About That MH370 Inmarsat Data…

Earlier this month France announced that it will reopen its investigation into the disappearance of MH370:

French newspaper Le Parisien reports that investigators are keen to verify data from Inmarsat — the British operator of a global satellite network — which tracked the aircraft’s pings to the southern Indian Ocean off Western Australia, where it is believed to have crashed.

I was happy to hear that, because for the last four years I’ve been making the case that there is one known way by which the Inmarsat data could have been falsified as it was being transmitted from the plane. This falsification would make the plane look like it was heading south when it was really heading north, and would explain why an exhaustive quarter-billion-dollar search of the southern seabed found no trace of the plane.

Of course, there are other reasons to suspect that the plane went north. One of the less probative but more elegant is the simple fact that when it was last spotted, that’s where the plane was turning. The above image comes from page 4 of Appendix 1.6E of the latest Malaysian report, entitled “Aircraft Performance Analysis,” prepared by Boeing. I think this appendix is one of the most important sections of the whole report, as the authority of the source is unimpeachable and its assertions are laid out with such clarity. In this image we see a summary view of what is known about the first two hours of the plane’s flight, based on a combination of secondary and primary radar as well as the first ping from the Inmarsat data. It shows, as I and others have pointed out, that after an aggressive turnback at IGARI, and a high-speed flight over peninsular Malaysia and up the Malacca Strait, the plane disappeared from primary radar and then turned to the north.

Some have proposed that this is best explained by the assumption that whoever was in charge of the plane wanted to avoid conflicting traffic on the airway, but that is absurd–there was no conflicting traffic, and anyway it would be very simple to avoid any such hypothetical traffic by flying at a nonstandard altitude. A simpler explanation is that they turned to the north because they were heading north.

The report has another similiarly compelling illustration that combines fuel-burn data with ping-ring distances to illustrate the various routes the plane might have flown, assuming a constant altitude and turns only at ping arcs:

This picture neatly illustrates a point that the DSTG arrived at more conclusively through the heavily application of mathematics: namely, the only straight-ish flight paths that wind up at the 7th arc at the correct time and distance for fuel exhaustion are ones that fly around 450 to 475 knots, and at relatively high altitude. This is where the Australians originally looked for the plane, and really it was always the only rational place to look.

The absence of the plane in this area could have told the authorities two years ago that something was up–and that would have been the right time to start being suspicious about the Inmarsat data.


479 thoughts on “About That MH370 Inmarsat Data…”

  1. @David

    That should of course read “appealing crime” not “appealing crime”. I may be an exceptionally clever guy with an IQ of 151, but even I can fall foul of auto complete after downing a can of Goose Island IPA.

  2. @Rob, I don’t think the second “@Rob” should be automatically considered a troll, more likely he is a recent arrivee who isn’t aware of your history of contributions. I’ll try to keep an eye on things to avoid confusion in the future.

  3. @Ventus. You say, “The French have NOT confirmed the flaperon came from FLIGHT MH370 of 8th Mar 2014. They confirmed it came from “the accident aircraft” 9M-MRO.

    Yes there is a clash between what the Malaysians say in their debris summary which I quoted and the DGA report, which includes, “On 03 September 2015, confirmation by the Paris Public Prosecutor’s office that the flaperon did in fact belong to aircraft 9M-MRO”.

    Without the document from the Public Prosecutor we cannot say, though I agree it does look like at least a slip-up by the Malaysians.

    I think the fire business has been addressed by others.

    About the evidence the left flap part came from 9M-MRO, you say, “Lots of assumption and assertions, but no proof.”

    The ATSB investigation found, “The flap manufacturer supplied records indicating that this identifier was a unique work order number and that the referred part was incorporated into the outboard flap shipset line number 404 which corresponded to the Boeing 777 aircraft line number 404, registered 9M-MRO and operating as MH370.”

    Your judgement as to what reasonably can be accepted as ‘proof’ for the purposes here is different to mine.

    As to the barnacles on the flaperon I agree this remains an anomaly. Jeff’s post below and that of others indicates this barnacle type attaches underwater, not at the surface.
    An explanation that after they attached the flaperon flipped continuously, so keeping all customers happy, is unrealistic since it is unlikely that conditions would see to. However that does raise the question as to how long they survive when not at least awash or dry and what that exposure does to their growth rate even should they survive.

    The impression I have is that while the DGA estimates it had 37 kg of buoyancy from flotation tests a month after recovery there might have been less than that when it beached, but that is purely speculation.

