Machinbird

Ballistic agglomeration: Descent in the Sea. I have never argued that there was an inflight breakup. Loss of the odd bit? PERHAPS

bearfoil, quote:
"If the decision had been made to turn around, why not a maximum rate turn to port off the airway at some point past LKP ??"

We hope soon to have an indication from the CVR of the point at which this flight ceased to be "ops-normal" from the pilots' viewpoint. Not sure what you mean by a "maximum-rate turn", but I cannot think of any circumstance where a crew already experiencing severe turbulence in a Cb or line squall would attempt to turn back. In heavy precipitation, the radar tends to be fairly useless until you come through it. The time to do a big turn is well before entry.

Once you are in it, you are going to avoid anything but the most gentle of turns. Prior to that, you can try to plan a zig-zag course between cells, staying clear of the worst stuff, but there's always the chance of a weaker/smaller cell eclipsing the big, nasty bugger beyond.

Doubt you would find any pilot who has actually decided to abandon a passage across the ITF (sorry, ITCZ) for lack of alleyways. In any case, the analysis suggests that this crew would have been anticipating a routine one. And evidence suggests that they had not deemed it necessary to secure the cabin in advance of this event.

gums and Machinbird,
As you both tend towards the deep-stall scenario, maybe resulting from a sudden pitch-up, wouldn't it be interesting to see how far the high-altitude, low-speed part of the envelope was explored during certification?

Hi Gums,
Right, we'll certainly know everything if the recorders are fully readable.

We already tried to put some numbers together in the previous thread around this stage: http://www.pprune.org/tech-log/39510...ml#post6356297

My point was that it looks not likely that an early loss of control (0210-0210:20) happened to AF447 based on those published data I'll repost below.

From nine cases studied, we'll find an altitude loss in the range between 12,000 and 32,000 feet per minute, and a ground covered distance range between 3 to 17 Nautical Miles, while no event lasted more than 210 seconds overall after the LOC event:
__________________
9 unrecovered upsets from cruise level (data Metron Inc., BEA study):
Date.........ft/mn..Level..Dur....Dist..Type....-> Cause
07.12.1995..32,000..FL310...57s...8 NM..TU-154B -> Spiral / roll upset
19.12.1997..29,000..FL350...75s...5 NM..B-737.. -> Unknown
21.12.2002..27,000..FL180...40s...2 NM..ATR 72. -> Stall / icing
19.11.2001..26,000..FL260...59s...4 NM..IL-18V. -> Spiral / diving
01.01.2007..20,000..FL350..105s...9 NM..B-737.. -> Roll upset
15.07.2009..16,000..FL240...90s...5 NM..TU-154M -> Spiral / roll upset
22.08.2006..14,000..FL390..166s...3 NM..TU-154M -> Stall-spin
23.03.1994..12,000..FL310..156s...3 NM..A-310.. -> Spiral-spin / roll upset
16.08.2005..12,000..FL310..210s..17 NM..MD-82.. -> Stall
__________________
AF447 - timed cases for one upset from cruise level:
01.06.2009..35,000..FL350...60s..~5 NM..A-330 (1 mn)
01.06.2009..23,000..FL350...90s..~5 NM..A-330 (1.5 mn)
01.06.2009..17,500..FL350..120s..~5 NM..A-330 (2 mn)
01.06.2009..12,000..FL350..180s..~5 NM..A-330 (3 mn)
01.06.2009...9,000..FL350..240s..~5 NM..A-330 (4 mn)
01.06.2009...7,000..FL350..300s..~5 NM..A-330 (5 mn)
01.06.2009...6,000..FL350..360s..~5 NM..A-330 (6 mn)
__________________
1 Recovered upset: B747 China Airlines (17.02.1985)
1. 1014:50 -> 40,442 ft -> roll upset (4.8 G pull up)
2. 1015:23 -> 30,132 ft -> unreliable attitude data (5.1 G pull up)
3. 1017:15 -> 9,577 ft -> recovery & climb

Total upset duration: 145 sec.
1..10,310 ft lost...33 sec..18,700 ft/mn
2..21,155 ft lost..112 sec..11,300 ft/mn
3..31,465 ft lost..145 sec..13,000 ft/mn (average)

__________________
"Upset" definition (from “2008 - Airplane Upset Recovery Training Aid Revision 2”):
An airplane upset is defined as an airplane in flight unintentionally exceeding the parameters normally experienced in line operations or training. In other words, the airplane is not doing what it was commanded to do and is approaching unsafe parameters.

