AF 447 Search to resume

[steamchicken]

Open447

I'm quite dubious about crowdsourcing in the pure sense, because the problem is too much like screening bags - people aren't very good at detecting significant rare events in a large number of insignificant ones. Further, as a lot of the seabed is going to be quite different oceanographically, most of the variation in it will be rocks and the reviewers will lock on to the rocks because they're what's available (anchoring; target fixation).


But there might be some role for it as part of a semiautomatic process - for example, if some kind of statistical model has thrown up x number of probables, there could be value in having the crowd work through them, as the targets probably aren't very amenable to automatic detection. If we were looking for the fin, for example, that would have some nicely defined corners, but we're looking for "stuff that looks engineered, but has been smashed" and that's quite subjective.

There are some interesting methodologies for this sort of thing; one would be pairwise comparison (a.k.a kittenwar). Rather than "is there an aeroplane here?" or worse "here are 100 very similar sonar images, can you spot a piece of wreckage?", this presents the user with 2 randomly selected probables and asks them to pick one - eventually, you should end up with the images sorted by the totality of the users' preferences, which should be a valid Bayesian search. There's a more formal version called the Analytical Hierarchy Process, where you make the pairwise comparison on the basis of parameters you choose in advance.

As well as kittenwar and Am I Hot Or Not, this has been used for quite a few hard problems - like examining British politicians' expenses claims, and making technology decisions in big companies.

Alternatively, if you were trying to build a statistical method of identifying targets on the seabed, you might use something similar to classify the seabed into different topographies, or to confirm that a zone was clear of wreckage so that you could use it to calibrate the model.

Depending how self-similar the seabed topography is at the relevant scale, it might be possible to use a Fourier analysis to compare similar areas of seabed - the answer to my assumption looks to be "not very", but it might be possible to classify it into relatively self-similar zones.

[aguadalte]

HN39:

Is there any merit in the BBC claim to have solved (part of) the mystery of AF447, with the theory of an NOAA meteorologist that a small convective cell may have hidden the 'big' one behind it on the Wx radar screen? Wouldn't the crew have avoided the small cell, and thereby the one behind?

In fact, this is not knew to us. We all know that a smaller cell, may hide a bigger one. Sometimes, this smaller cell, may appear to the crew, as not threatening the path we are flying. However, when the radar gets a first cell, and if it appears to be "blocking" the "view", the radar will display a continuous solid band at the further end of that same display, as wider as the shadowed angle, to alert pilots for that possibility.

[JD-EE]

Originally Posted by FlexibleResponse

Actually TiiberiusKirk makes a good point. GPS knows accelerations which integrated provide speed and the again provide distance in all axis. GPS routinely provides wind and true speed and direction and distance and thereby navigation information.

GPS does not track accelerations. It is a tool for calculating positions. Differentiating position with respect to time gives velocity to relatively poor resolution. Differentiating velocity with respect to time gives acceleration "sort of". The intertial units provide the acceleration. Neither directly gives velocity. Both GPS and the intertial units are combined with other data sources in Kalman filters to give reasonably good estimates of position, velocity, and acceleration.

None of them know squat about the surrounding air mass.

{^_^}

[auv-ee]

Alternate airspeed measurment

It seems that an alternative air speed measurement is needed to augment the pitot tubes, something that would be affected differently by environmental effects and other failure modes. I'm not suggesting replacing pitot tubes, because that would just introduce a different, but still single, failure mode, rather than introducing redundancy. What comes to mind is Doppler measurement of back-scattered energy, either EM or acoustic. A quick search reveals that an optical version of this is already being developed in the UK (though they seem to want to replace pitot tubes):

Emerald: Article Request - New optical airspeed sensor poised to cut airline costs

Likely other groups are working on this also.

[BOAC]

Not a lot of use when it is IAS you need.

[auv-ee]

Alternate airspeed measurment

Originally Posted by BOAC

Not a lot of use when it is IAS you need.

If you are responding to the use of Doppler measurement for air speed, that is an excellent point. It would be necessary to measure air density to go along with the velocity measurement. I don't know how the UK team plans to do that, though it could be derived from temperature, static pressure and humidity (any other significant factors?).

