A Comprehensive Survey of Possible MH370 Paths

This is the second in a series of articles that is dedicated to defining a new area for the underwater search of MH370. In the previous article, we presented Bobby Ulich’s overview of a new statistical criteria that supplements the criteria that many investigators have used in the past for evaluating candidate flight paths. In the present article, we present an overview of an exhaustive study principally undertaken by Richard Godfrey, with contributions by Bobby Ulich and me, to examine flight paths with the assumption that after 19:41, the flight was automated and with no pilot inputs. The results indicate that a flight crossing the 7th arc near 34.4S latitude merits a deeper investigation, which will be the subject of the next (third) paper.

What follows are excerpts from “Blowin’ in the Wind”, by Richard Godfrey et al. For more details, please consult the full paper.

Introduction

Following on from Richard Godfrey’s earlier paper entitled “How to play Russian Roulette and Win” published on 13th February 2019, which covered the first part of the flight and diversion of MH370 into the Straits of Malacca, Richard was contacted by Bobby Ulich, who asked the question “where do we go from here?” Richard Godfrey, Bobby Ulich and Victor Iannello came up with the idea to scan the Southern Indian Ocean (SIO) for possible flight paths of MH370 using a degree of precision that we believe has not been previously applied, and to use certain statistical checks on the presence or absence of correlations in the data. Each of us had independently developed a MH370 flight model using the Boeing 777-200ER aircraft performance data, Rolls Royce Trent 892 fuel range and endurance data, Inmarsat satellite data and the GDAS weather data. The goal was to find all possible MH370 flight routes that fit the data within appropriate tolerances. Additionally, the data would be checked using a set of correlations.

Method

Our assumptions about the automated flight after 19:41 are that there are 7 parameters that determine a possible MH370 flight path:

  1. Start Time
  2. Start Latitude
  3. Start Longitude
  4. Flight Level
  5. Lateral Navigation Method
  6. Initial Bearing
  7. Speed Control Mode

If you draw an arbitrary line of latitude between the area of the last known point and the SIO, MH370 must have crossed this line at a certain time, longitude, flight level and initial bearing using a particular lateral navigation method and speed control mode.

Having fixed the start latitude, the start time and start longitude can be varied for any given flight level, lateral navigation mode, initial bearing and speed control mode, and the fit to the aircraft performance data, satellite data and weather data ascertained. The flight model used in the wide area scan was developed by Richard Godfrey. First the altitude and air pressure at the selected flight level is determined. The GDAS weather data provides the actual surface air pressure and surface air temperature for a given position and time by interpolation. The air pressure for a given flight level is calculated based on the ISA standard surface pressure of 1013.25 hPa and standard surface temperature of 15.0°C. The geometric altitude for a given flight level is then approximated using the actual surface pressure and actual surface temperature. The altitude is used in the satellite data calculations, assuming the flight level is maintained between 19:41:03 UTC and 00:11:00 UTC. Similarly, the GDAS weather data is interpolated for the exact latitude, longitude and time to find the Outside Air Temperature (OAT) and wind at the given flight level.

The scan method for each Lateral Navigation Method (LNAV, CTT, CTH, CMT and CMH) and for each Speed Control Mode (Constant Mach and Long Range Cruise) requires stepping through each possible Initial Bearing (initially from 155°T to 195°T) in steps of 1°T. In Constant Mach (CM) the value was set initially at 0.85 and decremented in steps of 0.01 Mach.

[Note: LNAV = Lateral navigation (following waypoints connected by geodesics); CTT = Constant true track; CTH = Constant true heading; CMT = Constant magnetic track; CMH = Constant magnetic heading]

Once the Initial Bearing and (if relevant) the Mach has been set, the Start Time or Start Latitude is adjusted to minimise the RMS BTO Residual (BTOR) across the 5 satellite handshake points between 19:41:03 UTC and 00:11:00 UTC. The BTOR is the difference between the predicted BTO and the observed BTO. Then the Start Longitude is varied to minimise the RMS BTOR. Finally the Flight Level is adjusted in steps of 1 (standard altitude steps of 100 feet) to minimise the RMS BTOR. A full report is then produced for each scan. (The definition of GSE is found later in the paper, and the significance of the correlation coefficients is based on the work of Bobby Ulich and will be presented in a future paper.)

A number of MH370 candidate flight paths have been found over the years by various analysts resulting in Regions of Interest (ROIs) that have either already been searched or have been proposed for a further search. The table below lists some of the ROIs. The table includes some new ROIs which have been found as a result of the current systematic search. Some of the ROIs can be readily dismissed as the standard deviation BTO residual (<47 μs), standard deviation BFO residual (<4.3 Hz) or the calculated PDA (<1.5%) is too high.

Possible MH370 flight paths resulting in a candidate Region of Interest (ROI)

Systematic initial bearings from 155°T to 195°T in steps of 1°T were analysed, plus some exotic cases in steps of 0.1°T. All navigation methods were covered: LNAV, CTT, CTH, CMT and CMH, all speed modes: Constant Mach 0.80 to 0.85, LRC 0.7047 to 0.8408, MRC, ECON CI52 and all flight levels: from FL290 to FL430. The fuel endurance was allowed to vary around 00:17:30 UTC and the resulting PDA was noted. The PDA was allowed to vary from the nominal 1.5% and the possibility that the bleed air was shut off for part or all of the time was considered.

In total 1,372 flight paths have been analysed, of which 828 flight paths since the start of this systematic study on 17th February 2019. Start latitudes from 16.0°N to 4.3°S have been covered and the start longitudes were unconstrained. Start times from 18:41:00 UTC to 19:32:00 UTC, but the final major turn had to be completed before the 2nd Arc at 19:41:03 UTC was reached.

Discussion

A more detailed analysis reveals 3 candidate ROIs for further investigation: ROI 1 – LNAV180 CM 0.84 FL403, ROI 2 – LNAV 170 LRC FL350 and ROI3 – CTH170 LRC FL290.

From a pilot’s point of view, a LNAV path on a bearing of 180°T would require setting a final waypoint as the South Pole. This flight path passes close to waypoint BEDAX. The overall fuel endurance and range fits and for a Main Engine Fuel Exhaustion (MEFE) at 00:17:30 UTC, a PDA of 1.37% is calculated (the nominal PDA is 1.5%). The RMS GSE is 2.49, which fits the expected range between 1.0 and 3.0 knots. This flight path ends at 00:19:37 UTC at around 34.5°S near the 7th Arc. This area was originally searched by Go Phoenix but all possible sightings were reexamined and discounted. The search area was widened in later search by Ocean Infinity, but again nothing was found.

