Back to the Facts
 

Hi,

Browsing again the old BEA reports, I found this quote which should have been taken into consideration much earlier by most of us as for the initial state of the recovered floating bodies, next, enlightnig someway the nuisance of the various rumors circulating from "tabloids" reporting sources:
This would contradict most of the press releases (still today summaries):
1. Many bodies were recovered by the French Navy (about 3/5th by Ventôse);
2. They were clothed and well preserved;

S~
Olivier

News from BEA
 

Information, 9 May 2011

They have recovered one engine and the avionics bay.

Some BEA photo isolations from today's briefing:


http://i337.photobucket.com/albums/n...7-EngLift1.jpg


An A330 avionics rack (hat tip to PJ2):

http://i337.photobucket.com/albums/n...0-AVrack04.jpg


AF447's rack being hoisted aboard:

http://i337.photobucket.com/albums/n...AvBayLift1.jpg


Recovered rack components:

http://i337.photobucket.com/albums/n...47-AvComp1.jpg

Machaca;

Many thanks once again.

With the Avionics Bay comes the memory boards/chips of the FCPCs, FCSCs and possibly the QAR. If a card, (and I now suspect it is), it may be readable if not shattered. The card is placed in a slot so is somewhat "protected" unless the unit itself was crushed. What we see of condition of the boxes is encouraging. Let us hope for the best here as well.

I see from the photograph of the engine that half the stator-blade section is missing, confirming what was seen in the original photographs.

auv-ee:
Thank you for pointing me to your previous very interesting post on the subject.

Can we assume that a 20db signal to noise ratio is about the detection capability of TPLs ?

Since we are in the ITCZ intermittent heavy showers have a significant probability of occurring at any time. I read in a US Navy paper that rain drops impacting sea surface and implosion of air bubbles caused by rain produce noise in the 1-100 kHz band with max SL @ 20 kHz and SL can be up to 30 dB above sea surface noise though I think the intensity is lower at great depths. It seems that if the TPLs were flown at, say 1700 m from the nearest pinger a simple shower occurring at the same time would have prevented detection. As I understand phase I, any area was swept only once so perhaps the failure was only bad luck.

OTOH, it is very possible that both ULB failed in the crash.

If they were using a traditional detector, which consists of a narrow-band filter, a peak detector and a threshold comparison, then 20db would be about right. If they are doing more modern signal processing, then it might be considerably better (another 10-15db, perhaps).

Those numbers sound about right. I expect that the 30db increase is compared with sea state 0; the increase will be comparatively less if the surface already has breaking waves. Also, these high frequencies are attenuated with distance, as I noted in the earlier post; so, yes, the effect will be less at depth. Also, at 4000m depth, the rain noise is being averaged from a similar radius area of the sea surface; thus a rain shower that does not cover that whole area of the sea surface, will have a lower total effect.

Weak signals detection
 

If the half buried SSCVR ULB was transmitting as specified the chances to be detected with available techniques and adequate methodology were very high.

I am very familiar with "weak signals detection" using DSP techniques, intrinsically quite insensitive to noise.

In the picture below:

http://www.sdrham.com/images/zl6qh2001-09-22-1856.jpg

You have a signal that could be IMPOSSIBLE to be heard and with proper techniques can be CLEARLY detectable in the computer screen (High Signal to Noise ratio) using widely available FREE software PC (Win) compatible.

For example:

With this setup you could tune in a distant NDB (impossible to be heard, even it´s carrier) and it will come up "loud" and CLEAR in your screen

Another example:

You could have a CW 10W HF carrier operating in the 5 MHz* aeronautical band (dark hours) in the a/c, being monitored in DKR or REC with near 100% chances to tell you when disappear triggering an alert.

(*) 8 MHz in daylight.

Even with the current HF antennas (a/c and ground stations) very poor compared to the ones used by dedicated amateur radio in their stations.

PS

I will investigate what kind of Receiver they used in order to understand on their capability to succeed in cases like AF447.

And if the wreckage did fall in a hilly region it would be more difficult.