While specific values may vary among airplane models, the following unintentional conditions generally describe an airplane upset:
• Pitch attitude greater than 25 deg, nose up.
• Pitch attitude greater than 10 deg, nose down.
• Bank angle greater than 45 deg.
• Within the above parameters, but flying at airspeeds inappropriate for the conditions.

Turbine D, thanks. Read the report. Suppose it makes sense that the set of pitots, if exposed to the same conditions, would behave the same way. I wonder if modern systems would spot the short period of disagreement, as the icing was taking place before the data became consistently erroneous. Alternatively, disagreement would occur when the condition was recovering after the period when the system was listening to lies.

takata

hello.

"Upset" definition (from “2008 - Airplane Upset Recovery Training Aid Revision 2”):
An airplane upset is defined as an airplane in flight unintentionally exceeding the parameters normally experienced in line operations or training. In other words, the airplane is not doing what it was commanded to do and is approaching unsafe parameters.


"While specific values may vary among airplane models, the following unintentional conditions generally describe an airplane upset:
• Pitch attitude greater than 25 deg, nose up.
• Pitch attitude greater than 10 deg, nose down.
• Bank angle greater than 45 deg.
• Within the above parameters, but flying at airspeeds inappropriate for the conditions. "....

These are the limits of the autopilot supplied on the A330, as I understand. So the a/c will fly on auto pilot within this range, and beyond it (upset) the Pilots are to assume control?? The a/c is already 'upset'........ So this would qualify as a challenge, one would think.

woodvale,

You may not post often, but good post when you do !

My personal suspicion is that this will all pan out to have been a cockpit light on /Capt snoozing/read the papers/wx radar brightness dim/ don't understand tilt or gain so try to fly over it / CB encounter , followed by loss of control, either partially or not at all excacerbated by frozen pitots & Airbus gubbins.

But, the problem in any case,I fear, originated with US ,gentlemen, the pilots.

Hope not, but I fear it will be so.

A test pilot’s point of view
 

Concerning definition of upsets, stalls and their recovery some words from year 2000 from airbus chief testpilot captain william wainwright.

Download PDF Airplane upset recovery

Hi Bearfoil,

Come on Bear!
Those figures are certainly not A330 system related. They are taken from the whole Industry standards and issued in the courses of general training as a specific "upset" definition.

Each word/expression should have a clear definition (here boundaries to qualify as "an upset"). It's like your "en ligne de vol" stuff... or you will only be making some noise around the fuzz.

I'm with Chris Scott on this one.
I do not see any possibility to arrive at that point after 4mins without some kind of detour.
I had calculated before as did Chris Scott that this would require ~60kts ground speed and ~60kts vertical speed.
Given the Lift/drag dimension of an airliner and the available thrust I cannot see any way to achieve this (Btw. none of the planes in the russian study took longer than 2,5 minutes to arrive at the scene). This would be helicopter (/ light GA aircraft) territory !
Moreover you first would have to decellerate from ~470kts at 2:10:00 to less than 60kts (final speed would have to be even slower due to the higher speed at the beginning) in 4 mins without traveling further than 4nm.

My theory was a massive wing drop as initiating event for the turn.
Could have been intentional as well, however. We'll see.
hopefully.

RetiredF4;

Thanks for that "Airplane Upset Recovery" link.

Wainwright dealt in general terms with the issues, but seems to have ignored some of the FBW limitations that are inserted into the equation when not in Normal Law - see below.