An LNAV path on an initial bearing of 170°T starts close to Car Nicobar Airport (VOCX) and passes close to Cocos Island before ending at 00:19:37 UTC at around 28.9°S near the 7th Arc. The overall fuel endurance and range fits and for a MEFE at 00:17:30 UTC, with a calculated PDA of 1.17%. Notably, the Mean BFOR for this flight path is low at -6.87 Hz and is out of the expected range. The area around 28.9°S was searched by Ocean Infinity, but nothing was found.

The CTH path on an initial bearing of 170°T is unlikely as the fuel endurance and range does not fit well. The RMS BTOR is high at 79.6 μs and individual BTOR values are out of normal range. It is also unlikely that a pilot would switch to a True Heading mode. Normal operation is Magnetic Track and this mode is only used for short flight paths, such as during an approach or deviating to avoid bad weather. Switching from Magnetic to True compass mode is only normally done in the region of the north or south poles.

Candidate ROIs for further investigation. (Click on image to enlarge.)

Conclusion

All possible MH370 end points of flight routes in any navigation mode and any speed mode have already been searched, within at least ± 25 NM of the 7th Arc (partially ± 40 NM). This means that MH370 has either been missed in a previous search or recovered from a steep descent of around 15,000 fpm and glided out to an end point outside the previously searched area.

There is only one Region of Interest, where we recommended a further analysis and search at around 34.4 °S near the 7th Arc, following a flight route from close to waypoint BEDAX using the LNAV lateral navigation mode with an ultimate waypoint of the South Pole on a track of 180°T due south, in Long Range Cruise speed mode and at a flight level between FL390 and FL403.

This Region of Interest will be analysed in more depth in the next paper in this series.

1,394 Responses to “A Comprehensive Survey of Possible MH370 Paths”

  1. Richard Godfrey says:

    @All

    Jeff Wise has published an article entitled “The Mystery Behind the Missing Malaysia Airlines Flight Isn’t Solved Yet” dated 28th June 2019.

    https://onezero.medium.com/the-mystery-behind-the-missing-malaysia-airlines-flight-isnt-solved-yet-476f65d9b597

    In my view, the article is an example of outrageous cheap journalism. There may be gaps in the MH370 story, but there are bigger gaps in the Jeff Wise fantasy.

    I believe Blaine Gibson made genuine finds of MH370 debris, as did Johny Begue, Schalk Lückhoff, Neels Kruger, Liam Lotter, Milson Tovontsoa, Rija Ravolatra, Eodia Andriamahery, Jean Dominique, Suzy Vitry, Barry McQade, Jean Viljoen and others. 20 items of debris have been confirmed or are likely to have come from MH370. That these items of debris were flown intact to Kazakhstan, then subsequently damaged to simulate a crash, then subsequently exposed to marine life for a month and finally planted in 27 locations in 7 countries for 14 different people to find, is preposterous nonsense.

    Blaine has travelled to 185 countries and speaks 6 languages fluently including Russian. This is not proof of anything suspicious. I have travelled to 130 countries and speak 4 languages fluently.

    It is public knowledge that Blaine worked on various government sponsored projects between the US and Russia from 1998 until 2005. This included the US independent verification of the Russian reform of the Atomic Industries Enterprises (including the Atomic Weapons Program), which was based in the closed Russian Nuclear Cities of Snezhinsk, Zarechny and Obninsk. Blaine was based in Washington DC and Seattle, but travelled extensively to Russia. Blaine has not travelled to Russia in the last 15 years. There are many international verification programmes which include US citizens travelling to Russia, which are in operation even today. This is not proof of anything suspicious. I have also travelled to Russia. I have also worked on government sponsored projects.

    It is public knowledge that Blaine set himself up as a consultancy and trading company, called the Siberia-Pacific Company in 1992. The company is registered at his home address in Seattle, with Blaine as President. US trade with Russia in 1992 was 2.6B$ and in 2018 had reached 27.5B$. Siberia-Pacific has been inactive for the last 20 years and was dissolved last year. This is not proof of anything suspicious. I am a one man consultancy company registered at my home address. I have done business in 17 different countries.

    It is public knowledge that Blaine Gibson sold his parent’s home in April 2014 for $1,185,000. This would easily fund travels to various locations bordering the Indian Ocean. This is not proof of anything suspicious. It is not uncommon for children to inherit from their parents.

    One Russian and two Ukrainian passengers on MH370 is not proof that MH370 was sabotaged. There were 14 different nationalities on board (not counting stolen passports from Austria and Italy). It may come as a surprise to Jeff Wise, but Russians travel too. There are flights from Russia to 110 international destinations. The volume of International flights is only surpassed by the USA. This is not proof of anything suspicious.

    You may think what you will of the Jeff Wise fantasies, but his personal attack on Blaine Gibson is unfounded, deceitful, actionable and divisive. I am sure you would all join with me in defending Blaine Gibson’s integrity and his great service in helping to solve the mystery of MH370.

  2. Victor Iannello says:

    @TBill: In the PMDG 777 model, if you remove power from the left bus, you will see the message DATALINK LOST, among other messages. However, in the model, I don’t know a way to shutdown the ACARS without removing power to the entire SATCOM. There’s only so much realism you can expect in low-cost software.

  3. Victor Iannello says:

    @Richard Godfrey: Well said. Jeff Wise’s continued attacks on Blaine Gibson are despicable. Unfortunately, Blaine’s reputation is collateral damage in Wise’s futile attempt to prop up a long-discredited theory.

  4. airlandseaman says:

    Thanks Richard. Jeff Wise’s attacks on Blaine and William are totally without merit. I agree with Victor that those comments are nothing more than a desperate attempt to prop up Jeff’s nonsensical theories about MH370. He should be ashamed for his statements, and his acts.

  5. Paul Smithson says:

    “All possible MH370 end points of flight routes in any navigation mode and any speed mode have already been searched, within at least ± 25 NM of the 7th Arc (partially ± 40 NM). This means that MH370 has either been missed in a previous search or recovered from a steep descent of around 15,000 fpm and glided out to an end point outside the previously searched area.” Yet your table above shows a path 192 ending at 40.3S. My preferred route, as previously mentioned is ~191 and ends just shy of 40S. Neither searched at any width. I am also surprised that far from being infeasible on fuel, as has been so often asserted, this table appears to show feasible pda. I’m not clear why these non-searched latitudes of roughly 39.5-40.5S would not be at the top of the list!