[bearfoil]

Redundancy is a siren on the rocks when one is up against common fault. BA038 had redundancy, but at some stage a system must be parallel, hence the dual rollback. The closer the rubber gets to the road, a system folds into unity, up only against the odds.

Unreliable airspeed became common on this type, hence the acronym, and the training focus. Because three separate pitot static systems generate discrepant airspeeds does NOT necessarily mean they have all iced, together or in some aberrant sequence.

Sailing in the doldrums one experiences no "wind". The mass is expanding around you and rising. Slowly at first, then as this warmth gains altitude, it increases in velocity.

Had 447 entered the mother of all cells? Most likely not. An average one will do nicely relative to upset and airspeeds that rocket up and down. Was any of the Alert/warning/alarm prompts singled out and addressed/cured? Shall we use the RTLU as an impromptu altimeter, by its demonstrated limit after falling out of the sky? The fuselage of the 330 is massive, and the pitots are well away from each other. This attacks the theory of Ice/turbulence, and entertains the discrepant/local theory. 1/2, 1/3, 2/3? Look at the numbering of the Tubes and the ACARS.

Again the autopsies are mentioned. The Captain's remains do not suggest a position in the a/c. Neither do the other peoples' remains who are of necessity assumed to be unseated, mobile, and not just at impact. Specifics of trauma are limited to the determination by BEA that these were seated. Belted? An easy conclusion given the facts? Evidently not. I say again the trauma focussed on by BEA is patent in trauma exhibited by human beings in CAT or upset historically; in this case the posture of each victim is concluded to be known. Mistake?

[SaturnV]

HN39,
Perhaps more a function on how the radar was being used and less on whether cells were being hidden. From the first BEA interim report,

Flight AF459 (Airbus A330-203) passed at the level of the ORARO waypoint approximately 37 minutes after l’AF447. The sky was clear but the half-moon, visible to the aft left of the aircraft, did not make it possible to see the contour of the cloud mass distinctly. After flying through a turbulent zone in the head of a cumulus congestus formation at the level of NATAL, without having detected this zone on the radar, he selected gain in MAX mode. At about 2 h 00, he observed a first echo that differed significantly depending on whether the radar’s gain was in CAL or MAX mode. The TILT was set between -1° and 1.5°. He decided to take evasive action to the west, which resulted in a deviation of 20 NM to the left of the route. During this evasive action, a vast squall line with an estimated length of 150 NM appeared on the screen, which was set to a scale of 160 NM. The echoes were yellow and red when the radar was set with gain on the MAX position and green and yellow when the gain was on the CAL position. No lightning was observed.

ATLANTICO control, informed by the crew of their decision to avoid this squall line by taking evasive action to the east, asked them to return to the airway as soon as they could. This evasive action meant the aircraft flew between 70 and 80 NM to the right of the planned route. In addition, the crew was authorised to climb from FL350 to FL370.

Similar narrative detail is not provided for the actions of the pilots on the Iberia and Lufthansa flights, who also deviated.

bearfoil, there were 50 bodies identified through autopsy. The second interim BEA report says this:

Forty-three of the victims had fractures of the spinal column, the thorax and the pelvis. The fractures described were located mainly at the level of the transition vertebrae. The compression fractures of the spinal column associated with the fractures of the pelvis(2), observed on passengers seated throughout the cabin, are compatible with the effect, on a seated person, of high acceleration whose component in the axis of the spinal column is oriented upwards through the pelvis.
(2) Fractures of the pelvis can also be associated with the wearing of a seat belt.

Whether from deliberate or imprecise phrasing, the above language read literally suggests the 43 victims so described were passengers, with the results of the five crew autopsies omitted.

[syseng68k]

JD-EE,

That might need a little clarification.

While gps is primarily a system to calculate position, it can also be used to provide fairly accurate heading and ground speed data, in conjunction with appropriate software.

For example, if gps position is sampled at a constant, known rate, it only needs a bit of trig to calculate heading and for ground speed, a bit of arithmetic using sample rate (time) and change in position + scale accordingly. Not exactly rocket science.