I recall from the NASA Columbia investigation that bodies recovered from in-flight breakups, or from the failure of a pressure vessel at a significant height, the bodies in all previous recoveries were denuded; this was the case in Air India 182 and Pan Am 103.

The fact that bodies from the AF447 are fully clothed, I assume, rules out any in-flight breakup and explosive decompression?

From the PDF
One hopes that this apparent hubris has been mitigated in the interim. (Report is 15 years old, yes?) I say this fully acknowledging that the author is a better pilot than I.
The difference between education and training. ;) I applaud the approach advocated.
A boilerplate "unusual attitudes" recovery technique.
As discussed previously in various AF 447 posts ... and as seen in stall training in small single engine trainers.
JD-EE
If your suggestion is to be considered, there is likely a cultural (industry and/or corporate) issue that may be non-trivial to address .
I think you mean "for other flight duties" but I see what you are headed towards.

I learned radar work on a scope with more or less raw video (ship board radar) and was disappointed in later graduating to the APS-124, which only providing processed video to my cockpit display in the (then new) SH-60B. One factor that drove this choice was that the data from the radar had to be in a form that would easily go down a directional data link to the ship's Combat Information Center. That tech and design decision allowed the radar to share space in the cockpit with our other tactical displays, via a selection toggle, but it constrained how much one could get out of the radar itself. It also allowed either the pilot, or the crewman, to work the radar.

While I share your position on giving the pilot the choices to tweak the display and input, there is the design problem of competing with other capability for real estate in the finite amount of space provided for pilot attention on any flight deck.

Ergonomics at its most interesting.

Some of the folks who operate that radar have, in various posts over the past two years, pointed out that there is some room to "tweak" the display during a given trip. It may be "good enough" with enough operator training, habit, and familiarity. This points again to a cultural issue, possibly moreso than a design issue. :cool:

QAR media
 

(#1003)

Surprisingly in better condition than i first thought. Considering the bay location in the a/c.

zoom-boom scenario
 

I´m somewhat astonished concerning this scenario.
Flying at FL 350 with a still heavy aircraft, how much climb could be produced even with full power? Is there enough elevator authority available to get the climb rate going to reach such an AOA and how fast would airspeed degrade? Gums with his F16 could have done it, my F4 would just have produced a lot of drag on such an elevator input and would have picked up a descent rate instead of a climb rate.

Next question is, at what point of the events and with what kind of airspeed could this pitchup have started? Icing of pitots, therefore unreliable airspeed (not known to the crew jet), turbulence and autothrust of, stall onset at the low speed region (not clearly identified by the crew yet, might be mixed up with turbulence issues). Crew-action increase thrust leading to pitchup (underslung engines, unfavorable CG), with slowing down descent rate, but not curing the pitch problem (correct speed still not known to the crew, work overload, turbulence.....). Nose stays up crew reduces thrust again..........

At what forward speed would we define the upset (if it happened) from FL 350 to FL100 and then on down to the water? If the normal flying speed is 470 KTAS in FL 350, is the thing still flying with 300 KTAS? Per my definition the aircraft would be out of flying speed way before the previous mentioned 60 KIAS from other posters.

The going down in a upset condition would probably never be in a straight line, struggling with a falling aircraft in free airspace directional control would be my last worries. It might have done one or more turning/ yawing motions around the vertical axis, it is at least more probable in an upset than a straight line. That sums up for the missing time. And it might not have been a continuous upset, secondary stalls are quite a common encounter in well controlled training environment (planned, briefed, prepared and good weather).

I´m not saying, that´s what it was.

franzl

But if you keep an open mind, there is a small possibility of such - similar to the small possibility of both pingers failing (maybe about 1%). Looking at the ACARS, I too would have assumed no glide, but I would also have assumed at least one working pinger based on track record.

Really ? They failed to completely search it all (hence the phase two) within the pinger specification time despite using two TPLs, a nuclear sub, several surface ships and another sub + a ROV. What would you consider a "large" search area, and how much resource would you need to deploy to search it within the 30days ?