  6. paul smithson says:

    I am also surprised at how great the max/min and rms BTO errors are on your shortlisted 3. These are not particularly good BTO fit compared to other path solutions, c.f. Vi’s great circle paths paper, showing the most southern routes with best BTO fit (rms in the low 20s).

  7. Richard says:

    @Paul Smithson

    You state: “Yet your table above shows a path 192 ending at 40.3S.”

    The table was showing the list of all candidate regions of interest from various analysts over the years. “A number of MH370 candidate flight paths have been found over the years by various analysts resulting in Regions of Interest (ROIs) that have either already been searched or have been proposed for a further search.”

    You ignore the statement that follows in both Victor’s article and in the paper “Some of the ROIs can be readily dismissed as the standard deviation BTO residual (<47 μs), standard deviation BFO residual (<4.3 Hz) or the calculated PDA (<1.5%) is too high."

    The LNAV192 route that you reference has a standard deviation BFO residual of over 6 Hz. This is too high in my view and disqualifies the route.

    The reason that non-searched latitudes of roughly 39.5-40.5S are not at the top of the list is because in general the BTO residuals and BFO residuals are too high (please see Figure 16 and 17 in the linked paper).

    You further state "My preferred route, as previously mentioned is ~191 and ends just shy of 40S."

    I will be pleased to run a flight path LNAV191.0 CM 0.84 FL410 through my model if you wish, or would you prefer a different initial bearing, navigation mode, speed mode or flight level?

    In any case, at this stage we are only making an initial recommendation and all possible ROI candidates will be evaluated again in more detail.

  8. Richard says:

    @Paul Smithson

    You stated “I am also surprised at how great the max/min and rms BTO errors are on your shortlisted 3. These are not particularly good BTO fit compared to other path solutions, c.f. Vi’s great circle paths paper, showing the most southern routes with best BTO fit (rms in the low 20s).”

    The purpose of the paper presented here is to give the results of the wide scan search.

    Victor’s purpose in his article on great circle paths in 2017 was evaluating specific flight routes.

    Bobby has pointed out that the goal is not to find a flight route with all BTORs = 0 µs and all BFORs = 0 Hz.

    Again I repeat my offer to run a flight route of your choosing through my model, so we can compare the results of your model and mine.

  9. DennisW says:

    @Richard

    Using BFOR as a route qualifier is not sound from a mathematical standpoint for reasons I have voiced here several times (BFO data is neither stationary nor ergodic and STDEV should not be used). Because one can compute a BFOR does not mean that it has significance.

    Your CTT160 and CTT158 routes look very good otherwise.

  10. Peter Norton says:

    Richard Godfrey: “finally planted in 27 locations in 7 countries”

    I don’t think JW ever stipulated that. The debris doesn’t need to be planted at different times in different places, but it would be enough to throw it all into the water somewhere, preferably somewhere near ARC7 and let the currents do the rest.

  11. DrB says:

    @DennisW,

    You said: “Using BFOR as a route qualifier is not sound from a mathematical standpoint for reasons I have voiced here several times (BFO data is neither stationary nor ergodic and STDEV should not be used). Because one can compute a BFOR does not mean that it has significance.”

    You keep repeating the same mistakes and reaching the same, and incorrect, conclusion. This has previously been discussed at length on this forum by Mike and me in comments made specifically to you.

    Here (again) are the FACTS:

    1. The BFO reading noise comprises multiple components, including random electronic noise in the GES frequency measurement equipment, pseudo-random frequency errors impressed on the transmitted frequency by limited precision trigonometric functions in the frequency compensation algorithm in the AES, and frequency drift (frequency noise) in the OCXO in the SDU.
    2. The frequency drift of the OCXO is characterized by the Allan variance, and it is non-ergodic and non-stationary.
    3. The time scale for the OCXO Allan variance (i.e., the OCXO frequency drift) to become significant or dominant relative to the other sources of BFO reading error is days.
    4. Applying the specified Allan variance for the OCXO on short time scales (seconds to minutes) shows that the expected magnitude of OCXO frequency drift is smaller than the observed BFO reading errors by several orders of magnitude.
    5. On time scales of hours or less (i.e., during a single flight) the Allan variance effect is a negligible portion of the observed BFO reading errors. Thus DSTG’s Figure 5.5, which is the observed probability density function of BFO reading errors, is insignificantly affected by the OCXO Allan variance. That is because the mean value has been removed for each flight, which effectively removes the OCXO frequency drift characterized by the Allan variance.
    6. Thus, BFO best-fit residuals, for a single flight, which lie outside the boundaries of the the PDF in Figure 5.5 are indicative of route parameter errors (NOT OCXO drift).
    7. Thus, fitted BFO residuals are useful in evaluating routes during a single flight.
    8. Thus, Richard (and DSTG and everyone else who has ever built a route fitter) is correct to use the BFO residuals (other than the mean value) as a route discriminator.

  12. Richard says:

    @Peter Norton

    You stated “Richard Godfrey: “finally planted in 27 locations in 7 countries” I don’t think JW ever stipulated that.”

    That is exactly what Jeff Wise stated and he names Blaine Gibson and the Russians as the perpetrators:

    (1) Jeff Wise 14th April 2016

    http://jeffwise.net/2016/04/14/mh370-debris-was-planted-ineptly/

    “There is only one reasonable conclusion to draw from the condition of these pieces. Since natural means could not have delivered them to the locations where they were discovered, they must have been put there deliberately. They were planted.”

    “Given how little inquiry had been directed at the Réunion piece, whoever planted the most recent four pieces might reasonably have assumed that the public would accept the new pieces uncritically, no matter how lackadaisical their preparation.”

    “So maybe whoever planted the debris in Mozambique, South Africa, and Rodrigues weren’t lazy–maybe their understanding of human psychology simply allowed them to take the minimum steps necessary. Whether their calculation was accurate or not will now become apparent.”

    (2) Jeff Wise 29th April 2016

    http://jeffwise.net/2016/04/29/mh370-debris-questions-mount/

    “I understand that not everyone is ready to accept that the absence of marine life can only mean that the debris was planted.”

