A couple of interesting developments in the MH370 story.
- A number of independent investigators have been working to determine whether the debris found on Tanzania is from a 777, and if so, which part it corresponds to. Mike Exner (who supplied the link to the image above), Don Thompson, Ge Rijn, Victor Iannello, and Barry Carlson all pitched in and have found that most likely the object is inner 1/3 section of a 777 right outboard flap. Mike and Don have written up a report which you can download here.
- Victor Iannello has long been working on a post-FMT route by which MH370 could have continued to fly on autopilot without human interference and still wound up outside the current search area. Many have tried to find such routes in the past and found it impossible to make them match the ping rings without arbitrary changes in direction and/or changes in throttle setting. Victor has at last published his route. It achieves changes in direction by having the plane follow a magnetic heading, and achieves the change in speed by imagining that the plane is descending at the lowest possible automatic descent rate of 100 feet per minute. It’s quite a clever piece of work by an ever-creative researcher. Of course, the fact that it is possible does not mean that this is what the plane actually did, as Victor himself has pointed out.
- UPDATE 6/28/16 #2: Mike Exner has informed me “I checked those photos yesterday and confirmed the compressor is a Chinese model powered by 230VAC/50 Hz. Not from any aircraft.” UPDATE 6/28/16: Reader Greg Holwill writes: “I have attached photos of a fridge floating out at sea in Mozambique while fishing… I am situated in Durban in South Africa. The debris is located at my lodge in Mozambique.” Here are some of the photos he sent:
Doesn’t seem aeronautical to me, but would invite readers to share their thoughts. I could post more pictures if people are curious.
Just for the record, the FCOM says that at 310 kIAS the speed indication
changes to IAS. Your interpretation is that the commanded speed changes at the same time. Your interpretation may well be correct, but maybe not. The FCOM also says that the autothrust “respects” Vmo/Mmo.
Just for the record again, the DSTG comment on the ‘filtered’ speeds near the turn at IGARI from the radar data is: “The speed estimates vary dramatically during
the first turn, which is not an accurate representation of the aircraft speed at this time. It is likely due to the mismatch between the assumed linear Kalman filter model and the high acceleration manoeuvre performed by the aircraft.”
@Gysbreght: Since we last had this discussion about whether or not 310 KIAS is maintained after the cross-over altitude, I have verified with Boeing pilot Ed Baker that my interpretation is correct.
As for the Kalman filter, it would tend to smooth the temporal data set. The unfiltered data set is likely even more extreme than what was presented in the DSTG figure. That is why I have tried hard to get the unfiltered data set that was supplied by Malaysia to the ATSB.
@ROB
“Your inability to come to terms with the obvious, ie pilot suicide,”
why would someone fly that long and turn towards Australia if he wanted to do a suicide?
Suicide motive simply doesn’t add up.
@StevanG
True. I have not found any behavioral psychologists or other mental health professionals who have endorsed the notion of pilot suicide based on observables before and during the event. I regard it is as possible, but extremely unlikely.
Various posters having strong opinions on what the condition of the recovered debris can tell us are likewise over-stepping their credentials. The reality is that it is the responsibility of the forensic scientists who actually have the debris to make that determination. Whether the flaperon was damaged by flutter or water impact is still an open question as far as I am concerned. I can only assume that the lack of input from either Boeing or the BEA is due to attorneys doing what they do best – obstructing progress.
The excellent work done to identify the Tanzanian debris further highlights the fact that Boeing and the ATSB know full well where that piece came from, and for reasons known only to them have not made that information public. Very very annoying to me. I have lost a great deal of respect for a number of organizations involved in this effort.
@VictorI:
Thanks for your reply. Well, I don’t know “Boeing pilot Ed Baker” but, as I sait, it is entirely possible that you (and Baker) are correct on this.
Regarding the turn at IGARI certainly something very strange happened there, but not what is shown in figure 4.2 of the Bayesian report. My takeaway from this figure is that after the IGARI anomaly the speed did not show large variations, which is in agreement with the radar data published in Factual Information.
@VictorI – A couple of years ago, someone on PPRune posted fuel burn values for speeds other than LRC and and also included altitudes lower than typical cruise altitude. I’ve copied that table and posted it below for your use. Perhaps it will help you to determine endurance during the last few hours of your proposed scenario? (in case the table does not convert, the number 38.5 6.08 is associated with a 2000nm trip at FL100 that burned 38.5mton of fuel over 6h8m. These compute to an average speed of 326kts and a burn rate of 6.3mton/hr)
From Boeing Data for B777-200ER.