    We seem to have drifted way off the 4 cracks on the leading edge.


  4. @Rob. Assuming as you say that the choice of flight, fuel and turn back was all part of his plan and his purpose was to get as far as he could within those limitations, what he could have done is not fly as far north/tarry on the way as looks apparent and he could have step climbed etc to maximise range. Judging though from consistency of the apparent fuel exhaustion with there having been no pilot and the various arcs it looks like he didn’t.

    You say,”The ruse worked because they were searching the SCS for several days.” Well it looks reasonable to suppose he was unaware that INMARSAT would be able to extract enough information to bust his general later route but surely he would be aware that military radars would have tracked him back? He could not have supposed there would be a stuff up in timely unearthing of that or that this would never be discovered. That he gained a couple of days makes little difference to his plan being flawed, that is supposing his intent was to simulate a crash in the SCS somewhere, without trace. Besides wasn’t it so that unanswered calls from the ground disclosed he was in the air hours later? He would have to be unaware of that possibility also.

    You say, “He didn’t want to be busted before he even got beyond the Malacca Strait.” With the above in mind he should have known he was going to be busted anyway. All he gained was a few days and he could not even have counted on the stuff up which led to that, IMO.

    Separately, many pilots (and Larry Vance) had/have in mind a powered flaps-down ditching. That would not necessarily entail much loss of range were the glide at low power or engines shut down and restarted lower down (windmill/APU) though an explanation for the 7th arc transmissions would be needed for the former. Your view?

  5. @Rob
    Seems to me the SIR report indicates MY probably knew MH370 U-turned at IGARI by morning. So I agree with Rob there was an attempt to make it look like a crash at IGARI, but that was only for temporary purpose. Turning SDU back on at 18:25 marked the end of the complete stealth strategy.

    Razak hunted SCS four days, probably because he was covering up that he knew that a MAS pilot flew off with a B777.

  6. @Ben, these kinds of one-source stories are frustrating, because they show no journalistic integrity and exist purely for the clicks.

    There is plenty of physics, not to mention photographic evidence, to show us jet aircraft, rarely crash in pancake fashion, particular in jungles or forests, where wing meeting tree tends to have a destabilizing effect….

    Do a google image search for TU 154 and Polish president Lech Kaczynski, whose plane crashed outside of Smolensk, Russia in 2010.


    Or see Air France Flight 296 or PK 661 or the chartered Avro that crashed Rionegro in 2016. You just don’t put a plane down on a forested mountain side and leave it model perfect.

  7. @TBill

    The last sentence of your post September 5, 2018 at 8:39 AM would be closer to the truth if you deleted ‘MAS’ from it.

  8. @SteveBarratt
    I agree partially, MY at least probably covered-up that they knew the aircraft diverted. But there would be no reason to hide that radar path, unless there was something poltically-sensitive going on. Instead we had to wait for Inmarsat, and President Obama via his press secretary, to blow the whistle.

  9. @TBill: Victor said “Although I believe the hard facts are most consistent with a deliberate diversion under the control of the captain, I have to admit that I am biased towards believing the diversion began with the intention to deliver the patients safely to an alternate airport. I am not a fan of the scenario in which there was a controlled depressurization over Kota Bharu, but again, I freely admit that could just be my bias. Despite my bias, I’m trying to keep an open mind.

    The first officer could have been incapacitated by the captain. For instance, upon returning to the flight deck, the captain could have injected him from behind with a sleep-inducing drug”.(!)

    What a hoot! Hilarious or what? Victor has obviously been watching far too many repeats of “The Man From UNCLE”!

    Well, at least Victor admits he is biased!

    BTW ALSM and Richard, the 18:40 BFO can be best explained as showing a gradual climb of about 300fpm at that point, commensurate with the closing stages of the climb to 40,000ft, the final cruise altitude en route to S38. DrB will confirm the climb rate.

  10. @Rob
    Well I think we all have our own hypotheses and bias, and it is good to know where each other are coming from. Especially now we are at crunch time hoping the case does not go inactive.

  11. If I may return to my post of August 21, 2018 at 10:59 AM –

    I have determined that the second engine never failed until impact. That applies to all end-of-flight simulations in abnormal electrical configuration, where the autopilot was lost when the first engine failed.

  12. @Gysbreght. Interesting.

    Would you expand on that and on whether your conclusion applies more widely than these specific simulations?

    Presumably Boeing and the ATSB would be aware of this?