    (3) Jeff Wise 16th October 2016

    http://jeffwise.net/2016/10/16/what-if-zaharie-didnt-do-it/

    “Granted, we are still left with the issue of the MH370 debris that has been collected from the shores of the western Indian Ocean. Many people instinctively recoil from the idea that this debris could have been planted”

    (4) Jeff Wise 10th December 2016

    http://jeffwise.net/2016/12/10/is-blaine-alan-gibson-planting-mh370-debris/

    “I find it quite extraordinary that a purported piece of MH370 apparently washed up on the shore within half an hour of Blaine’s passing by the spot.”

    (5) Jeff Wise 10th March 2019

    http://jeffwise.net/2019/03/10/who-is-blaine-alan-gibson/

    “If the hijacking of MH370 was a Russian plot, and MH370 flew to Kazakhstan, then the pieces of debris collected in the western Indian Ocean must have been planted by the Russians in an effort to support the misleading southern narrative. Blaine Alan Gibson had demonstrated an uncanny knack for locating and publicizing this debris. Was Gibson somehow connected to Russia?”

    (6) Jeff Wise 28th June 2019

    https://onezero.medium.com/the-mystery-behind-the-missing-malaysia-airlines-flight-isnt-solved-yet-476f65d9b597

    “But Langewiesche finds the idea of planting evidence inconceivable — mainly, it seems, because a large proportion of the debris was found by an American named Blaine Alan Gibson. And here, I think, is where his story really goes off the rails.”

  13. Richard says:

    @Niels

    You asked “Could you perhaps help to evaluate the proposed route (21:11:02 onwards)?”

    The key parameters of your case C are as follows:
    – LRC to fuel exhaustion.
    – FL 335.3.
    – 21:11:02 position of 9.191°S, 93.692°E.
    – Constant track after 21:11:02 of 178.098 degrees.

    Here are the results for MH370 Flight Path Model V19.7 RG CTT LRC FL335.3 178.098 Niels Case C:

    https://www.dropbox.com/s/xmavpyw8hzixug4/MH370%20Flight%20Path%20Model%20V19.7%20RG%20CTT%20LRC%20FL335.3%20178.098%20Niels%20Case%20C%20Full%20Report.png?dl=0

    The BTOR, BFOR and PDA all appear to fit from 21:11:02 UTC onwards.

    I will examine similar flight paths CTT178.098 LRC FL335.5 from 19:30 UTC onwards with a full set of correlation coefficients.

  14. paul smithson says:

    Re BFO. I thought that the consensus now was that BFO error under ~7hz should not be considered as a “hard discriminator”, although lower BFO errors would be expected/preferred. Dr B says that oscillator drift occurs only over longer timescales and appears to be saying that BFO is “back in the game” as a hard criterion for solution selection. Do others with expert knowledge of oscillator characteristics agree? Dr B, does your view still hold in the case of an oscillator that has been power-cycled? With the possibility of much colder ambient temp (and temp gradients within the temp-stabilised casing)?

    Re Fuel. I am also confused (but pleasantly surprised) that 40S seems to be fuel feasible (PDA 1.3 – 1.4). In previous correspondence, Dr B was adamant that this was not even remotely reachable, even with packs off. Could one of the authors please explain how the estimate of fuel requirement for southerly paths has changed so radically?

  15. Richard says:

    @Paul Smithson

    (1) Re BFO please see the comment from DrB at 04:40 am today (above).

    (2) Re PDA it has not changed, Bobby, Victor and I all have different models.

    (3) I note you have declined my offer to model a flight path of your choosing.

    (4) What results does your model give for your preferred path ca. 191, in particular for BTORs, BFORs and PDA?

  16. DennisW says:

    @DrB

    The time scale for the OCXO Allan variance (i.e., the OCXO frequency drift) to become significant or dominant relative to the other sources of BFO reading error is days.

    You might want to check out figure 5.4 of the DSTG book. The random walk behavior of quartz oscillators begins at about 100 seconds.

    https://photos.app.goo.gl/uxMW4mxeoagvGr1S7

  17. DennisW says:

    @DrB

    You might also ponder why no oscillator manufacturer uses mean and variance (Gaussian statistics) to characterize oscillator behavior.

  18. DrB says:

    @DennisW,

    You continue to ignore actual facts.

    First you said: “You might want to check out figure 5.4 of the DSTG book. The random walk behavior of quartz oscillators begins at about 100 seconds.”

    The feature displayed in Figure 5.4 has already been explained (to you directly) by @airlandseaman. He was told that less-than-2-hour excursion was not identified on other flights and actually repeated ( i.e., re-traced) on the return flight. DSTG properly concluded it had a different cause than oscillator drift.

    Even if oscillator drift begins at 100 seconds, that does not demonstrate it was significant in amplitude. In fact, we do know that occasionally drifts of 16 Hz occurred over periods typically of several days (about 172,800 seconds). Over a period of 100 seconds, the same oscillator would exhibit an amplitude reduction by a factor of SQRT(100/172,800) = 42. So at 100 seconds the worst-case amplitude could be as large as 16/42 = 0.4 Hz, and thus be insignificant compared to other BFO reading noise sources (which have nothing to do with the OCXO).

    You also said: “You might also ponder why no oscillator manufacturer uses mean and variance (Gaussian statistics) to characterize oscillator behavior.”

    You continue to conflate oscillator frequency noise with BFO reading error. They are NOT the same thing. You continue to ignore the fact that the BFO reading error is NOT dominated by OCXO frequency drift during a single flight. The major contributors to BFO reading error DO have repeatable statistics, including mean and standard deviation.

  19. paul smithson says:

    @Richard. Thank you for taking the time to respond. However, your defensive tone implies that you read some sort of veiled offense into my remarks?

    1) Yes, it was to that post that my query referred. And you didn’t answer the question
    2) I’d like to hope that your different models don’t produce answers that vary +/- 2 percentage points PDA.
    3) Where did I say that I had declined your offer?

  20. DrB says:

    @paul smithson,

    You said: “Dr B, does your view still hold in the case of an oscillator that has been power-cycled? With the possibility of much colder ambient temp (and temp gradients within the temp-stabilised casing)?”

    As I have previously explained, there is no reason to think the frequency drift characteristics of the OCXO would be changed by thermal cycling. I do allow the BFO bias frequency to be different after 18:24 to accommodate the possibility that the SDU was cold soaked to -55C during the previous hour. We now know that the change in bias frequency compared to take-off was at most 3-4 Hz, so either the cold soak did not occur or it did not shift the OCXO frequency substantially.

