The Reasoning Behind the Continued Search of the Southern Indian Ocean for MH370

April search area small
The soon-to-be completed search area, in purple and orange.

Earlier this month, at a meeting between ministers from Australia, China, and Malaysia, the countries involved in the search for MH370 announced that, in the event that the plane was not found within the current search zone by the end of mission in May, the area would be increased “to extend the search by an additional 60,000 square kilometres to bring the search area to 120,000 square kilometres.” (The new area is outlined in red in the image shown here.)

I think it’s worth considering the logic behind this decision.

Last year the ATSB spent months carefully calculating the boundaries of the original 60,000 sq km area. What they wound up with was a rectangle about 1200 km long and ranging in width from 48 to 62 kilometers wide, straddling the 7th arc.

This area fit what the ATSB believed to be the most likely scenario for the final phase of the plane’s flight: that it flew straight on a southerly heading on autopilot and then shortly after 0:11 ran out of fuel — first one engine, then the second. After the second engine stopped, a backup system called the Auxiliary Power Unit (APU) would have kicked in, restoring a limited amount of electrical power. The plane’s satellite communications system would have rebooted, leading to the final “half ping” at 0:19.

As soon as the second engine failed, the engine would have entered a unpowered glide, much as the “Miracle on the Hudson” A320 did after its engines ingested a flock of geese. In this case, however, there would have been no pilot at the controls to guide the plane in for a smooth landing. What’s more, the power interruption would have turned off the autopilot. Uncontrolled, the plane would have gradually banked into a turn, which then would have grown steeper, devolving into a tight spiral dive that would have ended with the plane impacting the water at high velocity.

Let’s call this the “Unpiloted Fuel Exhaustion Scenario,” or UFES.

Under these circumstances, it would be virtually impossible for the plane to have traveled very far from the 7th ping arc. In a paper released last week, IG member Brian Anderson calculates that by 0:19 the plane would low and have been traveling downward at a tremendous rate:

… the 7th arc position calculation should assume that the aircraft was at or near sea level (the surface of the reference ellipsoid) at time 00:19:38, and hence only a little above sea level at 00:19:29, and descending rapidly.

The UFES, combined with careful analysis of the Inmarsat data, adds up to a clear and falsifiable hypothesis about the location of MH370’s final resting place: it should lie within a few kilometers of the 7th arc, and certainly be within the ATSB search area. The search team has consistently expressed absolute confidence in the hypothesis. Indeed, their language has only gotten stronger with the passage of time. Earlier this month, a Fugro executive told Bloomberg, “We’re absolutely in the right spot — all the analysis has been done. It’s actually getting more exciting as we get closer.”

The fly in the ointment is that the search area described in the area is now almost completely scanned, and the plane is not inside it. When the last square kilometer of the current 60,000 square kilometer search zone is scanned sometime next month, the hypothesis will have been falsified. The UFES will have been shown to be incorrect.

The failure to find the plane in the search area should not be regarded as failure. Rather, it is an important piece of information about the fate of MH370. It allows us to narrow down the list of possibilities going forward.

But it does force us to confront a difficult question: How do we best rationally proceed?

One approach would be to assess the different assumptions that underlie the UFES, judge which ones are most likely  wrong, and then examine the available alternatives. Perhaps, for instance, the 0:19 half-ping was not caused by fuel exhaustion, but some other event. In that case, the plane might have flown on for an unknown period of further time.

Perhaps aircraft performance calculations, which would rule out scenarios like that proposed by Simon Hardy, are incorrect and the plane really did fly far to the southwest.

Another alternative is that the airplane was not unpiloted, but that a conscious individual was actively flying the plane at the time of fuel exhaustion. If that were the case, then the plane could have glided a considerable distance before impacting the sea; the ATSB’s June report states “the aircraft could glide for 100+ NM.”

A fourth alternative is that, as I have suggested, the BFO data upon which the UFES is predicated cannot be trusted, because it was tampered with by sophisticated hijackers. Indeed, we know that the system that generates the BFO data was indeed tampered with, in a way that implies sophisticated knowledge by whoever carried it out. If such a spoof were carried out, then the plane would likely be many thousands of miles from the southern Indian Ocean—most likely, in Kazakhstan. That scenario would explain why no debris has been found in the southern Indian Ocean, and provide a link to what would otherwise be an extremely unlikely coincidence: the Russian shoot-down of a second Malaysian Airlines 777 just four months later.

The ATSB has as yet given no indication that they realize that the the UFES is incorrect, or what kind of scenario would be compatible with the new search area. They have specifically stated that they view the Kazakhstan scenario as impossible, but have not revealed any evidence that would rule out a spoof.

The only insight we have into the ATSB’s actual line of reasoning is by looking at the new search area that they have laid out. Essentially what they’ve done is to take the current search area and make it bigger in every direction. There doesn’t seem to be any operational principle besides “let’s keep looking.”

The ATSB’s proposed new search area might simply be an example of what psychologists call “perseveration.” This is a pattern of behavior often seen in individuals who are highly stressed to the point of panic, and, unable to come up with a solution to their problem, simply continue to repeat the same unsuccessful actions over and over again. (An example I cite in my book Extreme Fear is the case of Civil War soldiers who, after suffering a misfire in their muskets, simply continue to cram more and more cartridges down their rifle barrels; after one battle guns were found stuffed with 10 or 11 cartridges.)