  13. @David: I’ll have to write a paper to explain. Of course it applies only to the simulations conducted in 2016. In the abnormal configuration the second engine fails only if the fuel available to it at the first engine failure is exhausted before the impact, typically 4.5 minutes after the 1st flame-out.

  14. @Gysbreght
    Nice proof…in so many words it is a quicker dive, producing BFO’s something like we see (in theory)….but is there time for the SDU reboot sequence to give us the BFO’s?

  15. @TBill: “….but is there time for the SDU reboot sequence to give us the BFO’s?”

    No. THe SDU reboots and sends its log-on request two minutes after the loss of electrical power from the first failing engine.

  16. @Rob

    ‘..For instance, upon returning to the flight deck, the captain could have injected him from behind with a sleep-inducing drug”.(!)’

    As an anaesthetist (anaesthesiologist in the USA) gaining venous access from behind a patient is extremely difficult. Maybe IM ketamine or insulin but this will take some time to work. Also there is the issue of the FO’s mobile login which Captain Zaharie would have immobilised. A third party in the E/E bay would more likely have missed this.

  17. @ Gysbreght. Thank you for your analysis. Comments:
    1. The ‘hiccup’ in Case 1 Fig 3. A thought. TAS has a consequential rise (Fig 4), presumably due to a nose drop followed by recovery. Thrust would be from increase of the forward component of the weight vector.
    2. As was your intention you have demonstrated that the second engine did not fail before the crash if the first that failed carried the AC load. That is of course consistent with the endurance of each engine from Case 1, where the endurance gap was 16½ mins.
    3. I draw more from your work though. Looking at your Fig 4 and the ‘average’ speed between the right failing at 115 secs, the autopilot remaining operative until after the left’s failure at 1100 secs, in between the aircraft will have travelled some 90 miles. However in the case of AC loss shortly after right engine failure and loss of autopilot then, the aircraft will not realise that advance. That loss would affect getting to the 7th arc. Besides, since it would have failed 10 mins before reaching the 6th, which nominally is 6½ mins of engine running from the 7th, that would prejudice it reaching the 6th on time as well. To avoid the right’s failure slowing the aircraft before the 6th arc thus requires the fuel exhaustion gap to be no more than 6½ mins.
    4. Related, returning again to your Case 1 where the AC load was shared, your Figs 3 & 4 indicate that the aircraft started descending 6 mins after the right’s failure. Had it failed 10 mins before the 6th arc the aircraft would have been descending at the arc.
    5. In brief summary if a large gap of around 16½ mins between engine failures is valid, right engine failure resulting in autopilot loss would make getting to the 2 last arcs problematic. (Also, with the MH370 right engine consuming more than the left, supposing the 16½ mins to be typical of the electrical load being shared during the long leg south, that will be increased should instead the right be carrying all the electrical load. So without fuel transfer to the right tank or appreciably more starting fuel in that the gap would be more).
    Also, Victor Ianello has said that all 4 simulations based on this scenario yielded descents which were consistent with the descent rates and acceleration calculated from the final BFOs, if not their timing. So did 1 of 5 of the AC-shared scenario: Case 5. If the above rules out the right engine generating all AC at its failure then consequentially that rules out all simulations bar Case 5 as being compatible with those descents. It may be also that Case 5 is incompatible with the timing.
    6. I should add about that while your Case 6 is consistent with the BFO descent rates and acceleration it does so at approaching 4 mins after AC loss (your Fig 9), not the 2 the ATSB specifies. Besides, this is at above Mach 1. That is beyond the flight envelope and probably even the Boeing simulator data base so that outcome may be unreliable, as Boeing has warned, which raises a like question about others in that group.
    7. Finally, in Case 1 and its 5 brethren there could have been a left engine relight, that possibility not being catered for in the simulations. That too could affect their outcomes.

  18. @David:

    Thank you for your reply.

    (1) “Thrust would be from increase of the forward component of the weight vector.”
    No, the flight path angle FPA is part of the determination of specific excess thrust: (T-D)/W = SIN(FPA) + (dV/dt)/g

    (6) “Besides, this is at above Mach 1. That is beyond the flight envelope and probably even the Boeing simulator data base so that outcome may be unreliable, as Boeing has warned, which raises a like question about others in that group.”

    The Boeing comment about “In some simulations, the aircraft’s motion was outside the simulation data base” applies to Cases 5 and 6. Case 6 had the lowest maximum Mach of the four abnormals:
    Case 3 M1.01; Case 4 M1.03; Case 6 M.96; Case 10 M1.02.
    (Caveat stated in the paper applies).