Of course, ground speed != air speed, but in a typical flight, with small changes in wind force and direction, there would be a measurable relationship between the two, which could be used as an added data input to the various system filters. It may be "poor quality data", but still usable within it's limitations...

Regards,

Chris

[ettore]

air speed check

Just an idea, or two...:

Could variations of the air flow, measured at the engine's intake at a given rpm, give a clue about pitots' malfunction ?

Would air speed measurments at the engine intake, compared against the tremendous engine's thrust, be reliable ?

Sorry, I'm not an engineer...

[infrequentflyer789]

Originally Posted by auv-ee

It seems that an alternative air speed measurement is needed to augment the pitot tubes, something that would be affected differently by environmental effects and other failure modes.

Already exists - BUSS. Not fitted on AF airbuses. There seem to be some widely varying opinions of how well it works. AF rationale for not fitting it is here:

Air France - Corporate : The BUSS or "Backup Speed Scale"

I believe BUSS is standard on A380, where it was optional on earlier models.

[voltage]

Preface: I have 250H+ TT and study for my ATPL exams at the moment.

syseng68,

I can tell you from my limited experience that it would not be very useful. The relationship between TAS and GPS GS would be the wind, and then between IAS and TAS temperature, ambient pressure, altitude and humidity.

The only way to know the wind vector is to compare GPS vector with airspeed and heading information. If you have lost airspeed, you have lost all information about winds.

Winds do change, especially around convective activity. If the margin between stall and overspeed is small, GPS groundspeed is just about useless.

In my opinion, using optics or engine pressure readings sounds more promising as an ASI backup, because it tries to *measure* airspeed rather than deduce it.

[sensor_validation]

One for the conspiracy theorists:-

When the 1st BEA interim report was issued on 2nd July 2009 the English language version was only 72 pages long, it didn't have all the appendices in the French version. The version with the same filename f-cp090601e1.en.pdf (created 28th July 2009) now on the BEA website is complete and is 128 pages long, but doesn't include the exact final position ACARS message, which is in the earlier version

Code:

2:10:34 #0210/+2.98-30.59

This is the same location described elsewhere, but was it removed because

a) They missed the fact that this position was slightly off-track (see takata post earlier re eurocockpit analysis)

or

b) The position is more likely to be @2:10:30 (higher priority messages inserted into sequence), which puts it some 20 seconds or so after the cascade of errors started, not just before. I'm pretty sure mm43 posted before about positions quoted at half-minutes, but maybe that was based on Brazilian Air Force extrapolation data - they also had a position at 02:14z.

or

c) It just wasn't supposed to be there in the first place (wasn't in the first TV leak),

Can't say its very important, but I was sure I'd seen the #02:10 position report, but no obvious revision marks to the document...

For all the AF447 positions from 00:10 in ACARS 2 dp precision use in

Flight Paths of Flight AF 447 and of the flights that crossed the zone around the same time

see

http://www.bea.aero/en/enquetes/flig...data/AF447.txt

[auv-ee]

Alternate airspeed measurment

Originally Posted by infrequentflyer789

Originally Posted by auv-ee

It seems that an alternative air speed measurement is needed to augment the pitot tubes, something that would be affected differently by environmental effects and other failure modes.

Already exists - BUSS. Not fitted on AF airbuses.

The description of BUSS does not make it sound very appealing. I was thinking of something that would be a precise speed measurement, as valid as the pitots, a few of which would provide more inputs to the voting logic and the Kalman filter.

Originally Posted by bearfoil

Redundancy is a siren on the rocks when one is up against common fault.

The idea is to add a speed sensor based on different physics that would not be subject to the same failure modes as pitots. I agree about "parallel" redundancy, like multiple pitots, where a common fault is possible, or the case of "series" redundancy where the pre-existing failure of a backup is not evident until the primary fails (double pressure seals come to mind).

[auv-ee]

Originally Posted by JD-EE

[ROVs] are lowered in a cage and turned loose with a tether back to their transit cage. The tether's length is limited. So their ability to move along in a search is limited. They'd have to search an area and come back to their cage to be moved to the next block lest cords get tangled.