Perhaps the time has come for the taxpayers funding the SIO search to demand a more rational approach.

UPDATE: Soon after I put up this post, one reader commented via Twitter that continuing to search the southern Indian Ocean at least provided jobs and did no actual harm. I disagree. Apart from spending taxpayers’ money unnecessarily, it serves to put the mystery of MH370 on ice–allowing the authorities can say “we’re doing the best we can” while running out the clock on the public’s interest in the case. They’re buying the right to pretend, for one more year, that their assurances haven’t been empty and that the UFES is still a reasonable (indeed, the only reasonable) possibility. This means that more profitable lines of inquiry will continue to be ignored.

167 thoughts on “The Reasoning Behind the Continued Search of the Southern Indian Ocean for MH370”

  1. @Richard

    My belief is that the 200nm and 295 radial were directly derived from the radar data, and someone stuck it in the presentation without reconciling it with the waypoint overlay. The rest of the stuff on that presentation is an overlay created by someone who did not do a very good job. (Waypoints are not shown on a military radar screen. In fact, the screen shot shown is not even from the radar at Butterworth.) The waypoints were added. So what we have is the IG and ATSB drawing conclusions from an overlay on an overlay.

  2. Brock: It seems to me that you are [deliberately ?] misconstruing my points and that is clouding the issue, and perhaps your judgement.

    (i) I mentioned altitude only in relation to the possibility that a greater altitude at 00:11 might help resolve a potential discrepancy between BFO determined ROC, and observed behaviour in the Sim.
    (ii) I have not ascribed point estimates to altitude, speed or bearing at the 7th arc.
    (iii) I understand the Sim trials. I have viewed the videos many times and of cause had may discussions with ALSM.
    (iv) The 3 min 40 sec figure is important. It was stated by ATSB, and observations in the Sim trails confirm this.
    (v) It is simple speed and distance calculations to arrive at my conclusions. Speeds are constrained by what is plausible. At 00:11, 480 knots GS is over 500KTAS, and greater than any path model based on BTO and BFO that I know of. 429 knots GS is the lowest BTO and BFO speed that I know of. Check yourself and see which speeds are capable of reaching the 7th arc.
    (vi) The distance to the 7th arc is less if the path is a turn to the left immediately the 2nd engine fails. The Sim demonstrated turns every time.

    Chris: No. That can’t be implied. I am concerned only with simple speed and distance calculations from 00:11, and relating these to the observed Sim behaviour.

    This my last word. I don’t have the time nor inclination to continue to argue the simple analysis in my paper. If you have a contrary view please feel free to present your analysis in your own paper.

  3. The distance between the 6th and 7th arcs is 37 NMi, so the average groundspeed in those 8.5 minutes must be greater than 280 kts.

  4. Brian,

    I really appreciate the time and effort you’re taking to respond – and I certainly meant to portray your work as accurately as possible – apologies if I haven’t. I think the key misunderstanding is on what I meant by “point estimates”. Please permit me to clarify:

    I struggle with two aspects of this paper:

    1) Its conclusion rests on calculations which treat key inputs:

    – distance between arcs 6 and 7(FL0) = 56 nmi
    – deceleration between them = -0.315 KTAS/s
    – altitude at arc 7 = “very near zero”

    …as constant (i.e. point estimates) – when as you know (and state explicitly, in most cases), each is a variable, with a DISTRIBUTION of possible values. Basing a conclusion (e.g. “very likely…inside the 7th arc”) on the point-estimate outputs of such a model is dangerous. For example: if the straight-flight distance between the two arcs is 58 nmi instead of 56 (quite plausible, per BTO error data sent to me by ALSM), then left turns of much larger radius than your 10nmi cut-off become feasible. Yet whole FAMILIES of such scenarios seem to have been cut out of the picture prior to drawing your conclusion.

    2) Its key conclusion – “INSIDE” arc7 – seems COUNTER-indicated by the rest of the report:

    a) if, at arc 7, MH370 was “only a little above sea level at 00:19:29, and descending rapidly” as you observe, then impact MUST have been a very few seconds later

    b) if arc 7 is as hard to reach in the available time as you observe, then it follows by simple deductive reasoning that intersection was more likely to have occurred while MH370 was OUTbound, not looping back IN after a full 270-degree (or more) spiral.

    c) A “more likely to be OUTSIDE arc7” conclusion requires simply combining a) and b).

    Brian, I respect your time, appreciate the insights you’ve offered us, and am not being spurious (I’ve encountered several characters over the past year who ARE, so I don’t mind it if you took me to be one of them). I just want to make sure any analysis guiding this search is as robust as it can possibly be.

  5. This is all I could find on B777 fuel flow measurement:

    “The dual-rotor flow meter design was inspired by Boeing’s need to measure fuel consumption more accurately, during the development of the Boeing 777 aircraft. Great emphasis was placed on the accuracy of fuel consumption of the GE 90 engines, which powered the 777. By using the dual-rotor turbine flowmeter, Boeing was able to improve the accuracy to ±0.25% on a mass flow basis and determine the horsepower of the engines. Improved measurable fuel economy allows for longer international flights and guarantees the fuel safety margins required.”

