jcjeant

Hi,

That's unfortunate ...
I'm sure others will find better .. more brains .. more fun

You never noticed that the wheel is reinvented every year ?

CONF iture

Clandestino,
What I can see here is that you want to mix two things over and over :

1. Flying eyes closed can guaranty a LOC and you are perfectly aware of that. (average time needed seems to be around one or two minutes only ...)
2. But exiting a dive, to set things more obvious, be it in clear sky or in the clouds, can be done in many ways. From 10 deg ND to 10 NU can be down the hard way, the confortable way, or a variation of both. Only inputs will make the difference - Those inputs can vary on purpose during the manoeuvre depending what the pilot feels and is ready to feel. Wings are level. Sensation can absolutely be part of the elements that will guide adequately the inputs.

What don't you get here ?
Be explicit on your disagreement ...

I would not disagree as the altitude margin was confortable this time, but nevertheless, I see AF447 as a compulsory trigger to modify the UAS procedure to something that makes sense in any circumstances. What would or could happen if the 5 deg NU + CLB thrust is applied when the aircraft for any reason is already around the recommended max or simply at the FMS optimal a day when that optimal is pretty close to that recommended max ?

I like that 2.5 deg + N1 between noon and one (or whatever is the reference and general figures for cruise versus eng type) ...

Lonewolf_50

JD-EE ^^ It appears from CVR excerpts that such was not the case.
I don't see the "climb away from clouds" as an instinctive reaction, since his initial control problem looks to have been controlling roll.

The more likely explanation is that as soon as he ran into that sloppy roll feeling, and was oscillating back and forth in roll, his hand got a bit tight on the side stick and he was pulling while he was rolling.
see the chart in post number 342, airtren,
perm link = http://www.pprune.org/tech-log/46062...ml#post6656578

My own personal experience with upset and being behind the aircraft on instruments (albeit not in heavies) is that when one is out of trim, it is easy to get a tight grip on the stick (Thought foremost is "GET BACK IN CONTROL, MAKE IT DO WHAT I WANT IT TO DO!") and end up fighting the aircraft a bit until closer to power / attitude / performance desired.

That catching up period (particular event I am recalling was a trim runaway) takes time and you really have to force yourself to get back to the basic instrument scan: attitude, performance, attitude performance. In the AF447 cockpit, the standard scan was not able to reference a key performance crosscheck, which is airspeed. The lack of that cross reference means that his scan (practiced and habitual) would not be serving him as usual: what he was looking for to inform a correction was simply missing. (Granted, his attitude should have told him, his PNF did tell him ... but I digress).

A number of members have pointed out over six threads the problem of dutch roll being a characteristic of swept wing jets. This may have contributed to his roll control problem, which raises tension (MAKE IT DO WHAT I WANT IT TO DO!) and result in inadvertent back pressure, back displacement, on SS.

Pilots who have flown A330, and who have posted on these forums, have pointed out that at high altitude, the aircraft requires a light touch to hand fly.

Occam's razor suggests a case of overcontrolling, even if the weather was indeed a concern for the crew during that phase of flight.

HarryMann

Is there really any pressing need to suppress it once the a/c is off the ground and once above a nominal (chosen) speed, until it's deactivated again upon landing

Yes, the AoA vane can malfunction (we know that), but its extremely rare, there can be cross-checks for that condition (e.g. a trouble shoot test soon after take-off could be applied)...

and additionally, why are more cctv, extrenal web-cam style devices, not used so the crew can - at a glance - inspect all around the aircraft for actual physical configuration and incipient or extant problems.

I seem to remember suggesting that such a (potentially low-cost) camera installation would likely have help avoid the Madrid take-off configuration accident. Sometimes the aircraft industry seems overly conservative, despite being knwon as a high-tech leader.

I think the industry should make its mind up how much accidents really cost and what that worth is in much better thought through systems and safety double-checks.

The Mk I human eyeball seems to have been sorely undererstimated as a troubleshooting device... ?

airtren

HazelNuts39,

Indeed, the graphs at page 41 and 42 with their granularity illustrate better the variations - thanks for pointing that out, and explanation.