No, they did not "assume", given the lifetime (30days minimum, 40 expected) the pingers are known to be no longer working by this stage.

First search - pingers over 90% probability of working, larger area can be covered faster by looking for signals. Later searches - pingers <1% probability of still being working, so use other slower search methods over smaller area. Eventually got lucky.

It's easy to use hindsight to say they should have bet on the 100:1 outsider after you know the result.

From Tim Vasquez's excellent meteorology of nearly two years ago.

http://www.weathergraphics.com/tim/a...47-profile.jpg

http://www.weathergraphics.com/tim/a...7-cloudsat.jpg

IMO, the possibility of a downdraft beneath the anvil contributing to the rate of descent ought to be considered.

Vasquez's site for his AF447 work.
Air France 447 - AFR447 - A detailed meteorological analysis - Satellite and weather data

Watch out Tim's work was produced very early on when it was assumed the Brazilian navy had a location for the final ACARS message - which later turned out to only be extrapolated, and now some 30NM off. I think this moves the final transmission right into the heart of that Cb.

That's a terrific picture of the engine! You can see the 12 o'clock position facing the camera --their sling is hanging it from it's 9:00 -- and the fwd eng mount quite well... [*edit*Clearly*] it appears** the bottom forward half of the engine took the brunt, pulling the forward mount down, shearing those mount bolts and pushing the aft upper portion of the engine upward... Can't see what failed there, because the aft mount is gone.

RR_NDB
I believe the technique you are referring to is a phase lock loop system.
Phase-locked loop - Wikipedia, the free encyclopedia
Real electronic black magic. No doubt the EE folk will have more to say.

or in my day lock-in amplifier

Hi,

Sorry for the big size (so I post only the URL) .. but this give a better view (engine) of what explained above ... (magnified 150 % of the original BEA pic)

http://i.imgur.com/F0cQX.jpg

Digital Signal Processing
 

Machinbird,

PLL is just a "building block" that can be "hardware based". And are powerful.

DSP systems are IMPRESSIVE. You use the "computing power" to "remove" the noise (that is ramdom) to detect the signal (that is coherent) using Redundancy.

The results i observed are simply unbelievable. So i designed a "Propagation Tester" using it that can be understood in my PY2CW mobile Ham Station website.

Using the synergy between HF and Internet in "near real time".

I am curious to known what kind of Receiving Technique they used in the failed ULB location.

It's all very well for you experts to pontificate about the several sciences involved but I am simply amazed that they have got an engine up from that depth.

I know how big that engine is and I applaud their efforts.

Salute!

First the beacons:

- if they are using pulse code modulation or a variation of the spread spectrum technology, it's very easy to find the thing even with s/n approaching 1:1 Witness GPS with all the stuff basically on a single frequency. Same with your cellphones and such. You just have to know the "chip" sequence and rate.

Some of the RF gurus here can explain it better than I.

Second, about the pitch excursion theories:

- Doesn't take much to lose half your groundspeed in ten seconds if you are climbing at 10 or 15 degrees and not having motors going full grunt. Just a few degrees of flight path change at a reduced power setting will do the trick - think higher AoA and resulting increased drag.

If the air data went FUBAR and the jet thot it was going faster than it was actually going, I can see the thing climbing until reaching the AoA "protection". Unless the power was increased (unlikely}, a very rapid speed bleedoff would occur. The folks in the other incident being mentioned either took immediate action or the body rates were low enuff for the flight control system to counter the excursion (gotta find out more about that incident, but it sounds hairy). Not the "test flight" crash, but the two planes flying close together when one went yahoo.

- If you look at the jet I flew, we could reach 40 or 50 degrees AoA before the stability issue came to play. From then on we had little or no pitch authority commanding a nose down attitude or gee. Remember we were flying at a very rear C.G. So if we got real slow, real fast at a steep climb, then we didn't have enuff control authority to get the nose down. We "parked" and then started down at a great rate. Wasn't some violent "snap roll" or "nose slice" or "pitch up" ( also flew the VooDoo and it had the violent pitch up LOC). We couldn't enter the deep stall on purpose at pitch attitudes of 5 or 10 or 20 degrees without being real slow and with some harsh rolling or skidding commanded by the pilot.