    (ref. http://www.exactflow.com)

    Comment: This is about flight test instrumention to check the accuracy of the standard fuel flow measuring system. Not that the observed difference in INDICATED fuel flows is in the “Deferred Defects Log” of 9M-MRO. The first step in a “defect investigation” would be to check the calibration of the fuel flow transducers. In the field there are three ways to accomplish this:

    – monitor the fuel quantities in each tank before and after several flights, checking that no fuel cross-feed was done; or
    – cross-exchange the left and right fuel flow transducers; or
    – remove both transducers and send them to a specialized lab for re-calibration.

  6. To followers of Mike Chillit’s map tweets, as an FYI:

    (mi = “statute miles”)

    Mike is suggesting Fugro Discovery is now searching NW of the area previously bathy scanned. My sense is that this is not true.

    Back before I was banned by Mike for “trolling”, we had a profitable DM exchange in which the map below (see link) featured. I believe I managed to convince him that a) his white line marking arc 7 was roughly 3mi inside the precise arc7 (FL0), and more like 9mi inside the precise arc7 (FL350).

    I ALSO believe I convinced him that 2 or 3 lines = 6-9mi of bathy scanning along the NW edge had at some point had dropped off his graphic.

    I believe Mike at one point corrected the former, but I don’t think he ever corrected the latter. As such, I believe Disco is still searching “charted waters” (though it is surely very near the edge, now).

    I also believe my sense of things is supported by the detailed maps Richard Cole has tweeted.

    Here’s my map, which simply aligns 2014 & 2015 versions of Mike C.’s maps:

    https://drive.google.com/file/d/0B-r3yuaF2p72UUJfbjVOcEdTdUU/view?usp=sharing

  7. @Richard Cole, posted May 3, 2015 at 6:08AM
    @DennisW, posted May 3, 2015 at 7:21PM

    Thank you for both your posts as itemised above.

    I have spent some time examining the Lido Hotel radar images, and plotting on Google Earth etc. I have come to the conclusion that the furthermost (upper-left-hand) label “295R 200nm from Butterworth AB” should actually read “275R 200nm from Langkawi”; the last radar point (if 10nm beyond MEKAR is to be believed) is actually on a bearing ~273.5 from Langkawi, at a distance of 204nm. In that sense, I agree with Mr Cole, insofar as this label is thus (simply) typo-ridden in the heat of the moment, and that thus the “10nm beyond MEKAR” final radar point is most likely still correct.

    The bottom-left-hand label “PERAK ISLAND 279R 89nm from Butterworth AB” is broadly correct by my amateur judgement; I get a bearing of 283° at 88nm.

    If the later radar is from Langkawi, would that not suggest that the radar source is in fact as much commercial, as military?

    DennisW, I’m afraid I am struggling to understand your comment “That location is well inside the 18:25 ping ring which causes me to doubt it.” Surely the “10nm beyond MEKAR” location is about 35nm short (along airway N571) of ping ring 1828, and thus OUTSIDE the ping ring? I am using Mike Exner’s ping rings (from https://www.dropbox.com/s/2olyztvzo5aj043/ALL_ARCs_2015-04-06.zip?dl=0, dated 19 April 2015). I can upload a graphic to illustrate if you like.

    In all other respects, that is to say, the radar data from IGARI to Pulau Pinang, I agree with your comment “The radar data has some serious problems and I have no idea what they are.” particularly in the light of @oriondt’s analysis from 30 April 2015.

    Thanks in advance, gents.

  8. @BT-77

    I am using an arc radius of 3630km for the 18:25 ring. What is the radius you are using?

    Best,
    Dennis

  9. Jeff – I’m a bit disappointed with some of the expert responses to this post which was as timely as it was pertinent. Noone complained when the initial 60,000km2 zone was set. Indeed the crunchers were mainly standing by waiting for their moments of vindication, so why would the analysis simply be “flawed at the margins” when it could be very wrong. But if it’s not in this area and it’s in the SIO, then it didn’t spiral in? Would that be an unreasonable conclusion? If it didn’t spiral in then we know the “scaffold of assumptions” is creaky?

  10. @DennisW

    Thanks for coming back. Mike Exner’s rings give me a radius (more specifically the distance from I-3 F1 sub-satellite position to 1828UTC ring at 6°50’21.50″N, 95°48’25.70″E) at 3558km (or 1921nm). Clearly this is the reason for our different views! 🙂 What source did you use for your rings? Cheers – Bill

  11. Dennis, BT-77,

    I don’t know where Mike got 6°50’21.50″N, 95°48’25.70″E, again with the ridiculous accuracy, but I will not be surprised if it is based on the “assumed fact” MH370 flew to IGOGU, or along N571.