  19. @David
    > Alt Config looks good explaining 00:19 BFO’s
    > As VI noted last month, we are assuming no fuel leveling by the pilot to pump fuel over to the right tank to equalize.
    > David’s comments about Normal Config difficulty reaching Arc6 and 7 I think that depends on heading. Heading due South in passive flight (like Boeing simulations) necessitates almost 480 knots if I recall to cross Arc6/7. But there are shorter pathways as low as 300 knots, if you take the short “leg” instead of the slanty “hypotenuse”. But overall Alt Config looks better at the moment, which I guess could get over the Arcs at full speed.
    > Gysbreght- what is counter-intuitive is we have the Left Engine running but the aircraft is nonetheless spiraling hard left. But I guess the rudder is over-powering or something like that. I’d say wind was pushing it but the No Wind case does the same darn thing.

  20. @TBill:

    We have Cases 3 and 6 spiralling hard left, and Cases 4 and 10 spiralling hard right.
    Wind has no effect on direction of turn.

  21. @David
    I made a 30-sec YouTube video of a flight sim hypothetical “skipping” crash that I thought might illustrate how the flaperon trailing edge could get damaged in a flaps-up hard “ditching”. Speculative only as this goes against the prevailing wisdom of a catastrophic hard dive, and would require either the pilot taking charge after the wings dipped in Gysbreght examples, or somehow the aircraft was more stable in the passive case.


  22. @TBill: The key to the question why the airplane turned left or right must be found in the starting conditions that have been selected by the ATSB and specified to Boeing.

    The ATSB has now released the results of the simulations conducted by Boeing, but does not reveal the conditions that they (the ATSB) have selected.

  23. @TBill. I think you might have misunderstood. Terminology maybe.
    It is those of the ‘alt’ configuration which have a problem reaching the 7th arc and even the 6th because they turn off course some 90 NM earlier than the normal configurations. See, “However in the case of AC loss shortly after right engine failure and loss of autopilot then, the aircraft will not realise that advance. That loss would affect getting to the 7th arc.” etc.

    The “AC loss shortly after right engine failure” is synonymous with alternative configuration.

    The point is that there is only 1 of 10 of those simulations which is likely both to reach the 7th arc and prompt the BFOs there, the alts being ruled out. As I noted, that is one of the 6 normal configs, Case 5. However even that one’s timing may not be consistent.

    Besides the alt simulations having little chance of being compatible with both the arcs, the Gysbreght ‘alt’ example fails on descent timing and is at a speed which most likely takes the analysis beyond where the data are representative. Gysbreght has added (I think) that other ‘alts’ likewise are at speeds likely to be unrepresentative.

    The normal configurations have a maximum 1 in 5 chance, if this small selection is representative.

    I should add that there is a corollary to the increase in right engine fuel consumption if it carries all electrical load for some time. That is the left could have done the same, reducing the endurance gap between the two beneath the nominal 16½ mins.

  24. @TBill. My third last para graph should begin, “Besides …..with both arcs and descents, the Gysbreght…”

  25. @TBill. About your video, I am asked to create a channel if I want to ‘upload’. Would you check please?

    A ‘skipping’ crash would require a strong aircraft to avoid break up during the bounce but also I would have thought the engines would stop that even if they separated in the process.

    For there to be much of a spread any initial skip might need repeating. Intuitively that to me would seem unlikely even were seas calm.

    Still maybe your video will be help.

  26. What deductions can be drawn from the content and existence of the new ACARS msgs released after Victor Ianello’s blogging? Any thoughts?

    And do we know—for certain—that the reports of contact with the aircraft and reports of “muffled voices” from a nearby plane are indeed mistaken?

  27. @David
    …will check tomorrow on video. Basically suggests to me that a flaps-up, high speed “ditch” could explain the trailing edge damage and separation. Seems me the active pilot argument has to be a long glide to water surface, as I think @Rob was saying.

    …the muffled voice is unsubstantiated, and if true, unkown if it was a reply form MH370. So there is nothing useful we can extract from that account.

    As far as ACARS msgs, the jury is still out, but the main thing so far is that there was useful data (an unreported attempt to contact MH370) that was withheld. In itself that does not help too much, but it implies there is more hidden information out there.