So I can see entry to a stall or even a stall/spin if the jet started an extreme climb.
++++++++++++++++++++++
I feel most of us wish to discover the root cause of this accident, then come up with some procedures by the crew to mitigate the situation. If we have to change the flight control "laws" while we're at it, then so be it.

Contrary to popular belief, the Viper FBW system had several modes that changed the priority of the air data, body rate, AoA, and gee inputs. Nevertheless, the system was way less complicated and had zero inputs from the autopilot to bias the basic control laws - it acted as a very limited "pilot" command when seen by the computers. More on the autopilot later upon request, but it wasn't quad redundant and was basically a device to reduce workload while pulling out an approach chart or cruising along to and from the target.

later,

Gums sends...

-

Real EE black magic = kalman filters (especially when extended). PLL is the simple case.

Once I thought I got all the math and completely understood how they worked, whilst the elders advised just treating it as black magic... now, I am older and wiser and know how little I knew :)

Retired F4 - The 330 has a big wing and two powerful engines. Cruise power settings are modest in comparison to other types such as the 320
Imagine the folowing scenario- pitots ice up generating an overspeed. The aircraft pitches up and the athr retards the power to idle,add in a powerful upcurrent as a cb is penetrated adding to the pitch up. The aircraft has a lot of energy and due to the big wing a lot of manouvrability at altitude Whilst the crew is reacting to the autopilot dropping out and the aircraft fbw law degrading the aircraft stalls. If the aircraft has been cruising at 15-20kt above VLS and assuming a rapid decelleration the crew may have had only ten seconds or less to counter the stall but having penetrated a cb they are still in an updraft. The aircraft continues to climb. The crew pitch down to counteract the stall but the speedtape is still showing an overspeed. VLS is now overlaying the Barbers pole and the crew struggle to interpret what what has happened.Their computerised FBW protected aircraft had turned against them.
The cockpit is full of warning chimes and the ecam is full of failure messages due to the icing up of the ADR. The aircraft exits the CB core and is still in a
stalled condition but now is in the downdraft of the CB and it rapidly descends towards the sea and to their demise

Speed/height trade
 

The mass doesn't enter into the equation. I see gums answered most of the questions you raised. Here is just a graph illustrating speed traded for height at constant total energy: Zoom Climb

Signals 20 dB or more below the noise floor can detected using this method!
 

Gums,

Some tutorial info:

DK8KW Longwave Information (Slow-CW)

Do i understand the graph correct?

Flying S&L at FL 350 with 470 TKAS, i can zoom the A330 to FL430 by trading 250 Kts of my TAS, provided thrust equals drag? To initiate this zoom, how much additional drag would be produced and how would that effect the result, if thrust is not increased?

What would be the rate of climb and AOA? We discussed the possibility of a sudden pitchup-scenario, not an easy and gentle climbing maneuver, which the graph probably represents, didn´t we?

Why does the mass not enter into a "useful" equation? I´m aware of the title "Constant TE Thrust=drag", bat higher mass means higher drag, therefore higher thrust to maintain TE (drag=thrust) is necessary, isn´t it?

At what altitude would the speed be below stall-speed and what would the stall speed be?

Trade kinetic energy 1/2 m v^2 for gravitational potential energy mgh - the m's cancel.

Noise
 

I think you are confusing wide-band vs. in-band noise. Note that I said in answer to DJ77 in post 1005, that a traditional detection starts with a narrow band filter. This accepts and compares only the signal and noise that is within 100-200Hz of the target signal; the bandwidth of a 10ms pulse being about 100Hz .