  21. Richard says:

    @Paul Smithson

    When did you accept my offer or answer my question regarding your model?

  22. DennisW says:

    @DrB

    The feature displayed in Figure 5.4 has already been explained (to you directly) by @airlandseaman. He was told that less-than-2-hour excursion was not identified on other flights and actually repeated ( i.e., re-traced) on the return flight. DSTG properly concluded it had a different cause than oscillator drift.

    I don’t recall that exchange, but I will take your word for it. Was the cause ever explained or are you choosing to simply ignore it? If we had the data from previous flights we could characterize typical BFO error associated with oscillator drift. As it stands I would not use BFO as a path qualifier.

  23. DrB says:

    @DennisW,

    DSTG determined that particular event had a “geographical dependence”. They never identified a specific cause, nor were similar occurrences noted. DSTG does not say whether this event was included in the histogram in Figure 5.5, which is what I used for my BFOR reading error model.

    If you ever create a route fitter, you are free to choose whatever criteria you wish to use. However, as I have said before, the major value of the BFOs is to require a Southern Hemisphere solution. With only that, the answer is unchanged. The only ROI which is consistent with all the data is LNAV 180. Adding the standard deviation of the BFORs produces stronger rejection of more southerly and more northerly routes, but these are also rejected without the BFORs.

  24. DennisW says:

    @DrB

    “determined” = “speculative”. No matter. I am off this subject. Be careful :-).

  25. paul smithson says:

    Dr B. Thank you for your response. I understand you to be saying that a step-change of bias freq after cold-soak is conceivable, but not drift of growing amplitude. I’m not sure what the basis is for your confidence that “We now know that the change in bias frequency compared to take-off was at most 3-4 Hz”? Could you please elaborate?

  26. Niels says:

    @Richard
    Thank you for evaluating “Case C”. The results are encouraging. The main difference with “Case A” is that for each chosen FFB (in the range of 150 Hz – 153 Hz with 0.5 Hz steps), I minimized the variance(track) for the interval 21:11 – 00:19 instead of for the interval 19:41 – 00:19 [by varying the starting lat, lon only]. This results in a much straighter track in the 21:11 – 00:19 interval.

    Furthermore I select the flight level for each FFB/path by subsequently minimizing the mean error in TAS (error defined as the difference between calculated TAS and expected TAS for LRC). The path (FFB) that is finally selected as “optimum” is the one having the smallest RMS(delta_TAS). The key parameters (FL, mean track and 21:11 position) are directly extracted from the data sheets.

    Note that the path estimator indicates a path roughly from BEDAX through ISBIX to the 21:11 position (ISBIX pass around 19:57 UTC) and a settling at the near 178 degrees track in the vicinity of the 21:11 position, which is less than 100 nm north of BEBIM. The speed profile indicates that perhaps the FL was gradually increasing until about the 21:11 position as well (and then settled at a constant level). The indicated 19:41 position is 2.36, 93.75 (BTO calculation for 11 km) at a GS of about 432 kts.

    Finally: this is all based on a “best fit” interpolation of BFO and BTO data, which not necessarily represents the “true” BFO and BTO curves.

  27. DrB says:

    @paul smithson,

    You said: ” I’m not sure what the basis is for your confidence that “We now know that the change in bias frequency compared to take-off was at most 3-4 Hz”? Could you please elaborate?”

    This conclusion is based on the fact that all ROIs which show > 1% probability of being the True route also have 19:41-00:11 mean BFORs near -3 Hz.

  28. paul smithson says:

    Dr B. Thanks for the response. But is there not a degree of circularity in that reasoning? I’d be more comfortable with some experimental evidence of the behaviour of the equipment in question.

  29. Victor Iannello says:

    @Paul Smithson said: I’d be more comfortable with some experimental evidence of the behaviour of the equipment in question.

    There is experimental evidence of a shift in frequency bias after a power up for the SDU on 9M-MRO. From a previous post:

    It turns out there is an effect called “retrace” that causes oscillators that are powered down, cooled, and powered up to shift in frequency, and there are indications that a retrace shift of about -4 Hz occurred while 9M-MRO was on the ground at KLIA before the MH370 flight. A similar shift, but in the opposite direction (up) might have occurred due to the inflight power cycling.

  30. TBill says:

    @Victor @Richard
    Catching up on my reading, do we have handy link to “How to play Russian Roulette and Win”…trying to recall if I read that paper.

  31. DrB says:

    @paul smithson,

    You said:”Dr B. Thanks for the response. But is there not a degree of circularity in that reasoning?”

    Actually, no there isn’t. The few ROIs which show a significant probability of being the True Route, when the BFORs are not included at all in the statistics list, have mean BFORs near -3 Hz. Those same ROIs also have the best match to the expected BFOR standard deviation.

  32. Peter Norton says:

    @Richard:
    re debris: thank you, I stand corrected

  33. Mick Gilbert says:

    @DrB
    @Richard

    Gents, like Paul, I was very surprised to read that there are sub-39°S termini returning feasible PDAs. I’m sure that I can recall previous discussions where DrB’s fuel flow model effectively limited termini to around 38.5°S.

    Is there a different base assumption between DrB’s earlier work and Richard’s most recent work that explains the difference?

  34. Julia Farrington says:

    @Andrew Just to say thank you very much for your helpful reply to my comment on the previous blog about the 737 Max. I have only just seen it. I appreciate the time you took to explain. Thanks! Julia

  35. Richard says:

    @Mick Gilbert, @Paul Smithson

    The fuel endurance and fuel range calculations are complicated and based on many assumptions.

    The last ACARS message at 17:06:43 UTC shows that the aircraft Gross Weight was 480,600 lbs (rounded to the nearest 100 lbs) and the Fuel Weight was 43,800 kg (rounded to the nearest 100 kg). This would make the aircraft Zero Full Weight (ZFW) 217,997 kg – 43,800 kg = 174,197 kg. The Malaysian Safety Investigation report states the ZFW is 174,369 kg. What was the precise amount of fuel at 17:06:43 UTC? What was the ZFW? What was distribution between the fuel tanks?

    The last ACARS message was at 5.299°N 102.813°E at 35,004 feet and before the diversion after waypoint IGARI. The flight path between 17:06:43 UTC and 18:22:27 UTC is not precisely known, the flight path between 18:22:27 UTC and 19:41:03 UTC is not known at all and the flight path between 19:41:03 UTC and 00:19:37 UTC is not precisely known. What speed mode was selected? What navigation mode was selected? What Flight Level was selected? Was there any climbs or descents? Was there any step climbs? Was there an intermediate landing or attempted landing?