Following the cause/effect chains a little further:

a) Pitch command -> Elevator action -> Pitch Attitude

The "integrated" angle of the Elevator NU is roughly 1 degree, perhaps less, in the time interval 2:10:08 - 2:10:16 in which the pitch attitude goes from approx 3 to 10 degree NU, which is a 7 degree delta.

b) Thrust -> Pitch Attitude

The thrust is on a decreased level, on a downward gradient variation rather than upward, if any gradient variation.

c) Elevator -> Normal Acceleration

The Normal Acceleration show 2 consecutive aggressive spikes of roughly 0.6-0.8g excursion upward/downward, corresponding to two spikes of roughly 1 degree in Elevator NU.

d) Turbulence -> Normal Acceleration

We don't have a graph of turbulence alone; it would be interesting to see.

But the variations in Elevator angle later on the time scale on the lesser granularity graphs - not present on page 41/42 - seem to correspond to lesser amplitude Normal Acceleration variation.

If that were indeed the case, would that leave room for the interpretation that the higher amplitude of the Normal Acceleration spikes between 2:10:08 - 2:10:16 indicates a turbulence contribution to that of the Elevators?

Not that it matters much in the larger perspective of the sad things happening during those 2-3 minutes....

MurphyWasRight

Seems obvious to me so I must be missing something:

Once stall warning/condition is detected for more than 5 seconds (or other appropriate small number) latch the warning and do not remove it until valid (unstalled) airspeed and AoA are seen.

In other words once in a stall prove to me that we are no longer stalled.

Oh and while we are at it flash STALL on all displays every 10 seconds or so.

HazelNuts39

The graph on page 42 compares the recorded normal acceleration to that calculated from the control inputs without turbulence. That should give you a fairly good impression of the turbulence contribution.

Lyman

ICING of the Pitot Probes as some explanation for UAS is a reach. It requires too much of a chaotic system by way of lined up cheeses.

At ~2:10:01, 447 enters a substantial UPDRAFT. Immediately, she does what all large (conventional planform) aircraft do when encountering a robust UPFLOW, she PITCHES DOWN, and DESCENDS. She has a LEFT WING displacement UPWARD due to TANGENTIAL ENTRY into the rising column. airtren's accel trace shows this at 2:10:01 - 2:10:05 from 1.3 down to .7.

This accounts for her aspect (NOSEDOWN) at a/p loss, and why the PF felt he had to ROLL LEFT, and CLIMB.

What about UAS? takata tells us the AD's three failed simultaneously. If so, that really makes icing suspect, but makes clear the loss of indicated airspeeds due to a consonant loss of AIRMASS to enter the Probes. All three started sampling ambients that had changed heading relative the airframe, and by 90 degrees. So the indications declined by > 30 knots in one second, and the ADs were rejected.



The PULL? The NOSE was down by three off 0, and 5.5 off cruise. That's nearly six degrees to make up, and he was too swift, STALLSTALL. Also, the a/p was actively inputting NU to counter the PITCH down. So PITCH is iffy at the a/p drop, and PF needs to fly, and he did. No judgment.


There is more, as I see it, and it is in airtren's graph.

Lonewolf_50

That can explain the initial nose up, if the aircraft was not Straight and Level at the point of "you have it, robot taking a timeout, old chap."

It does not explain subsequent nose up inputs that lack successful counter-corrections toward a more level attitude.

mm43

Now explain why AoA vane #1 got stuck!

A seagull embedded on it?

GerardC

Pilots who, actually, flew "ALT law" at high altitude, reported difficult control of the a/c.

From what we see, bank angle control required VERY large ss inputs.
In this situation, it seems difficult to criticize large ss NU inputs by PF.

Imho, the initial nose up is a reaction to the loss of 360' (35,024'-34,664') in just 4 seconds (02:10:05 / 02:10:09) that is "only" a 5400 ft/minute rate.
Any pilot flying an RVSM route will react very quickly to this kind of altitude variation.

I am afraid we will never have the answer to that question.

OK465

No. Just the opposite.