Long Range Cruise. Engines RR Trent 892
All up Weight at start of distance, 200.0 Tonnes
Time in hours and minutes, fuel burn in 1,000s kg
Distance ————– Flight levels
(nm)_______100__________200___________300____________400
_________Fuel Time______Fuel Time_______Fuel Time________Fuel Time
_600_____11.5 1.50_______9.1 1.37________7.5 1.28_________6.6 1.22
1200_____23.3 3.18______18.8 3.12_______15.6 2.52________13.9 2.38
2000_____38.5 6.08______31.2 5.21_______26.4 4.45________23.5 4.19
@VictorI
I wasn’t necessarily thinking about technical failure.
Imagine hijackers having received only basic training to pilot a 777. They remove the pilots from the cockpit. Lock themselves in.
Someone from the crew gets in the EE bay, and switches off the flight director’s computers (or what ever necessary to disable autopilot). It gets much more complicated for the hijackers. They weren’t trained to fly a 777 by hand.
Soon they would loose control.
They would pitch down when stall warning is heard and pull up when pull up terrain warning sounds (flying a 777 at night is not easy without autopilot).
Anyway all speculation. But if you allow ROD/ROC of this magnitude, then you also need to consider flight paths that are much much more north than the current search area.
@VictorI
Yes, the raw radar data would be very useful. As far as the Kalman filter itself concerned, I commented early on after reading the DSTG book, that it seemed like an odd choice for that task. When you have all the data you are ever going to get (and it is a limited set at that) you can filter it any way you please. The strength of the Kalman filter is to combine the “innovation” new measurements with the weighted ensemble of old measurements to produce an “optimal” (in a least squares sense) estimate of the “new” position. When you have a complete set of measurements (past and future relative to the next data point) it makes no sense to use a Kalman filter.
Oh well, it is what it is. Depending on how the Kalman filter is tuned it may or may not act in a low pass fashion. The filter may actually amplify excursions from the true flight dynamics. Of course, it seems unlikely that we will have a chance to make this determination for ourselves.
A record exists online of the flights (carrying paying passengers)
for many months prior to 8th March that 9M-MRO made.
Does anyone know of any online record of the flights prior to
8th March that Captain Zaharie Shah made?
À plus tard.
@Lauren H: Thank you. I’ll try to make sense of it, although the weight varies from the starting weight of 200 MT and the speed varies from the average speed.
@DennisW: Thank you. I am assuming that the Kalman filter smoothed the data because when @sk999 digitized the filtered speed and track data from the DSTG report and numerically integrated over time, the resulting path did not show the sharp turn to the left after IGARI that has been presented to us graphically as radar data. We need to see the unfiltered data to understand what is real and what is an artefact of the filter.
@Victor!
I strongly agree that smoothing is the most likely result of the DSTG filter
Collected some B777 flap info – might be useful. https://www.dropbox.com/s/m5g4r42l6kw8ig0/B777%20Flap%20Info.rar?dl=0
@StevanG
At the risk of being a bore, I must repeat the aircraft never turned toward Australia.
Z carried out an act of political terrorism. This act required suicide.
Is that really so difficult concept to grasp?
B777 flap info – RAR file link above does not seem to want to work.
Try this one:
https://www.dropbox.com/sh/tndzz3ow4ipbd77/AABqZm72n43J1BdXEW5W6HUpa?dl=0
@ROB, You’re entitled to speculate that Zaharie conducted his vanishing act as the most inffectually passive-aggressive act of terrorism in history, but please don’t act indignant that others might fail to recognize it as self-evident truth.
Since Victor mentioned the DSTG radar-based speeds: after IGARI the groundspeed was 525 kt +/- 25 kt. Those speed variations correspond to altitude variations of +/- 1160 ft. Therefore the airplane was obviously not flown by autopilot/autothrottle, nor was there an attempt to avoid radar detection. I would expect that an experienced pilot handflying a 777 would be able to keep altitude variations to within +/- 100 ft.
Also, of course, an experienced 777 pilot would not be handflying, but would use the autoflight systems.
@Gysbreght: An altitude variation of +/-1160 ft is not what the DSTG report says. I quote from Ch 4, where a specific reference is made to “low altitude”:
In addition, the ground speed observed by the radar prior to 18:02 is relatively high and implies that the aircraft would have been at low altitude. This is likely to result in poor fuel efficiency, and in order to maintain flight for the duration indicated by the satellite data, the aircraft would have had to slow and increase altitude at some stage to conserve fuel. This is also consistent with a potential speed change between 18:02 and 18:22.