Machinbird;
The boot wouldn't have helped that much! Low IAS and only 7.9° of rudder allowed by the RTLU!

Parameters and data to FDR
 

What kind of info from WX radar (if any) goes to the data recorder?

Question:

The only information on the crew "decision making" to go through WX is the one that (may) is recorded in CVR or QAR?

Hi gums,

were it for the horizontal distance alone I would agree.
The problem is that the combination of horizontal speed and vertical speed does not fit at all for an A330 @210t coming down from 35kft.

The average RoD of 6000 ft/min is the typical spin RoD of a GA light aircraft. As I calculated earlier for an A330 @210 t I would expect RoD of ~20.000 ft /min at altitude slowing down to ~14.000 ft/min at S/L as terminal velocity e.g. in a flat spin.

That is the problem I have with the 4nm and 4,5 mins. I simply cannot see it descending with 60kts at a rate of 6.000 ft/min. At 60kts (~30kts IAS at altitude) the required lift for that RoD is not there.

Forgive me if this is an ignorant question (I'm not a pilot), but it seems to be a strong possibility that the pitots iced rapidly due to supercooled water. If that is the case, could ice have also formed elsewhere on the aircraft in a similar manner and could that have potentially played a role in making the situation worse?

takata posted the following summary of an A310 LOC.
Date.............ft/mn....Level....Dur....Dist..Type...-> Cause
23.03.1994..12,000..FL310..156s...3 NM..A310...-> Spiral-spin / roll upset

From what I understand, the aircraft departed FL310 cruise altitude into a gentle right-hand turn which then became a high bank roll and dive with the aircraft turning through 180 degrees before making a partial recovery. Control was finally lost at around FL130 and the aircraft did a combo spiral dive and finished up spinning in.

The point I am making is that any number of manoeuvres, controlled or otherwise, could conspire to place the aircraft close to the point that LOC occurred.

Henra, did you forget the vertical component of velocity? They add vectorially. And as the velocity vector is largely pointed down, drag (which is substantial) helps support much of the weight of the aircraft.

Two fragments taken from BEA Report #1; Appendix 1; Meteorological study undertaken by Météo France: Meteorological situation in the tropical atlantic on 1st june 2009 - :

To this can be added that transport category airplanes are designed and demonstrated to be able to cope with large concentrations of supercooled water as can be found at lower altitudes.

Right.
But, more vertical component means also much less duration (a faster rate of descent). You can't have both. Now, we are facing the opposite: a very long duration (the longuest ever recorded if LOC occured so early) but a very short distance from departure point. This will translate into a lot of horizontal component but in a very tiny airspace. Those figures should be somewhat in the range of what is already known or something very new happened... or is just wrong with the basic assumption.

henra theoretical: if the bird is without any control but starts with an extrem climb mayby 60 deg, this (~balistic) path will reach a maximal high of ~12700m (full change the kinetic horizontal energie into potencial energie), while reducing the speed to very low untill it will deap stall and if it then fall with a stable AOA of 50-60 deg. back to 10500m and afterwards down to zero.....

this path could need ~100 sec longer than yours and can end within a range of 8000m

Good Point Olivier
The reason for a very long LOC event is fairly obvious, the Airbus FBW system was still in control. But how was the airspeed lost so quickly? Logically the aircraft departed controlled flight with a strong upward vector.

Now if someone can match up these endpoints, we will likely come very close to the cause of the loss of AF447 without even using the recorder data.

Looks like great minds think alike.:} Grity is right on target.

No, I did not, at least not intentionally (may be I was a bit unclear?!)
I said horizontal component was 60kts. Vertical was ~60kts as well.
But even if you combine both you end up with ~85 kts along the trajectory.
And these 60kts and 85kts is even TAS not IAS, so when entering it into a L/D diagram, you would have to use IAS which is roughly TAS divided by 1,8 at altitude, i.e. ~35kts horizontally or ~50 along the trajectory, increasing to 60/85kts at S/L.
That's not really much.
You would have to enter the L/D diagram of the A330 at an Alpha of ~45° with 85 kts and see if the combined Lift and drag higher than the 2100 kN (+ thrust of the engines).
That would mean ~1500kN along the trajectory, if no thrust of the engines available (idle, stalled) and ~ 1800kN at Cruise Thrust along the 45° path.
The projected surface at 45° being roughly 520 sqm, at 44m/s you get a resisting force of 620 kN at Cd = 1.
This means the Fuselage + wing would have to have a drag coefficient of ~ 2,5 - 3.
That's definitely too much.
I would expect something around 1.