The pictures you have shown appear to be of signals that are at least 10 and maybe 20db above the in-band noise (I don't know what the color scale is for those images). Your sonograms are performing the narrow-band filter operation, by using an FFT to break the spectrum into narrow slices, then separately displaying each slice.

gums points out correctly that GPS works at very low SNR; but GPS uses a long pseudo-random code that an auto-correlation algorithm can detect. Unfortunately, these old pingers just send a 10msec tone burst. One can certainly do better using techniques possible with a DSP than with the traditional detector, but I'm certain that the performance of a system like GPS, which uses engineered signals, cannot be exceeded when the signal is a tone burst filling only 1% of time (10msec/1sec).

SaturnV,

Tim Vasquez's paper is indeed an impressive document. As I don't have to remind you, he did not entirely rule out the possibility of what he calls a "Warm sink" by the northern face of that Cb, which I guess marks the "front" of what the French still call le "Front Inter-Tropique". I suppose that this warm sink would amount to a downward kink in the tropopause. And we all know what flying in and out of the "trop" can be like.

The increase in SAT would lower the Mach for a given TAS. The wind would change, and the BEA commented that there was a weak jetstream of 280/85 at latitude 10N to the west of track. This MIGHT indicate a switch from very light headwind to a tailwind on the north side of the ITCZ. The downdraft also might cause loss of performance. The engine thrust would be lower for a given N1. Turbulence would be inevitable. The combination of all these factorss might present significant handling problems, even without the loss of reliable IAS/Mach indications.

Vasquez's flight-profile suffers slightly from lack of available data: I refer particularly to the isotherms, which he had to draw perfectly horizontal. Interestingly, the altitude scale is for true altitudes, and he shows AF447 at 35000ft on the -43C isotherm. On his modified SKEW-T for the "most probable" case (Fig 11), he has substituted a true altitude scale for the scale of pressure (millibars) on the vertical axis. The diagonal isothermic lines show -43C as being at about 37000ft, which I interpret as being the true altitude at FL350 (due to the ISA deviation).

If you suddenly encounter even warmer air (albeit only affecting the higher levels – not the whole atmosphere below you), you need to climb even higher to maintain FL350...

Chris

Takata

.13 Medical and Pathological Information
Sailors from the Frigate Ventôse recovered about thirty bodies. A visual examination of the bodies showed that they were clothed and relatively well preserved.
This would contradict most of the press releases (still today summaries):
1. Many bodies were recovered by the French Navy (about 3/5th by Ventôse);
2. They were clothed and well preserved


A bit confused by this. Unable to spot the difference, or are my eyes failing me?

promani;
The facts associated with that statement are most likely (from my experience) somewhat different, i.e.

"ballonné et endommagé par la morsure de requin étrange"

auv-ee;
Not a good basis for PLL capture, bearing in mind that the ULB oscillator is most likely R/C with inherent phase shifting and possibly small frequency changes as the battery voltage drops.

mm43
"ballonné et endommagé par la morsure de requin étrange"

"bloated and harmed by the bite of strange shark", is my translation. You could be right.

Yes.
You control AoA with the elevator, that changes your load factor, which determines the rate of change of rate of climb. In other words, if n=load factor, c=rate of climb, and t=time, then dc/dt=(n-1)*g
That depends on how you use the elevator, i.o.w. how much 'g' you apply.

I assume we are in agreement. That's another way to put the point I was trying to make.

promani;
"bloated and damaged from the odd shark bite" seems slightly clearer!

auv-ee;"Singing from the same song sheet" is another way to put it.

Google Earth
 

The terrain patterns shown in the LKP maps here are different from the terrain depictions in "regular" Google Earth. Can anyone explain?
For example, the abbyssal plane described for the final debris location looks like very broken/steep terrain in Goole Earth, not a flat sandy bottom.

However, I should point out that an uncertain frequency is one problem that can be aided with improved signal processing. In the traditional detector one would make the filter bandwidth wide enough to be sure to include the maximum drift of the signal, thereby including more noise. With DSP techniques, it's easier to simultaneously track multiple narrow bands to find one that has a signal. This is likely part of RR_NDB's point.

Googleearth