I think they are saying that if the true airspeed occurred at high altitude, the Mach number would be too high for sustained flight. At low altitude, the IAS would be high, but there would be enough engine thrust to maintain this, even if it inefficiently burned fuel. I don’t they are referring to speed/altitude changes from zoom/climbs and zoom/descents, but I really can’t be sure.
This concept of a low and fast flying aircraft has been mentioned before when fuel consumptions estimates were first made in the spring of 2014.
What a mess…
The authors of “Bayesian Methods” make several statements that illustrate poor understanding of airplane flight characteristics. Perhaps they should stay within their field of expertise – Bayesian methods. That statement makes no sense at all. The maximum TAS is at the Mach/IAS cross-over altitude. The ATSB interprets the average speed as M.82 at FL320. Before IGARI the airplane maintained FL350/M.82.
At a very early stage it was suggested to use all available means to look for debris. Many merchant vessels passes this area. If all of these vessels were informed using Inmarsat weather one could have spotted something. Guess where it could have come down that no one reported any floating debris .See link below
https://www.google.co.id/search?q=world+maritime+shipping+routes&client=safari&channel=iphone_bm&biw=320&bih=441&prmd=inmv&source=lnms&tbm=isch&sa=X&ved=0ahUKEwio4d7H_MjNAhWBvI8KHXp3DRoQ_AUICCgB#imgrc=NhUD3fxIUFOY0M%3A
@Gysbreght
“Those speed variations correspond to altitude variations of +/- 1160 ft. Therefore the airplane was obviously not flown by autopilot/autothrottle”
Very interesting! So why was the plane hand flown? Could it be that to disable what they wanted to disable, the perps inadvertently turned the autopilot off?
While the plane was overflying land one could assume it’s easier to keep her straight and level (city lights etc…). But once over water… no target points anymore.
Is there enough radar data to actually map (even roughly) where those speed/altitude variations happen?
Could the mid flight logon be an attempt by the perps to re-enable the auto-pilot?
All the experts still seem very reluctant to take into account the possibility of a (human) controlled flight after FMT till the end. Why?
Why propose a set -100ft/min descent after FMT and not stating this was set by a pilot?
Then explain another way how this could have happened.
I assume there is no other way.
But I like to hear another explanation then for I think it’s crucial.
IMO a problem now is many experts are still very reluctant to consider a (human) controlled flight till the end.
And I believe that refusing to look at this possibility will hampers progress.
In this stage of the case with the new found debris, the visible damage, the combined and related pieces there is good reason to take a possible all (human) controled flight, glide and attempted ditch into account in new scenarios.
A (human) controled flight also after FMT would offer a lot more plausible possibilities IMO.
The refusal to take this into account (for no apparent reason IMO) might turn out to become a missed change.
I hope the experts take notion of this.
@Brock McEwen,
My predicted end points for “straight” routes have shifted a bit NE up the 7th Arc. This has nothing to do with the searched area or ATSB fuel limits. It is simply a result of the inclusion of the elevated temperature effect on fuel consumption in my fuel model.
@VictorI,
I have had success using bicubic interpolation of the weight/speed/altitude/fuel burn rate tables in the FCOM. This allows my fuel model to interpolate all parameters quite accurately.
@DrBobbyUlich: OK. Thank you. I have in the past tried to match the data semi-empirically from first principles, but could not get to an acceptable error over a range of conditions.
@all,
Here is an example of a post-FMT route which satisfies all the satellite data: a True Heading of ~179 degrees with ECON mode and Cost Index = 52 at Flight Level 322 (this is the Flight Plan speed mode at a constant altitude):
https://drive.google.com/file/d/0BzOIIFNlx2aUNGxVNnB2b3lESUU/view?usp=sharing
This route is a best-fit for constant true heading using a new method. Instead of minimizing residuals, I match their statistical properties (RMS and Peak) to the probability density functions provided by ATSB (for BTO and BFO) and estimated by me (for air speed error and LNAV error based primarily on the inaccuracies of the wind data). No waypoints were used to constrain the fit (which also finds the best flight level and PDA).
The average engine PDA is 1.2% for a second-engine flame-out at 00:17:29.
The FMT occurs from 18:46-18:48.
This route passes very close to BEDAX and essentially directly over ISBIX.
It closely matches the statistical properties expected for all parameters, so it is a very acceptable fit.