    The fuel used is dependent on time, aircraft weight, fuel flow coefficients, temperature deviation from ISA, Mach, Performance Degradation Allowance of each engine, speed mode (Constant Mach, LRC, MRC, ECON, etc) and other loads on the engine (Integrated Drive Generators, Bleed Air System). We are told in a Rolls Royce report that the right engine fuel consumption is greater than the left engine fuel consumption. Was there any fuel transfer between tanks? What was the actual PDA of each engine?

    The TAS for a given Mach depends on the Outside Air Temperature (OAT) at the Flight Level chosen. The Ground Speed depends on the navigation mode, Wind Speed and Wind Direction. The range will depend on head/tail/cross winds encountered over the whole flight path. I use the GDAS weather data, but this is only available every integer degree of Latitude and Longitude, every 50 hPa of Pressure Altitude and every 3 hours in time. A quadrilinear interpolation has to be made. I analysed the dynamic in the GDAS for the MH370 solution space from 15°N to 40°S and from 80°E to 105°E, from 150 hPa to 350 hPa and from 18:00 UTC to 03:00 UTC+1. (1) As you would expect, the OAT changes predominantly with Altitude by on average -10.37°C per 50 hPa (worst case -17.1°C), but little over 1° Lat/Lon or time (worst case -3.6°C over 1° Lat). (2) The Wind Speed changes predominantly with Time and Altitude (worst case -120.74 knots over 3 hours), but less over 1° Lat/Lon (worst case -34.68 knots over 1° Lat). (3) The Wind Direction changes significantly with Time, Latitude, Longitude and Altitude (worst case ±180.0°T). I analysed the ACARS vs GDAS for MH371. During one stable part of the cruise, the differences between the GDAS model and the actual ACARS data were as follows: The OAT Actual – Predicted had a Mean Error of -1.0 °C and Standard Deviation of 2.5 °C. The WSPD Actual – Predicted had a Mean Error of -0.8 knots and Standard Deviation of 1.3 knots. The WDIR Actual – Predicted had a Mean Error of +1.6 degT and Standard Deviation of 6.8 degT. What was the actual weather data for a particular flight path? What is the accuracy of the GDAS data used to model a particular flight path?

    I do not claim to have made all the right assumptions or correct calculations. Bobby’s fuel model is probably more accurate than mine and Bobby certainly simulates the end of flight more precisely with the INOP after the first engine flame out prior to 00:11:00 UTC and MEFE after the second engine flame out at around 00:17:30 UTC more precisely than I do, with the resulting engine restart attempts, APU start, speed reduction and altitude loss. We agree quite closely on the aircraft and fuel weight at 19:30 UTC. There are small differences in the temperature and winds over the course of the next ca. 5 hours of flight. There are large differences in the end of flight simulation.

    I am genuinely open to see whether a flight path as proposed by Paul Smithson at 191° to 40°S fits the data we have. As you can see, there is a huge difference between a LNAV191, CTT191, CTH191, CMT191 or CMH191 path. There is also a huge difference between a LNAV191 CM 0.84 FL410 and a LNAV191 LRC FL350. There is no generic statement that all 191 flight paths fit the fuel data or not.

  36. Andrew says:

    @Julia Farrington

    You’re very welcome!

  37. Richard says:

    @TBill

    You asked: “Catching up on my reading, do we have handy link to “How to play Russian Roulette and Win”…trying to recall if I read that paper.”

    The link to the paper can be found in this comment:

    https://mh370.radiantphysics.com/2019/01/12/mh370-flight-around-penang/#comment-21779

  38. Richard says:

    @Paul Smithson

    You stated:”@Richard. Thank you for taking the time to respond. However, your defensive tone implies that you read some sort of veiled offense into my remarks?”.

    Thank you for taking the time to criticise the new paper.

    Your offensive tone was not veiled. You started to throw rocks at the new paper, less than 2 hours after publication.

    Since you meanwhile complain, that I was wrong to say that you had rejected my offer, when are you actually going to answer my question regarding the key parameters of your proposed flight path and accept my offer to model your flight path? I only need to know navigation mode, flight level and speed mode.

    I wanted to compare your results with mine.

  39. TBill says:

    @Richard
    Very nice paper…I like that Feb_2019 “Russian Roulette” paper very much. I will have some questions upon further review. Briefly, I am thinking, yes, Left and Right buses off at IGARI to go dark and maybe cut off DCDR recording session, and perhaps other side benefits. This potentially ties in with William Langewiesche’s recent article where, if I undertsand, he suggests all power off at IGARI.

  40. TBill says:

    Correction: DFDR (flight data recorder)

  41. DrB says:

    @Mick Gilbert,

    You said:”Gents, like Paul, I was very surprised to read that there are sub-39°S termini returning feasible PDAs. I’m sure that I can recall previous discussions where DrB’s fuel flow model effectively limited termini to around 38.5°S.
    Is there a different base assumption between DrB’s earlier work and Richard’s most recent work that explains the difference?”

    Neither Richard’s paper nor his comments say the PDA is acceptable for routes ending circa 40S. In fact, those routes have unacceptable BTO residuals, unacceptable BFO residuals, and unacceptable PDAs.

    Richard’s subsequent comment aptly summarizes all the aircraft and environmental factors which affect fuel consumption. There is a Region of Interest (ROI) at 191/192 degrees for “straight” paths. This route requires very high true air speed in order to roughly match the BTOs. Very high air speed results in a serious degradation in fuel mileage, and, using my fuel model, this produces a large fuel shortfall (~5%) for this ROI, ending circa 40.3S. This fuel shortfall cannot be fully compensated by turning off bleed air and one IDG. If one ignores the BTOs and flies slower at MRC, the range is maximized and 40S becomes marginally possible with bleed air off. However, the disagreement in BTOs is hopelessly large. So, it’s one thing to draw a maximum range arc, and then extend it by several percent assuming bleed air was off the entire time after circa 17:30, but such a route is incompatible with both BTOs and BFOs

  42. Richard says:

    @TBill

    You stated: “@Richard Very nice paper…I like that Feb_2019 “Russian Roulette” paper very much.”

    Many thanks!

    It is encouraging to receive a compliment.

    This is especially true since the “Russian Roulette” paper took 6 weeks to produce and the “Blowin’ in the Wind” paper took 4.5 months to produce.