    My recent overlay of “Lido snapshot”, waypoints and Bobby’s ‘hook’ [Posted May 2, 2015 at 7:12 PM] show that extrapolation position of MH370 would be approximately 10 km south of NILAM, which corresponds to approximately 6°40’N latitude. My estimation of the respective ping rings are consistent with the groundspeed of approximately 500 knots between 18:22 and 18:25. The discrete nature of BTOs, JS was actively discussing a while ago, is clearly visible.

  12. @BT-77

    Your question deserves an answer. I calculated the rings about a century ago, it seems, and stuck the values in a table. I checked with Steel et. al. at the time, and my values were very close. BTW, I was assuming a spherical earth and a geostationary satellite position (both assumptions introduce errors). It has never been an issue until now. I suppose I will have to go back now and use the actual satellite position and aircraft approximate latitude to get a proper estimate. My bad. I should have done that before concluding anything relative to this discussion. Sometimes I feel like I am going backwards!

  13. @Oleksandr

    Thanks for your post May 5, 2015 at 8:47AM. To answer your points:
    1. You’re quite right, the loc I gave does have ridiculous accuracy. I’m sorry, I was just reading off what Google Earth told me.
    2. Yes, that point is the intersection of Mike Exner’s 1828 ring and N571. Do you not think it reasonable, given that (according to the ‘Lido’ image) MH370 went from VAMPI to MEKAR, it might continue (some way) along N571?
    3. Your extrapolation position is not far from my interpretation of the last known position according to the ‘Lido’ image, ie. some 40km apart.
    4. I confess I am a little puzzled by two of the green rings on your ‘combined(lido_wps_npp).png’ image. The inner ring labelled 18:28:14 I concur with, it’s the same as Mike Exner’s that I am using. But of the other two; 18:27:03 is OUTSIDE 18:25:27. Am I missing something? Is it to do with the BTO analysis you mention? If so, fair enough; that’s way above my level of expertise!

    Thanks Oleksandr.

  14. BT-77,

    Thanks for your comments. My comments/answers are below.

    1. Earlier both Dennis and I has some issues with regard to the precision of the impact location pointed by some members of IG (indeed the plane was not found there), so both of us are oversensitive to this topic.

    2. It might continue along N571, but it might not. It is the working hypothesis of IG, which is being now presented as a fact. This generates a loop of self-supported assumptions (e.g. AP, N571, constant altitude, cruise speed), which are consistent with each other.

    3. Yes, but the important issue is the ‘hook’, the beginning of which appears to be close to the extrapolated position at 18:25. Given that the image was taken 30 minutes later, the beginning of the ‘hook’ may be in the same location as MH370 at 18:25 due to the wind.

    4. Yes, I am aware of this issue. I included BTO numbers in parenthesis, which were indicated in the “Update to Signaling Units Logs”:

    18:25:27.421 BTO=17120-4600=12520.
    18:27:03.905 BTO=12560.

    A long time ago JS noted discrete nature of BTO, with 20 microseconds rounding errors. Assuming the rounding is ‘standard’: 17120 can be anything between 17110 and 17130; the correction 4600 can be anything from 4580 to 4620 (because it is the difference between two samples on the reboot used for calibration); 12560 can represent anything from 12550 to 12570. Hence, the maximum 18:25:27.421 BTO can be 17130-4580 = 12550 microseconds, which is the same as the minimum 18:27:03.905 BTO. Note that 18:27:04.405 BTO is 12520 (not shown in my plot).

    In summary, I think it is simply an artifact of discrete arithmetic; the aircraft was somewhere in between these two arcs. This example demonstrates accuracy of the BTO data.

  15. @Oleksandr

    1. It reminds me of a tongue-in-cheek comment we used to make at work – ‘accountants like to be exactly wrong, engineers prefer to be roughly right!’ Perhaps I’m becoming an accountant :-/
    2. I quite agree, it might not have continued along N571. We need to keep an open mind. Your contrail and ‘smog’ analysis, along with Dr Bobby Ulich and Kirill Prostyakov et al is most illuminating in this regard. Keep it up.
    My only comment re the ‘Lido’ image is that I believe its inbuilt inconsistencies may be explained by some simple typographic errors and mislabelling. In this way, it would therefore become a more concrete piece of evidence. I can present a report on this if you think it would be useful.
    3. See 2 (1st paragraph) above.
    4. Good, thanks for your further explanation! I understand the broad principles involved, but do not pretend to be a BTO/BFO expert.

  16. BT-77,

    Thanks. Sure your thoughts on “Lido image” will be appreciated (well, may be heavily criticized).

    Just keep in mind that earlier [February 1, 2015 at 4:26 PM] Don shared radar coverage map:

    i.imgur.com/TBYllAA.jpg
    (add http in front)

    The ‘blips’ shown in the “Lido image” likely came from at least two radars (there are several reasons to think so).

  17. @Oleksandr

    Thanks for Don’s radar coverage map link – I was just looking for that.

    “well, may be heavily criticized” – Heh! Fair enough. I’ll put something together over the next couple of days.

  18. @BT-77

    I thought I responded earlier, but I must have gummed something up.

    I calculated the ring values what seems like a century ago. They were in good agreement with Steel et. al. So I just stuck them in a table. Never been an issue until now. I used the nominal equatorial position of the satellite and a spherical earth. I’ll go back and use the actual satellite position, and the WGS-84 ellipsioid. Stay tuned. My bad. I should have done this before drawing conclusions relative to a close call.