My major reservation is that it requires a fairly brief and rapid descent timed to coincide with the 18:40 phone call.
And I think it’s a bit strange. Some time ago this -0.1 descent rate was discussed also.
I then asked the question which altitude the plane would end up starting this descent rate shortly after FMT till 7th arc if it was a set descent rate from that start.
I remember the one comment I got; with a neg. descent rate the plane would fly upside down..
Now the same issue is brought as a key element of a possible scenario.
Never mind. Just wanted to mention it.
@jeffwise
@sevenG
Inffectually act of terrorism in history? Why not just immediately commit suicide?
Diving a plane into mountain certainly attracts attention but it does not last long – 10 days max and the press shifts to a new story. But creating a mystery with grieving relatives refusing to accept their loss, creating an opportunity for inept politicians to expose themselves as being inept, gives the act of terrorism some longevity in the media … over 2 years in fact.
A good mystery needs to have a few clues to gain interest but cannot have a ready scapegoat – as that diverts attention and brings early closure.
So one possible scenario is fly a plane deep into the SIO, run it out of fuel to minimise any oil slick, ditch the plane carefully to minimise floating debris and take a pill or step out of a door. Passengers have to die early in the flight to avoid any interference or messages being sent.
Was that planned from start or was that Plan B – adopted in flight when Plan A did not seem to be working? Perhaps Plan A failed and then there was no viable alternative but to disappear protecting family and cause.
Finding the plane now, and solving the mystery, gives a second bite for the media and more exposure.
So ineffectual act? … time will tell.
Why not just get the suicide over rather than prolong it … delay to best manage the finishing touches just as ROB has postulated.
All this is conjecture – it is just my opinion.
@DrBobbyUlich:
“I have had success using bicubic interpolation of the weight/speed/altitude/fuel burn rate tables in the FCOM. This allows my fuel model to interpolate all parameters quite accurately.”
In view of many unknowns I don’t think that increasing the numerical accuracy of interpolating between tabulated values has much added value. LRC (Long Range Cruise) speed is a unique function of weight and altitude, defined by the characteristics of engines and airframe. Interpolating in a LRC table doesn’t tell you anything about fuel consumption at speeds that differ from the LRC speed. You could of course interpolate between LRC and Holding speeds, but with only two speeds “bicubic” interpolation isn’t going to help either.
@VictorI. Thanks for your reply: “The trajectory before 19:41 could be modified to fit Kate Tee’s observation. I have deliberately avoided doing this because it is not clear what she saw.”
Yes, I agree it is best to avoid this at the moment. I’m inclined to believe her story that she saw a glowing plane but the detail is too vague. However, even if we take the radar and ISAT data as correct, IMO we can only speculate as to what event(s) changed an apparently deliberate flight approx. westward into a flight approx. southward. So until any real facts come to light I am keeping her observation on my “table” of possibles.
Anyone else notice the “orange glow” of the wing fire in the video of SQ368 (Singapore Airlines B777-300) today? One problem with a fuel fire is how did it still have enough fuel to fly for 6 more hours?
@ALL
Referring to the recovered debris for a minute, any parts of the inboard flap, RH or indeed LH, are conspicuous by their absence.
The RH flaperon turns up, also a chunk of RH outboard flap, plus a piece of flap fairing, two flaperon closing panels and a piece of (presumed) 661FB trailing edge panel adjacent to the outboard flap fairing, but nothing from the inboard flap. To me that is unusual, especially when you see how far it extends below the wing when lowered. And no bits from the LH wing, either.
One possible explanation is that the RH inboard flap wasn’t extended when the aircraft ditched (assuming of course that it ditched, as opposed to crashed). There is built-in protection against asymmetric flap extension, but if for example, there had been a problem with hydraulic pressure and the pilot had needed to revert to secondary (slower) electrical extension in the critical final moments, then possibly the RH flap might not have extended.
If there was non-extension of the RH inboard flap immediately prior to ditching, the aircraft would yaw to the left and roll to the right. In this scenario, the right wing trailing edge dips into the water, and the flaperon and outboard flap are ripped off. The RH stabilizer “no step” panel section is ripped off. Cabin door 1R is broken in, releasing the interior panel section. However, nothing breaks away from the LH wing.
Pure conjecture, possibly, but it could explain why such a oddly unrepresentative assortment of aircraft parts have turned up.
@jeffwise
I accept what you say about my conjecturing. I realized afterwards that I had possibly overstepped the mark. Apologies are in order.