At Cd = 1 it would mean ~80m/s => ~160kts along the trajectory. =>
~110kts horizontally + 110 kts vertically

On top of all that you would additionally have to shed the initial kinetic energy of it flying @470kts.

There is not so much left of the FBW to control in ALT2 Law. Wasn't it one of the first events to be triggered at 0210?
Or you'll have to start your sequence even earlier, which is even worst considering this timming issue at stake.

I'm believing that those end points can only be matched if every starting assumptions made is basically correct. If one discard something, like ALT2 law, one will certainly end with two fairly unmatchable end points.

When this "zoom & boom scenario" is considered, it will seem obvious that those fully monitored engines (from Brazil take off to 0214:26) will trigger at least "ENG STALL" ECAM warnings and all related ACARS.

Remember that F-GZCP engines were working "fine" during all the sequence covered - dixit the BEA as well.

But all very dependent on power setting. If my aircraft had just lofted me to the edge of space, I would put the throttles wherever the engines seemed happiest just to keep them running and the lights and hydraulics available.

Do you have specific data on engine operating envelope-Olivier? I wouldn't be surprised if the engine authorized envelope would be exceeded, but the stall margin envelope is typically very dependent on power setting.

And without Direct Law, I don't think a deeply stalled Airbus has a hope of recovery. It is a THS thing.

SaturnV, as a woman with pierced ears I didn't think of the screw on type that would, indeed, probably come off.

Mr. Optimistic, "Any clues as to what the northerly outlier is ?"

Not that I know of. But it if is the bare APU it might give a solid hint of what direction the plane is going.

Post-Stall and Spin Dynamics of Large Transport Aircraft
 

@mm43 et el

Recent NASA papers:

https://engineering.purdue.edu/~andr...2007005369.pdf

http://ntrs.nasa.gov/archive/nasa/ca...2008034489.pdf

http://ntrs.nasa.gov/archive/nasa/ca...2005208658.pdf

http://ntrs.nasa.gov/archive/nasa/ca...2008033763.pdf

Thanks. If they are going to put up videos wish they would make them legible ! Probably says 'a rock'.

Engines during upset
 

Takata,

Remember:

2:13:16 ~ 2:13:41 Possible "Loss of Signal" with satellite

bearfoil, most of the debris is pretty random. There is a good chance that the heavy stuff (engines and maybe APU) dissipated its horizontal velocity component fairly quickly in water. So they will fall fairly straight down with little dispersion from current or anything else.

If that is true, the Northerly outlier is the APU, AND the plane hit tail first then left wing, then we MIGHT be able to say the plane was heading more or less SSE to SE. That's a lot of ifs. So as noted I'm sticking to relatively random debris field modified by underseas currents as most likely. But I am putting in my one cent's worth that AF 447 may have been heading generally SE-ish.

Also, it's worth remembering that sections that sank relatively intact at the surface will have imploded on the way down (or possibly even exploded if they sank with trapped air inside), further complicating the picture.

woodvale, a few decades ago Amateur radio equipment was infected by this make it stupid simple philosophy. It (cough) missed a great deal in the assumptions it made. With older equipment and experience found I could dig out signals others didn't dream were there. (One fellow had to redesign a piece of a black world project because I could drag out and recite to him the unencrypted speech being sent. That was a MAJOR oops.)

The Amateur radio world got over this affliction. Military radio as a general rule has not. Aircraft radios are also minimal user interaction and control as well. This applies to the flight controls as hope that the airliner manufacturers may get over the make it stupid simple philosophy, too. If you are going to design an aircraft that gives up control when the automation gets confused but not before then you need the pilots to have the most wide open control possible. Maybe that will be a takeaway from AF-447.