  19. @ Brian Anderson:

    In your paper “The last 15 minutes of flight of MH370” you write in the “Introduction”:

    “The aircraft was flying on autopilot during its passage south over the Indian Ocean until the second engine failure”

    Then you write under “Fuel Data”:

    “… this difference (in fuel consumption between engines) is 0.8 per cent. Using this figure it is possible to deduce that the Right engine failed close to, and perhaps a few seconds before, 00:11 UTC on 2014 March 08.
    (…)

    Then you write under “Flight path after the first engine failure”:

    “The simulator trials mentioned earlier illustrated clearly that following the first engine failure the auto-throttle increased thrust in the remaining engine, altitude was maintained, and the (longitudinal) pitch increased as the speed reduced so as to provide adequate lift. The deceleration observed was noted, and the Indicated Air Speeds (IAS) were converted to True Air Speeds in Knots (KTAS) for illustration in Figure 1, below. The observed trend is linear, with a deceleration of approximately 19 knots per minute. (dy/dx = -0.315 ´ 60 = -18.9). A deduction from this result is that the aircraft would still have been well above the best single engine speed and would have continued to fly at the same altitude for some minutes after the engine failure.”

    “Turning back to what we know about the actual flight, the BFO at 00:11 suggests that at that instant the aircraft was descending at about 250 feet per minute. Together then, the BFO descent rate, the estimated time of the Right engine failure, and the simulator trials, may all be reconciled if the altitude was greater than FL350, in which case it is possible that a shallow descent commenced just prior to 00:11.”

    This last quote seems to be inconsistent with the others. If the airplane was on autopilot, and the first engine failure occurred close to, and perhaps a few seconds before, 00:11 UTC, how could the aircraft be descending at 00:11 UTC?

    After the first engine failure the aircraft on autopilot would either be decelerating at constant altitude, or it would descend at constant airspeed. The rate of descent in the latter case is related to the rate of deceleration in the former. For example, for the observed rate of deceleration of 19 kTAS/min, the corresponding rate of descent is about 620 ft/min (depending on TAS). The transition between the two conditions takes about 1 second. You do not state the weight, altitude and temperature when the 19 kTAS/min was observed, presumably 17,400 kg, FL350, ISA temperature?

    If the 1st engine failure occurred at 00:11, then at the time of the second engine failure 229 seconds later according to your figure 1 the speed would have been 387 kTAS (224.6 kCAS/M.672 at FL350 ISA), still above the single-engine drift-down speed of 219 kCAS, so a descent starting before the 2nd engine failure would not be expected.

  20. @ Brian Anderson:

    P.S.
    If the altitude was greater than FL350, the rate of deceleration would have been greater than 19 kTAS/min, and it is possible that a descent commenced prior to the second engine failure at 00:19:29 UTC.

  21. @Brian Anderson:

    You also wrote in the Introduction of your paper:

    “We know that each B777 aircraft is subtly different (as are most other aircraft) with respect to rudder trim in particular. One might say that they may be slightly ‘bent’ in terms of aerodynamics. ”

    That may be true in general. However, the production tolerances of a B777 are extremely tight, and do not explain the unusual behavior observed in Exner’s simulations. That behavior is different from other simulations and remains unexplained. A flight simulator uses an aerodynamic database which, if anything, represents somewhat idealized aerodynamics of the airplane type. It would certainly not represent an extremely ‘bent’ airframe.

    How was the simulator set up initially? What were the autopilot/autothrust modes? How were the 1st and 2nd engine failures simulated? How was the simulator reset after each test? Since dual engine failure causes flight control reversion to secondary mode, was normal mode restored prior to each test? Are you sure the airplane did not stall before it went into a spiral dive?

    Lots of unanswered questions.

  22. @BT-77

    I did revisit the 18:25:27 ring. The value you are using, 3558km, is very close to what I calculated today (3528 km) using a greater standard of care than my initial “casual” number. It does make a substantial difference relative to the radar data at 18:22. I am humbled.

    Link to 3528km calculation (“Where is that Ring”) below.

    http://mh370corner.blogspot.com

  23. @DennisW

    On the contrary, it is me who is humbled, sir. You have recalculated your ping rings in less than a day. I don’t think I could calculate my way out of a paper bag….

    Flippancy aside, I am glad your rings are at least somewhat closer, and doubtless more accurate. Thanks for posting to your blog. I couldn’t comment on whether yours or Mike Exner’s are better; I only wish I could.

    I have taken a screengrab of my version of Google Earth, showing the FI.pdf 1822 position on airway N571, and Oleksandr’s position at the same timestamp. See https://www.dropbox.com/s/qedd5pq45jfm1xl/2015-05-06_Google_Earth_screengrab_BT-77.pdf?dl=0

    I should add that I am using a generic static subsat pos of 0°S, 64°30’E for the measurements on my screengrab. Please correct me anyone, if this is wrong. (I realise the satellite “wobbles” slightly in reality.)