I think much to much is being made of the fact that we have more pieces from the right side…so far. As more debris is found, that trend will probably reverse. But if one needs a reason now, the best one I can think of is that the aircraft was turning, and quite possibly inverted at the time of main fuselage impact. That is what I observed in every simulation. It is also what Boeing observed in their sim’s. Logic suggests that the debris should be more random if there was an attempt to ditch.
BTW…It would require very poor pilot skill to ditch in such a way that the interior of the aircraft was shredded along with the exterior. It is time to acknowledge that the impact was a very high energy event.
@Ge Rijn
“And I think it’s a bit strange. Some time ago this -0.1 descent rate was discussed also.”
“I remember the one comment I got; with a neg. descent rate the plane would fly upside down..”
could flying upside down really mean heading North?
This projected flight path is very close to Drifting modeling that was posted on http://www.duncansteel.com/ I wonder how much time and money it would cost for ATSB to search this area
@Ge Rign: There is nothing strange about the -100 fpm path resembling the -0.1 FPA path. I generated both with similar assumptions. In both cases, I was attempting to find an automated flight path the ended well north of the current search zone.
@DrBobbyUlich: Thank you for sharing a new path. I think it is helpful to demonstrate any path that ends outside of the search zone and satisfies the satellite data. Because of what the drift studies suggest, I am most interested in paths ending further north. However, there might be rationale for searching north of the current zone over a substantial length and small width.
If I am reading your table correctly, it looks as though the true heading varies by about 1.7 deg. Do you think that is reasonable?
@Lauren H
Yes a fuel fire and +6 hours flight seem incompatible (I’m no expert); also no signs of fire on debris that I can see. [How lucky for SQ368 passengers and crew to escape that horrendous fire with no injury; is it normal procedure to put fire out first before passengers disembark?]
@Gysbreght,
I think you misunderstand my interpolation method. I am not doing what you are criticizing. Let me explain in greater detail. My bicubic interpolation method is first used within a single table to interpolate the fuel flow at LRC (or whatever speed the FCOM table is for) depending on values of altitude and weight that are in between the tabulated row and column values. It also interpolates the LRC Mach number in the same fashion (again, within a single table). I do not do any interpolation between tables (i.e., between LRC and HOLDING, INOP, etc. tables). For ECON mode only, I use a separate set of polynomial equations that scales the ECON speed and fuel flow for a given Cost Index relative to LRC values.
In summary, my fuel model is only accurate (and only used) at the speeds given in each FCOM table, plus the interpolation between LRC and MRC using the ECON mode equations. The ECON equations are not given in the FCOM, but I have derived them based on articles on this subject written by Boeing and Airbus. Since the range of interpolation between LRC and MRC parameters is only about 5% or less, I believe these equations, while approximate, are probably accurate to better than 1%.
@VictorI,
The question you raised regarding LNAV reconstruction accuracy is relevant to any route fitting exercise.
For Great Circle and True Track routes, I do not allow any variation along the route. Only a single initial true bearing is fitted, and it suffices to completely define the True bearings of the subsequent legs.
For True Heading and Magnetic Track routes, I allow a standard deviation of 0.5 degrees among the LNAV bearings of the post-FMT legs. The error sources are different in these two cases. For True Heading, the errors come about because of our limited knowledge of the cross winds. There are errors in the wind speed and wind direction data, and the coarse time and spatial sampling, especially in altitude, will not allow us to exactly reconstruct what is fact is a somewhat meandering path which I approximate with a series of rhumblines for each leg. There is an additional source of error as well, because I sample the wind data at the ends and in the middle of each leg. So I have a limited number of wind values to integrate for each leg to figure the average angle difference between track and heading. Taking all these error sources into account, and doing some approximate calculations, I estimated a 0.5 degree RMS difference between the actual path and its reconstruction for True Heading routes.
For Magnetic Track LNAV, the error comes about because the magnetic declination model I use is in 1 degree cells in latitude and longitude, and the values are in integer degrees. This quantized magnetic declination model prevents 100% accuracy in reconstructing Magnetic Track routes, and again I estimate about a 0.5 degree RMS accuracy.
For Magnetic Heading LNAV, the reconstruction error will be larger because you have two error sources: about 0.5 degrees RMS from the magnetic declination model errors and about 0.5 degrees RMS from the wind errors. I simply combined (RSS’ed) these two errors to estimate a 0.7 degree RMS reconstruction error.