Meantime, lobby for a switch on the radar front panel that may involve an "ARE YOU REALLY SURE?" interaction with the pilot that enables full control of the display including any raw mode that could exist. Maybe that MISS will be fixed. The pilots are being offloaded for other flight controls. So they have time to operate the radar in detail modes; and, they have time to learn how.

(And off hand I find it a little unsettling that airliners are working to become so simple to fly even al Qaeda militants can do it with minimal training.)

{^_^}

post #925, captplaystation, excuse the presumption, but perhaps this will be the one that redefined the whole approach to automation.

The system doesn't eat ACARS. This quote is to explain a possible delay in the time stampings, but not a message disapearance.

If the statellite link is lost, ACARS in the pipe are kept until the link is restablished. The protocole takes 6s with full aknowledgments at both end.
As the satellite was obviously working until 02:14:26 with no ACARS processing interrupted up to this point (it is known from the Sat operator), only very late ACARS in the maintenance system (not already processed) was kept by it until the end of transmissions.

henra - stopping from 470nm/h in 4 nm is a fairly modest (by automotive standards) 1/3g. Now, for a plane that wants to fly that's an absurd horizontal deceleration. For a plane in an upset condition that may be entirely within what could be expected. A very sharp pullup might do it, at least to my untutored visualization.

This does hint strongly that the upset happened a very short time before the LKP report. A very sharp pull-up followed by a fast drop may explain nobody managed to get to their life vests and people were still out of their seats perhaps queued at the potties. Once the upset started they'd be tossed around the cabin and not able to get back to their seats.

This does favor delta92's scenario, I suspect.

Um, the MISS, Make It Stupid Simple, philosophy may have struck again.

takata, think the point being made was that there was a loss of link integrity for a short while as per Svarin which may correlate to a manoeuvre

Interesting ?

Now more : both PRIM1 and SEC1 messages appear right after the "possible loss of signal" window.

Interesting ?

On another point, if the last position report was at 2:10.x when the a/c was already 3nm west of track then since there is at least another 4 minutes of flight based on acars, and at cruise speed this translates to a lot of ground, the a/c must either already be decelerating or the timeline for height and speed loss gets tighter AND the ground covered has to be negated in a subsequent manoeuvre.

I understand that this is for the zoom part of your scenario.
But what about the boom part of it?
Would you keep full throttles, once having already a very large AOA exceedance , and all your way down during about four minutes?
from, say 40,000+ feet down to sea level?

Not at hand.
But say 50-60 degrees AOA without much forward speed remaining at all (in order to stay into your 8000 m zone) will make such an angle for the airfoil to bypass in order to reach the compressors that I really doubt of the tolerance. I'm not even talking about all the tropical storm ice/water you will ingest at lower than cruise levels in the process.

What make you think Direct Law may be lost?
The rudder would be still limited (could be an issue at low speed for stall recovery) but other surface control would still move freely.

That is what troubles me with the simple scenario of pitch to a fatal point and actually slide down backwards at least part of the way. Wouldn't that put the candles out? And isn't a tail down stall recoverable using the elevators? Maybe they were in the middle of recovery when they hit?

{^_^}

Old article here giving some general details about GE engine monitoring

MRO USA: Engine diagnostics: GE opens the envelope

It is highly likely that everything was OK when the engines last 'phoned home'. Interesting to note that the data packet sent to GE must include basic flight information

I think you missed a bit of the intro:Now, the document is 10+ yrs old, the work that went into it, 15yrs or more, since then we've had crashes like Perpignan and we know that it's perfectly possible to stall these a/c when things aren't working right, so maybe it's time to disagree with "There is no need...".

Crash QTR
 

Before boxes info, one may ask:

Among the recovered bodies may be BEA or else one has the crash time?

An old mechanical clock would be nice;

Would simplify our current discussion :)