  24. Kevil – I am surprised that more people are not too concerned with the extremely high speeds on Figure 1.1F of the FI Report. Note, however, the speeds in the Military Radar Report on page1.1.3.a are not consistent with the speeds around the same time as those in Figure 1.1F. Also, no one seems to be concerned about the different altitudes listed in 1.1.3.a. I guess the feeling is radar cannot provide accurate altitudes.

    Oleksandr – With an American aircraft manufacturer coupled to a British manufacturer of the engines and and a peanut gallery from all over the world, the units of measurements used here is a mixture of Metric and American units. (Even the Boeing manual uses a mixture of units in a single table.) However, most contributors to this blog use Pressure Altitude in feet (e.g. 38,000) or Flight Level in hundreds of feet (e.g. FL380) to express how high up a plane is flying. Please join us. BTW, FYI, depending on conditions (e.g. long runway/certain temperature/humidity), full throttle on a B-777 is not always used during Take-Off.

    Brian – A year ago I calculated that the true radius of a Ping Ring was dependent on the altitude of the plane. I was happy to see your calculated values matched my own.

    In a post around April 04, 2015, I said my calculations showed the flameout of the right engine occurred at about 00:06, about 9 minutes before the flameout of the left engine. Almost all of the difference between our respective calculations is I used a constant 1.5% higher fuel consumption of the Right Engine when compared to the Left Engine. You used 0.8%.
    I have revised my calculations using 0.8% and came up with 0.05329 h (or 3min 12sec). (LRC,AT & AT) Could you review my methods and assumptions to see if this makes sense?
    X=flight time of both engines
    Y=flight time of the Left Engine alone after Right Engine flameout
    X+Y=7.75h (from 16:30UTC to 00:15UTC)
    Right Fuel Tank=24,800Kg
    Left Fuel Tank=24,900Kg
    A=Right Engine Average Fuel Flow in Kg/Hr = 24,800/X
    B=Left Engine Average Fuel Flow in Kg/Hr = 24,900/(X+Y)
    Here is where I think we differ: After flameout of the right engine, the Auto Throttle increases the thrust in the Left Engine. This increases the flow rate to the left engine. For this flow, I used 5573 Kg/Hr from the table on page PI.23.8 LRC for a 180,000 Kg plane at 27,000 Ft. (You can vary this number as you like. Note, back in late December, VictorI calculated 5.46 mton/hr total flow for both engines at 00:15 but that was before it was suggested that the Right Engine flamed out before the Left Engine.)

    AX/BX= 1.008; and:
    BX+5573Y=24,900; and AX=24,800; and
    X+Y=7.75; solving for Y, I get 0.05329h
    Is my math correct?

    Victor – Using your track offset puts MH370 even closer to Kate Tee’s sighting (based on the her boat’s GPS log being off by exactly one hour).

  25. If it’s of any comfort I also get 3528km for the 18:25:27UT ring, from a completely separate calculation path.

  26. @Richard Cole, DennisW

    Thanks for your corroborative post Richard.

    Now that I’ve eaten some humble pie and moved my subsat point to ~1.6°N 64.53°E (as per VictorI twitter post 2Feb2015 for 1822UTC), I get 3530km (1906nm) off Google Earth/Mike Exner 1825:27ring.

    Thanks all.

  27. Gysbreght:

    As I said a couple of days ago, I do not have the time, nor inclination to debate every point ad nauseum, especially when arguments are developed from points taken out of context and distorted. Please feel free to publish you own paper. Where did I mention an extremely bent aeroplane? Mike’s sims were consistent, and at no time did the simulated aircraft stall.

    Lauren H: I used the published numbers to calculate the fuel remaining at 17:07, i.e.L 21969kg, R 21831kg so there is probably a difference there. Then I used the difference in flow estimated at 0.8%, and estimated the increase in flow after the 1st engine failed at 3680 kg/hr, or approx 20% increase.

    Others in the IG have done a lot more detailed work on fuel endurance etc. All I was looking to establish, and confirm with ALSM, was that the 1st engine failed shortly before 00:11.

  28. @Oleksandr and BT-77,

    The rounded BTO values make me think that we can’t really establish motion in the 18:25-18:28 time frame. The ring changes could be either the result of rounding or changes in altitude. Theoretically, the plane could have been going a different direction.

    My feeling then, and now, was that the rounding was caused by the sensor, not the computer. I came to this conclusion because I could find no rational reason for the software to truncate the raw numeric values.

  29. @JS

    Thanks for your additional comments. I agree that we cannot establish motion 18:25-18:28.

    Do you not think the rounding/truncation might have happened upon publication? Either purposefully, or a “let’s make this column narrower” stylee…. (Apologies if this topic has been covered already.)

  30. ATSB confirmed months ago in a private email that the BTO values are truncated, not rounded. I have reported that here and elsewhere many times.

    It should be noted that the distinction between truncation and rounding is not important because the calibrated BTO values are the same in both cases, after the bias calibration process.

    BTO values are recorded with 20 usec resolution because that is what the Channel Unit engineers deemed to be sufficient, given the inherent system noise level. That design decision is regrettable in retrospect. I prefer to digitize observations to at least 10 dB into the noise. That was not done here. But nothing was rounded off or truncated after the fact. It was a design decision taken years before MH370.

