-flying per leg.
Three legs per month means perhaps 20 minutes of hand-flying, or about 200 minutes (just over 3 hours) per year of handling the flight controls.
Again for domestic the opportunities are vastly greater and more difficult to estimate but my guess would be around 80 to 100hrs of actual hand-flying per year for busy domestic schedules.
Hand-flying is rarely given/taught/provided for in recurrent simulator sessions which is a shame because it is the only real opportunity to try one's hand especially with Level D simulators providing the opportunity to practise visual approaches.
chrisN, all aircraft i've flown i always want to see how they handle at altitude so i don't get a big surprise one day. I think everyone should hand fly a 2000 ft level change at 38000 ft +every now and again. Sim's are no use as they handle the same at 5000 ft as they do at 41000 ft. which actually i find criminal, as the authorities don't insist the algorithms being programed. Also these days the snitch box (QAR) will tell on you for taking the automatics out at height, but luckily I fly for a decent airline that will turn a blind eye to the ah hem the tech log knocking the auto pilot off. As for Bonnin, i do not wish to talk ill of the dead but i expect he had none or very very little stick time at high altitude.
[crosswinds] even in gusty conditions (like here, apparently)?
Crosswinds while flying 2,500 feet over Rio? And if that's a crosswind landing, it the most stable, dead-nuts straight down the centerline, wings-level one I've ever seen...
They are so many strange things in this upset, that we really cannot judge the situation without the ORAL CVR : The first question to be answered is why the BEA refuses absolutely to give it.
Sim's are no use as they handle the same at 5000 ft as they do at 41000 ft. which actually i find criminal, as the authorities don't insist the algorithms being programed.
I would change your simulator manufacturer and/or change the tech pilots/engineers your company sends out to the vendor for factory acceptance testing.
In addition, if possible, change your monitoring authorities for good measure.
@ OK465
" [...] change your simulator manufacturer [...] the tech pilots/engineers your company sends [...] for factory acceptance testing. , [...] your monitoring authorities"
What are the odds of requesting a block altitude and getting it like we used to do in the military? Wouldn't need to do it very long. Even 10 minutes worth would be significant.
Perhaps with a little planning, you can get your hand flying time in that way.
Re, "Sim's are no use as they handle the same at 5000 ft as they do at 41000 ft. which actually i find criminal, as the authorities don't insist the algorithms being programed. "
Are you certain of this? There are areas in which simulators do not have data (behaviour at/during the stall, discussed in earlier threads and verified with CAE) but I would have thought that in Level D sims at least, aircraft behaviour at high altitudes vice low altitudes would have been easily reproduced and even required. Just one example..., one may pull hard on the controls, (here, the side stick), and may approach or even exceed the stall AoA in thinner air while at 5000ft that likely isn' t the case. I would think that would have to be reproduced accurately.
CVR !