  43. Richard says:

    @Peter Norton

    You stated “@Richard: re debris: thank you, I stand corrected”

    No worries! It takes a Gentleman to say, “I stand corrected”.

  44. Niels says:

    @Richard & co-authors
    Many thanks for sharing your new paper, summarizing past months systematic scan for flight paths that fit the data “within appropriate tolerances”. It is an impressive and important effort, which deserves careful reading. I’m still studying some parts more in detail, however for now on one point I would like to ask you to explain a bit further.

    In the discussion part, the ROI 1 is mentioned (LNAV180 CM 0.84 FL403), which seems to be a bit high on the STDEV BTOR.

    In the method part, at slightly lower FL390 and LRC speed setting, fig. 4 and fig. 5 give slightly lover values for STDEV BTOR / RMS BTOR.

    In the conclusions reference is made to “Long Range Cruise speed mode and at a flight level between FL390 and FL403”.

    Should I read from this that some optimization is still ongoing regarding the LNAV180 path? What is the typical margin on the 7th arc latitude?

  45. Mick Gilbert says:

    @Richard
    @DrB

    Thank you, gentlemen, for your separate responses. Much appreciated, as always.

  46. DennisW says:

    @Richard

    It is encouraging to receive a compliment.

    It is, indeed. Even more important is a careful read, and considered feedback. As a VP in a previous life responsible for a significant P&L, I nevertheless sat through every design review and code review. It sent a strong message to the team that I considered them to be important, and I took the time to prepare with diligence.

    As you know (or should know), I have a high regard for your work. The fact that I am critical of your (and DrB’s) recent efforts is not a reflection on your competence (or DrB’s). It is simply my deep seated belief that a reliance on BFO as a path qualifier is a mistake. I would be happy to be wrong, and there is ample precedence for that.

    However, to not comment would be out of character for me. I am too old for that.

  47. Mick Gilbert says:

    @DrB

    Bobby, I understand that you are saying that Richard’s LNAV192 CM0.838 FL355 route terminating at 40.30S has unacceptable BTO and BFO residuals but the PDA for that route is 1.2835%.

    Given that the nominal PDA is 1.5%, is 1.2835% unacceptable?

  48. DrB says:

    @Mick Gilbert,

    My LNAV 192.26 route at M0.84 and FL355 ends at S40.3 and has a PDA of -4.1% for a MEFE at 00:17:30. That PDA is obviously unacceptable, and that is why I said so. The PDA will be reduced in order to accommodate the high air speed, relatively low altitude, and elevated SAT on this night on this route, all of which increase the fuel flow.

    Richard’s value for PDA, as given in the table in this post and which you quoted, is much different. We shall investigate this difference and report back.

  49. Victor Iannello says:

    @Mick Gilbert: To look at the PDA for a path that has unacceptable BTO error doesn’t have much meaning (unless you are questioning our understanding of the BTO values).

    In the previous article on great circle paths, which limited the survey to FL350, by observing Figure 2 and the results in the CSV file, you can see that for a track angle of 192° at 19:41, the paths cross the 7th arc near 40S latitude. If the speed mode is constant Mach number, the BTO error is acceptable (21.3 μs RMS), but the required speed is M0.842, which is substantially higher than LRC speed, and there is insufficient fuel. (For the time period after 19:41, LRC corresponds to an average speed of about M0.81.) On the other hand, if the speed mode is LRC, less fuel is consumed, but the BTO error increases to 91.7 μs, which is unacceptable. It is misleading to consider fuel consumption without also considering BTO error.

  50. Richard says:

    @Mick Gilbert @Paul Smithson

    My apologies!

    The entry for the LNAV192 CM 0.838 FL355 in the table above is not correct.

    This values came form an old version of my model by mistake.

    I have checked this flight path again with my latest version.

    Bobby is right that this flight path does not fit the fuel consumption, BTORs or BFORs.

    I also get a negative PDA for a MEFE at 00:17:30 UTC, high BTORs and high BFORs.

    I will run a thorough check on all the values in the table.

    This flight path candidate remains disqualified, even more so than before.

  51. Mick Gilbert says:

    @Victor Iannello

    Thank you Victor. No, I am not questioning our understanding of BTO values.

    I am not interested in fitting a route. What I am interested in is establishing the southernmost bound on the 7th arc based on the fuel modelling. I thought that had been established at around 38.5°S. Richard’s recent paper seemed to call that into question.

    Accordingly, what I was questioning was the apparent marked difference between the PDA values as originally published by DrB and those published by Richard for not overly dissimilar routes.

    DrB has acknowledged that difference and will investigate. I will standby to stand by.

  52. Mick Gilbert says:

    @Richard
    @DrB

    Okey doke, thanks Richard.

    That issue aside, your paper (like the many that have gone before it) represents a considerable body of work and scholarship. It was somewhat remiss of me not to have acknowledged that earlier.

    Similarly, I was remiss in not acknowledging DrB’s earlier work on the fuel modelling.

    So, a formal thank you to you both and a thank you to Victor for making the your efforts and the ensuing discussions widely accessible.

  53. paul smithson says:

    @Richard, I am glad that you have identified the reason for the apparent discrepancy on fuel feasibility of the 190-192 paths.

  54. Richard says:

    @Mick Gilbert

    Many thanks for your kind words and thanks.

  55. Richard says:

    @Niels

    You asked: “Should I read from this that some optimization is still ongoing regarding the LNAV180 path? What is the typical margin on the 7th arc latitude?”

    (1) Yes, we are still comparing models in a comprehensive test program. There may be minor adjustments to come.

    (2) Our 3 models are very close for the LNAV180 LRC FL390, but subject to a number of assumptions. Bobby is doing a sensitivity analysis and we will then be in a better position to state typical margins.

  56. DrB says:

    @Mick Gilbert,

    My work on the fuel flow modeling took over 1,000 hours and 10 months to complete. However, it would not have been possible to achieve the desired accuracy were it not for data provided me by ATSB under a non-disclosure agreement and also non-public B777 information provided me by a confidential source. Both were extremely helpful.

  57. Brian Anderson says:

    Some comments on PDA

    PDAs, in the way that they have been calculated in the recent work by Richard, DrB and Victor, are are a useful discriminator in validating, or eliminating, possible flight tracks. However, what is being calculated is not actually a PDA at all. Rather it is a variation, or delta, from the fuel burn that would normally be expected from Boeing data that would apply to the particular circumstances relating to speed, altitude, temperature, pressure, wind etc, etc, and will include actual differences in engine efficiency from nominal, for for the flight track in question.