  28. @Gysbreght. Load factors of 1.3 and 2.2 for your 2 cases are not going to break the aircraft. Supposing the possibly-unreliable data for Case 6;
    • do not distort its outcome much and,
    • other simulations’ load factors are also low and,
    • that the simulations are sound, despite them probably being inconsistent with the BFO derived descents; then,
    • there was no mid- air wing break up due to overstress.

    I do not think that striking the sea at a descent rate of around 400 knots in a dive of about 45˚ as in Case 6 would allow the sequence which apparently caused the outer flap internal damage; and the steep bank then would hardly matter. Also, at that descent rate, higher than SwissAir Flt 111, there would have been very small pieces surely.

    Which brings us back to a flaps-up ditching (for damage compatibility) after a pilot induced, or did not prevent, a steep initial descent (BFO compatibility) and then broke the right wing backwards and upwards in that ditching. (As you know I have theorised that a wing break is the as-yet best fit with the flaperon and outer flap damage).

    A piloted scenario like this has been your and some others’ view as to what happened. It does entail a glide and not a steep final descent at the end of that. Lack of IFE connection via satellite looks explicable. Putting aside also whether a pilot inexperienced enough to get into the high descent would be able to recover and assuming he could, where would he go, inexperienced and quite possibly with no plan, just reacting? A rhetorical question I think, the point being that the scenario does not help much with where next to search.

  29. @David: “Putting aside also whether a pilot inexperienced enough to get into the high descent would be able to recover …”

    Your needle seems to be stuck in a worn-out groove. Recovery does not require any special ability. All the person has to do is to relax the forward pressure on the control column and to make small roll inputs to counter the airplanes tendency to bank. As shown in the simulations, except for the tendency to bank, the airplane recovers itself.

  30. @David: “where would he go, inexperienced and quite possibly with no plan, just reacting?”

    His natural instinct would have been to avoid crashing into the ocean, hoping to survive somehow.

  31. @Niels: Regarding your questions on the VI blog:

    The chart in Anderson’s paper is for a given altitude, then MRC Mach varies with weight.

    The optimum altitude increases as fuel as burned. To remain at the optimum altitude would require step climbs during the flight.
    The MRC Mach at the optimum altitude is constant. Step climbs are required when under ATC control.

  32. @Niels:

    The ideal profile is what was done in years past, when oceanic traffic was few and far between, and “block clearances” could be given, which allowed the “drift climb” or “cruise climb” at constant Mach. The aircraft was trimmed to constant Mach and automatically climbed very slowly and gently as weight reduced with fuel burn. U-2’s still do it, but only in the free to play playground, high above controlled airspace.

  33. @Ventus45: Thanks for that addition.

    I could have added that the optimum altitude is defined by a constant ratio of weight to atmospheric pressure, Weight/delta, where delta is the atmospheric pressure ratio altitude/sealevel.

  34. @Gysbreght
    Assuming a hypothetical straight glide scenario in the alt config, presumably we have about 15 minutes of fuel one engine (0.25 x 400 knots ~= 100 nm) followed by powerless glide from say FL220 about 50-75 miles. So we are possibly somewhere in the range 175 nm from Arc7 worst casse scenario.

    Agreed? or better estimate from you? I am assuming, in this case, that right engine fails about 2 mins before Arc7.

  35. @David: You commented on 11 September:
    “The ‘hiccup’ in Case 1 Fig 3. A thought. TAS has a consequential rise (Fig 4), presumably due to a nose drop followed by recovery. “

    Good point. I’ve edited that aspect in the paper I posted 10 September.

  36. @Gybreght
    OK thank you. I forget if @David mentioned this already, but in the Alt Config the left engine saves even more fuel due to not driving the IDG (left generator), on the other hand, the APU is I guess sucking fuel out of the left tank. So for now I assume it’s a wash, but it would nice if ATSB could tell us how fuel was left.

  37. @David: Compare Figure 13 CAS to the description of a simulator exercise Andrew reported on VI on March 17, 2017 at 9:24 pm:

    “The aircraft maintained level flight in VNAV PTH until the stick shaker speed (168 KIAS). The aircraft then pitched down and initially accelerated to about 180 KIAS before stabilising at 172 KIAS with a descent rate of 850 ft/min”

  38. @Gysbreght
    One less likely possibility in Alt Config is that autothrottle is lost so the left engine could go to idle. But I believe your analysis suggests full throttle is maintained on Left Engine until FE, in this simulation.

  39. @TBill: I think in the Alt Config the autothrottle is lost and the remaining engine stays at the thrust it had prior to the engine failure, not idle and not full thrust.

Comments are closed.