(my bold required)

airtren

Indeed, it's a fairly good impression. I should not be that conservative in equating the real plane with the theoretical/simulated model ran on the Airbus labs computers, and the would be values, with the corresponding values calculated by that model, based on feeding/running it with the known AF 447 command and headwind parameters.

It's also not that difficult "to imagine" based on the page 42 graphs, the graph of the vertical component of the turbulence; if I have time, I may try to generate one for my own curiosity....

By the way, this is an opportunity to mention and praise BEA for its professionalism in having these simulations, calculations, and analysis done, as they are extremely important in understanding and confirming that the plane acted within its intended design boundaries.

GerardC

I respect your opinion, sir, but what about 4/5 ss deflection to counter a mundane 9° roll at 02:10:30 and a "right limit stop" to counter a 9° bank at 02:11:11 ?

OK465

Exactly.

The difference between "utilized" and "required".

If you let go of the SS in a 9 degree bank in ALT 2, the bank will actually tend to shallow out.

(It's not just an opinion.)

airtren

Posted on this before, and concur again.

Audio message can be continuously disruptive, visual can be too, if done in certain ways. Being disruptive once, or twice, to gain attention is perhaps enough. A certain very simple, small, visual message can be very strong without being disruptive. An A/C Error Status Panel with 3 color coded LEDs, to indicate and remind the type of Error messages that reflect the current state of the a/c.
(1) Emergency/Fatal - maybe RED , (2) Warning - maybe Yellow, (3) Informational - maybe Blue.

Lyman

Lonewolf. It is on the graph, imo. The airframe was not S/L. The g accels, the PITCH, and Hazelnuts report of a .6 degree PITCH up moment/sec. So yes, I think safe to say the airframe was active, and on its way up, coincident with PF's first and emphatice NU pull. "Additive" controls, by happenstance, then.

No, I haven't gotten to the inter-relationships of command, result, deflection, TRIM, feel, etc. for the climb. But I believe it is all in there, and hang what PF thought, or BEA, or anyone. The traces are real, they don't lie, and they don't buy/sell/fly aircraft.

A thorough dissection of airtren's graph, then? If you need work in silence, I'll read only, and nort disturb. It IS that important to me.

Gerard C. I think the "initial noseup" started well before the frame you quote and was unrelated to the 360 foot "READ"/discrep in altitude. It had begun in A/P NORMAL, and the PILOT was responsible for piggybacking on the original .6dgree/sec. -imho. The g shows a sharp reduction, as the a/c descends ('turbulence/updraft').This descent: ap commanded, no?

mm43. Howdo. "Explain the stuck AoA vane....." WRG/FAULT ??

mm43

PIO would seem to be the answer, but don't forget that post 2:10:51 the aircraft wasn't within a 'normal' flight envelope.

Worth looking at is the yaw damper, the rudder and the lateral 'g' traces. Once in the 'mush' the aircraft was 'nodding' and rolling/banking in a synchronized manner as well as 'fish tailing'.

Pitch Attitude, Rudder Position and Roll Attitude traces have been overlaid in the second trace in this graphic.

AlphaZuluRomeo

Well... perhaps it was too obvious for my deranged brain...
Good catch, thank you. So far I cannot find why it would not work (unless you freeze your 3 pitots during an actual stall... what a bad day, then)


Are you talking of the same pilot (still flying an RVSM route by the way) who (apparently) failed to react as quickly as before to a 2500-3000 ft altitude gain lasting "just" ~60 seconds with "only" a 7000 ft/minute (top) rate?
Sorry, I don't find the comparizon very conclusive...


---------

[edit/addendum] Self-quoting myself (sorry about that) but I just read (here) that the A380 apparently carry MFPs (Multi Function Probes) for Pt + TAT + AOA. Why this choice on the bigger bus? Are those MFPs something "new"?

Mad (Flt) Scientist

Yes they are indeed something new. They use differential pressure data on a network of pressure ports to infer flow direction as well as total and static pressure 9hence airspeed and altitude). The probe body is itself fixed, though, so they don't have inherent high angle capability in the way a swivelling head might appear to - though even a swivel has to have stops beyond which it can't travel.

(Edit - ignore remark about static. Other manufacturers apparently integrate that ...)