Finally, just to make sure that one leg did not dominate the RMS error budget, I also applied a peak error limit of 1 degree. Thus for all routes the peak-to-peak LNAV error will always be less than 2 degrees.
To answer your specific question, yes I think these are reasonable estimates of the LNAV reconstruction errors, based on the methods I described above.
I will also mention that the wind errors affect the headwinds/tailwinds, and thus the difference between air speed and ground speed. These errors will affect the RMS residual speed errors remaining after the BTO data are satisfied, and I estimate the magnitude to be ~1 knot RMS. It would be larger in general, but for the post-FMT route the winds are mostly close to East or West, so the headwind/tailwind magnitudes are much smaller for a southerly route, and the reconstruction error in air speed should be a fraction of the total wind speed error (which I estimate at several knots). Thus my objective fitting function seeks to achieve a 1 knot RMS air speed error compared to the speed model being used.
@ Lauren H,
Yes, Kate Tee noticed it and commented on twitter that it was the closest to the glow she had seen that night, that she had come across since.
@MH @Victorl
‘could flying upside down realy mean flying north?’. I don’t know. Do you?
The discussion on this stopped soon afterwards.
As I understand now set -100ft/min descent and -0.1 FPA are two different things.
I meant the first at the time. Probably this is was why I got that answer.
Never mind as I said. I think the proposal from @Victorl is interesting and important.
For it suggests a credible possible flightpath combined with a more north/east crash area and maybe more important; opens up renewed interpretations of data, witness accounts (Kate Tee) and evidence (debris).
@airlandseaman
With all respect I like to object your conclusion it must have been a very high speed impact at this time and accept this as a fact.
‘Very poor pilot skills required to shred the interior’ you say.
I ask you to take a look at the interior pictures after the ‘low speed’ crash landing of Asiana 214:
http://www.nycaviation.com/2013/07/photos-inside-the-asiana-214-wreckage-and-cleanup/
The exterior parts that seperated during this ‘low speed impact’ are conform the ones they find now from MH370. Even a door got seperated hardly damaded in this crash landing (the red left aft door).
Further I like to remind you of the ‘perfect ditch’ on the Hudson.
Although a perfectly executed level ditch on a fairly smooth surface only the left engine seperated among with several control surfaces (including an outboard flap).
The right engine stayed attached but some control surfaces also seperated or got badly damaged:
https://www.youtube.com/watch?v=xdcyG49tsno
I suggest it’s too soon to exclude a ‘low speed impact’ and to keep this possibility in mind.
@ventus45
The diagrams of the flap hinge construction you posted via dropbox are quite interesting.
There is one thing that might be of particular importance I noticed.
The follower link of which one end is still attached to the found flap section is in two diagrams shown in its extended flap position.
In the picture above in Jeff’s article you can see this follower link end also is an angle which it would be if the flap was extended.
Assuming this follower link end got fixed in its position when the flap seperated it would suggest the flap was extended when it seperated.
Offcourse I don’t know if this follower link end is fixed on this debris in this position or still free moving but I think it could be important to keep in mind.
‘nough said about the fluttering inverted supersonic spiral dive!
@Ge Rijn
I have great respect for ALSM’s integrity in these matters, but even a cursory examination reveals the high speed crash scenario to be totally untenable. If the interior was shredded as he suggests, then where is the debris to back it up? One piece of interior partition that came from just inside of door 1R, and one LCD frame that, as you point out, could very easily been dislodged during a ditching attempt.
The point I wanted to make is this: The thing that’s most significant about the debris recovered so far is that it indicates something went very wrong as the aircraft was in the process of ditching. There was a loss of control. If the APU ran out of fuel just before ditching, there could have bren a problem extending the flaps. The APU is reckoned to have 13.5 minutes maximum worth of fuel, once the LH engine flames out. If this indicates he only glided for about 13 minutes, it puts an upper limit on the distance he could have glided. Certainly it would be less than 100Nm, more likely closer to 70Nm. The take away from this: the wreckage is possibly only a few miles south of the current search zone.
This serves to illustrate just how useful an analysis of the debris can be to the search effort, I one is prepared to make the effort.
@ROB:
RE: “The APU is reckoned to have 13.5 minutes maximum worth of fuel, once the LH engine flames out.”
The 13.5 minutes worth of fuel available to the APU has been determined “in a standard flight attitude (1° pitch)”. With both engines flamed out, the flight attitude and/or acceleration would be different and most of that fuel would not be available to the APU because it moves forward in the tank.