Agreed that the cage or depressor weight (2-body launch and recovery) is the usual configuration for a deep ROV. However, it is not so that the ROV has to return to the cage to move the ship. When the ship moves, additional scope must be payed out from the ship to keep the depressor at the right altitude, and the ROV can tow its neutral tether as the depressor moves. When the ship stops, the additional scope is retrieved as the depressor settles under the ship. With a rough bottom, it is necessary to ensure that the neutral tether is not dragging on the bottom, lest it fetch up on something. I don't know if this is attempted with more than one ROV/ship; that would surely be more tricky. Agreed though, for reasons already discussed, that ROVs are not the right tool for a wide area survey.

[HarryMann]

Alternative Air speed sensor
Hot-wire anemometer, hot film anemometer? Would need a filtered air feed (which might ice uip, unless it was designed like a Dyson bagless vacuum), but they last for ages in windtunnels and engine Air Flow Meters
With temp can give dyamic head fairly well I believe (Ptot)

[alexd10]

Hello!

Maybe I am missing something, but since airspeed is essentialy neded for assesment of the lift produced by the wings, would'nt be more straightforward to measure/calculate the lift? It should be possible at least by measurement of the stresses in the wing structure. Or by measurement of the pressure on a section of the wing and by integrating the pressure on that closed curve.
Had anybody heard of such an approach?

[HazelNuts39]

engine P1

Originally Posted by ettore

Would air speed measurments at the engine intake, compared against the tremendous engine's thrust, be reliable ?

Not a bad idea at all. Very likely each engine has its own probes for measuring intake total pressure P1 and intake total temperature T1, which are used by the engine electronic control system (FADEC). The difference between Pitot pressure and P1 is engine intake total pressure loss, which is quite small in cruise. I don't know if the engine's P1 probes have ever experienced similar icing problems as the airplane's pitots.

HN39

[JD-EE]

Originally Posted by syseng68k

While gps is primarily a system to calculate position, it can also be used to provide fairly accurate heading and ground speed data, in conjunction with appropriate software.

With processing accelerometers provide you with distance covered despite their primary data being acceleration. Similarly you can differentiate GPS location (and time) information to get velocity and a second differentiation gives acceleration.

Note that with GPS you are differencing two quantities that have significant error dimensions relative to the difference obtained. So the error band on velocity is large. The error band on acceleration is even larger.

I have a pair of GPS receivers a few feet to my right. They are for precision time keeping rather than navigation. But I note the wandering of their time signals relative to each other with stationary antennas. "A few feet" is the translation I get on errors at about 1ns per foot. (That is accurate within a three percent.) If a plane is proceeding at a ground speed of say 300mph that's about (mumble 88*6 feet per second or about 440'/second.. If you take a reading once a second you have a speed range of perhaps 430'/second or 293 MPH to 450'/second or about 307 MPH. That may be accurate enough if the plane is not in distress. If the plane's dynamics change on a second by second basis, as with turbulence, it might not tell you quite as much. At a tenth second you have 23% error bands without some smoothing rather than 2.3%.

I'm not sure what speed accuracy is needed for being happy within a cockpit. I suspect 2% is acceptable and 10% is not. GPS is perhaps marginal. Over a 10 second interval GPS velocity determination is, of course, much better than my bogey numbers.

Another detail to remember, if you are dealing with high dynamics or simply high speeds, is that GPS tells you where you were at the time of measurement not where you are at the time you read it. To the extent that the distance traveled during that interval matters this detail is important to keep in mind. Fortunately the various filters in the system "take up the slack" on this detail.

She knows more about GPS than she would like to bother with. You guys have, for more than a couple decades, been relying on some work I performed while at Rockwell International in Anaheim on the satellites as designed in the late 70s. At that time it was a pasttime of mine to "annoy" AF officers over their silly "dithering" pointing out differential techniques render it pretty much moot. It does provide a nice way to gain precision from the oscillators, though.

[JD-EE]

auv-ee
I was concerned with that towing scenario and potentially snagging the ROV to cage tether on rocks given the mountainous terrain.

{^_^}