  31. Gysbreght:

    Regarding your list of questions about the simulator trials…

    Your confusion seems to be rooted in the assumption that what we observed in the simulator was unusual. The opposite is true. A spiral descent (or “dive”) is normal and what we should expect. Virtually all airplanes end up doing this with no control inputs to correct the tendency to start a turn. This is what we observed in the simulator, and I have confirmed it in 2 of my own airplanes. The direction of the turn can be dependent on how the plane was trimmed before fuel exhaustion. But even if the plane has no manufacturing imperfections and no trim adjustments, it only takes a small external force from atmospheric turbulence to nudge the plane one way or the other. Any weight imbalance (like cargo positions, fuel, etc.) can also influence the direction. The weight of the wings (dry) is also a factor. They are never exactly the same. Once the turn starts, the natural tendency is for the turn to increase in bank angle over time. This is what we observed in all our simulations. According to ATSB (private correspondence), it is also what Boeing observed in their simulations.

    The simulations do not prove anything, but they do provide a better understanding of the few MH370 observations we have. As Brian notes, with engine flameouts at ~0011 and ~0015, it is hard, if not impossible, to make it from the 6th to the 7th arc in a straight line. Therefore, a turn must have developed. The geometry precludes a right turn, so a left turn must have developed after fuel exhaustion and loss of (most) electrical power.

    One final note. The B777 is so smart, it tries to restart the engines if they flameout. It does this automatically with no human involvement. In one case, we observed a brief successful restart of one engine. It only lasted about 3 seconds. But that asymmetric burst of thrust was enough to increase the bank angle to nearly 90 degrees. We do not know if MH370 had a brief engine restart. But if it did, the spiral probably ended in impact very quickly.

  32. @ Brian Anderson:

    Can you please answer one more question, and then I’ll shut up.

    What were the conditions of weight, altitude and temperature at which the deceleration of 19 kt/min was recorded?

    Thanks in advance.

  33. @ALSM: If achievement of arc 7 is so difficult as to require a left turn, then it follows – by strict application of the rules of geometry – that this turn must have been GRADUAL.

    Because any path curvature significant enough to bring the flight path at point of intersection anywhere NEAR tangential to arc 7 (heading back NE) has a LONGER flight path (in terms of total arc length) than would a straight path.

    Because (math).

    Please explain how MH370 could have intersected arc 7 – at “very near zero altitude”, per Brian’s paper – and at this SLIGHT cumulative leftward heading change, per mathematical REQUIREMENTS of your “had to turn left” claim – and still have managed to turn around and head back NW enough to impact MORE than the 11 nmi inside arc 7 already searched.

  34. @ALSM

    While on the subject of BTO, do you understand the source of the bias term (Table 2 p55 of the August ATSB report)? Not sure why it is called bias since it has substantial variability. Basically my question is what is the underlying causality, and how was it determined on the ground in KL?

    Also it would seem that using the average bias in subsequent calculations would result in potentially much larger errors than the 20us quantization.

    Thx in advance.

    BTW, after correcting my erroneous ping ring radius I can find no other “solution” to the radar data than what you have concluded. I still have some troubling issues there, but they are probably not important to tidy up.

  35. @ airlandseaman:

    Thanks for replying, I appreciate that very much.

    You wrote: “A spiral descent (or “dive”) is normal and what we should expect.”

    You are touching upon the crux of the matter. I am addressing the difference between the ATSB descriptions of “eventually a spiral descent wouls develop” and “the airplane entering a descending spiralling low bank angle left turn” and Brian Anderson’s “Observations from the simulator trials suggests that following the failure of both engines the aircraft will bank into a turn almost immediately.” and ” … the bank angle will continue to increase until the aircraft enters a spiral dive. At that point the bank angle may have increased to 90 degrees.”

    “As Brian notes, with engine flameouts at ~0011 and ~0015, it is hard, if not impossible, to make it from the 6th to the 7th arc in a straight line. Therefore, a turn must have developed.”

    Yes, that is true for the assumptions used, i.e. a fuel flow asymmetry of 0.8% and the IG’s trajectory.

    One final note. The B777 is so smart, it tries to restart the engines if they flameout. It does this automatically with no human involvement. In one case, we observed a brief successful restart of one engine.

    Yes, I agree that the B777 is very smart. The Auto-restart is decribed in the FCOM. However, it is not evident that the engine auto-restarts on an empty tank, and the violent ‘yerk’ you describe is impossible with a real engine that needs to spool-up after re-light.

  36. @ALSM

    Duh! Forget on how bias is measured. I can see that it was simply computed to make the numbers work. Question on source of bias remains.

  37. DennisW

    The bias is determined at the time of the calibration in acordance with the equation I published back in May 2014.

    https://www.dropbox.com/s/e2ij54k8voxdej8/Deriving%20Net%20L%20band%20Propagation%20Delay.pdf?dl=0

    The bias is an artifact of the way the Perth GES Channel Units (mod/demod) measure the time delay between a transmitted pulse and a response from the AES. It is a constant embedded in the hardware (CUs and AES). And yes…we do see jitter on the measurement as described in the Dec 23 report. Thus, individual measurements can have +/- 40 usec error on top of the bias error (bias error is much smaller because it is avergaed over many observations).

  38. Proof of my prior claim, under the assumption of constant turn radius:

    Angle between assumed straight flight path & arc 7 at point of their intersection: 48°

    Distance from assumed flameout at 00:16 to arc7 (FL(0)): 21 nmi

    (both are taken directly from Brian’s paper)

    radius of circle tangential to both flight path and arc 7: 21 * tan(48°/2) = 9.35 nmi

    circumference of circle between points of flameout and intersection:

    2 * pi * r * (180° – 48°) / 360° = 21.5404 nmi,

    …or 2.6% LONGER than the straight line distance.

    If we need to save distance vs. straight line flight, we need a much more gradual turn – which puts MH370 heading SE at arc 7 (FL0) intersection.

    The only way to make ALSM’s latest post-flameout “concept path” to work is as follows:

    a) turn left sharply (toward arc 7)
    b) straighten out (to save distance)
    c) shed altitude rapidly (10K ft/min)
    d) turn sharply left again (to asymptote to arc 7)
    e) intersect with arc 7 at very near sea level
    f) continue sharp turn – but suddenly stop rapidly shedding altitude
    g) straighten out again (to prevent spiralling in within searched area)
    h) fly an ADDITIONAL 12+ nmi NW (to escape the searched area)
    i) impact outside searched area

    The absurdity of this path is self-evident.

    If the IG has interpreted the signal data correctly, then at least SOME aspect of that data is invalid.

    PERIOD.

  39. Brock: Once again you are taking selected pieces of information out of context in order to prove whatever you are trying to prove. As I said before, I’m over it.

    It is you that has constructed the “absurd” turn scenario. Close observation of the sim videos shows that such behaviour is indeed possible.

    Gysbreght: No.

    This is my absolute last contribution to this blog. I’ll not even bother to open it again so there is no use responding. There are too many people here hell bent on proving everyone else wrong, maybe to satisfy their own egos, but not contributing to the prime cause. I say to these people, if you have something to say, then publish your own paper.

  40. @ALSM – if the bias was embedded as a constant in the hardware, why was it necessary to calculate it using averages? This part never made sense to me. Or is the issue that a bias of 500 was embedded, but it had a tolerance of +/- 5, and the actual constant was never known?

  41. @Brian: it appears you are not aware of how ALSM is USING your paper’s conclusions to construct a “concept path” ending some 14nmi inside arc 7 @ FL(0). It is HIS path I am disproving – not yours.

    There is no need for umbrage – we are ALL trying to help direct the search as best we can.

  42. @ALSM

    Thx, but I don’t need a tutorial or a history lesson. Believe it or not people do not read every post you make. Try being a human being once in awhile. It will work wonders.

  43. Brock – It’s getting desperate. They have calculated their way down to a search area and now they have to calculate their way out of it. The race is still on. It’s not just the relatives with suspended lives.

  44. @Matty et al – I find it interesting that the search area has now excluded not only the area covered, but it has also mathematically excluded the entire theory of the last few minutes. I mean, it almost sounds like it isn’t there! Almost, because they are more confident than ever that the last grain of sand in the search area is hiding a starfish, a beer can, and of course a B777.

    I want to say that maybe it’s just a little to the left or right of the search area. But this was the area defined by so much calculation, by experts. I’m just not convinced that so many experts could ALL be “sorta right.” I don’t think there are that many ways to be wrong if the BTO/BFO data is right. Everybody is within a few miles of each other and yet all are wrong.

    That can only mean the data is wrong. I don’t think they blew the calculations. I think we’re back to a software bug in the log or a spoof.

  45. JS – Ditto. Before they set off I opined somewhere that if it doesn’t come off we’re in a situation of not knowing how wrong it is. In my own uneducated way I’ve sat here and sprouted a whole lot of what-ifs because the confidence level was out of this world. Have the what-ifs come home to roost? How could all the gurus be surprised and heaps of laymen not one bit?

    Been an interesting one to watch. Some have an impenetrable conviction in their rightness that will be upheld right to the end – whatever that looks like. Others have shifted focus. Either way the expanded search will cover a lot of the off-shoot scenarios so it’s still a case of oceanic pin the tail on the donkey. People will stake their credibility how they see fit. They can thank the Chinese for a 2nd towing season because I’m sure the Australians and Malaysians wanted to wriggle out. They aren’t paying and it’s great PR back home. The benevolent republic seeks it’s lost sheep…..

  46. Some posters here seem to forget that there has always been a wide search area. It contains the locations that the airplane can have reached within the known data and performance limitations.

    Since it is impractical so search that very large area, the ATSB’s search strategy team has defined a smaller “priority search area” where the airplane location was deemed to be most likely, based on certain assumptions, which have been clearly stated. If the airplane is not in the priority search area, the assumptions on which that area was based need to be reconsidered.

  47. Dr. Bobby Ulich’s recommended location has never been searched by air, surface or subsea methods.

    ~LG~

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.