    PDA, on the other hand, is the Performance Degradation Allowance used in flight planning to determine the additional fuel to be uplifted, making an allowance for the age, and hence the reduction in efficiency from a new (maybe nominal) engine. The PDA for every engine can therefore be different, but by definition the PDA will never be negative. Because it is not unusual for engines on a particular airframe to have significantly different Time in Service, I suppose it is also possible that the PDAs might be averaged for flight planning purposes.

    I think we know, both from ACARS data, and from other sources (and DrB is privy to other information) that the “average” PDA for MH370 is of the order of 1.5%. I think we also know that it would be unusual for the PDA of any engine to exceed 3%. But it can’t be negative.

    If, as a result of the modelling it is determined that the delta in fuel burn for a particular track turns out to be -x%, then that in itself is interesting. But the reason is not necessarily just related to the PDA itself. It is the variation in overall performance of the aircraft from that which would be expected, bearing in mind the variability (or uncertainty) of all the other factors that affect fuel burn for the flight track being modelled.

    I think understand how the PDA is being used in the modelling, (Richard and DrB . . I am open to correction if I have it wrong), but perhaps it would be more appropriate to define some other term to describe the variability in overall fuel burn. (DeltaF) Actually this new variable can of course be compared to the known PDA range for MH370.

    Am I being overly pedantic? Perhaps.

  58. Niels says:

    @Richard, DrB
    What I don’t understand in your fuel flow analysis / PDA estimates (and using this as one of the path selection criteria) is how you estimate the initial (19:41) fuel quantity. Can you please explain?
    Imo, either it is an unknown (in which case you have an extra degree of freedom in the problem), or it is determined by calculating back and assuming fuel exhaustion (somewhere in the 00:11 – 00:19 interval) in which case you have to work with a fixed PDA.

  59. Richard says:

    @Brian Anderson

    Switching off an Integrated Drive Generator (IDG) or shedding load from an IDG and/or switching off the Bleed Air System can reduce the fuel consumption of an Engine. I did think of introducing a new term such as Performance Improvement Allowance (PIA) for this case, but found it more confusing having 2 terms PDA and PIA, rather than allowing the PDA to go negative. I agree we are abusing the original definition and intent of the term PDA.

  60. Brian Anderson says:

    @Richard,

    I think two terms are unnecessary, because even if you introduce a new one which is allowed to go negative, the PDA in the model is still an amalgam of the real PDA, and other factors that cause a change in the nominal fuel burn. (e.g. load shedding etc).

    It is sufficient to define a new term, say DeltaF, which replaces the current PDA. Then variations identified in the DeltaF, which might be positive or negative, can be compared with the real PDA, and possibly explained by such things as load shedding.

  61. Andrew says:

    @Brian Anderson

    Interesting point – the overall performance degradation from the book figures is related to both the engines and airframe. Airlines use airplane performance monitoring (APM) programs to determine a performance degradation factor for flight planning purposes, based on cruise performance reports sent from the aircraft. The same factor is entered into the FMC to provide the pilots with accurate in-flight predictions of the fuel remaining at flight planned waypoints and destination. Boeing refers to it as the FF (Fuel flow) factor, while Airbus uses the term PERF (Performance) factor.

    As an aside, it is possible to have a FF/PERF factor less than zero, because the book level reflects the fleet average of new aircraft and engines. There is inevitably a scatter that leads to performance above and below the book value for newly delivered aircraft and indeed, some older aircraft. We have several B777-300ERs (not new!) in our fleet that have FF factors below zero. Airbus uses a PERF factor as low as -2.0% for new deliveries with certain airframe/engine combinations.

  62. Brian Anderson says:

    @Andrew,

    Yes, it’s an interesting question. I thought that the PDA number related to a particular engine, and stayed with that engine. i.e. increased progressively through the life of the engine.

    I understand the concept of the FF, or the PERF, in that it includes variabilities in the airframe too. However, it must be indeterminate on day one, or at least until sufficient cruise data, or data related to a typical flight profile, is collected. Then if an engine is changed out the data collection must begin again.

  63. Andrew says:

    @Brian Anderson

    There is a PDA for each engine, used mainly for engineering purposes. The FF/PERF factor is more relevant for estimating fuel burn, because it reflects the difference between the airframe/engine combination and the book figures. The PDA normally increases through the life of the engine, but the FF/PERF factor might go down if engineering work is done on the airframe that reduces drag (eg seal replacement, flight control/door rigging), or there is an engine change that results in a lower PDA.

    The actual FF/PERF factor is ‘indeterminate’ on day one, as you mentioned. For the A330, Airbus has standard factors that it uses at delivery, depending on the engine type. For example, the A330-343 with Trent 772B-60 engines uses a PERF factor of -2.0% at delivery. That factor is adjusted once the aircraft enters service and data is collected using the APM program. The factors are regularly updated to reflect changes in engine/airframe performance, as determined by the APM program. One of our A330s has a PERF factor of -2.7%, and that’s a 15-year old aircraft!

    Some references that might be of interest:
    Getting to grips with aircraft performance monitoring
    Aircraft performance monitoring from flight data
    Airbus Regional Seminar – Aircraft performance monitoring

  64. Victor Iannello says:

    @Andrew, @Brian Anderson: To be clear, the value of 1.5% that we use is the fuel flow (FF) factor, as per the fuel analysis for MH370. We really don’t know the PDAs for the engines. We also don’t know the drag factor, which is used by the FMC to plan and manage the descent. That said, semantics aside, I don’t think that there is a fundamental error in calculating a PDA (FF factor) based on the estimated time of fuel exhaustion, and then comparing that calculated value of PDA (FF factor) to the value that is expected.

  65. DrB says:

    9M-MRO Fuel Flow Factors

    1. A parameter called “PDA” is used by MAS to predict the fuel loaded for each flight. It is used in the MH370 Flight Plan, and its value is +1.5%, which is a very reasonable value for an average of the fuel flows of 9M-MRO’s two engines in cruise. MAS may be using the term “PDA” in a different fashion than other airlines or manufacturers, who want to keep track of engine and airframe efficiencies separately.
    2. As far as I know, there is no other fuel flow efficiency factor used in the MH370 Flight Plan.
    3. I believe the “PDA” value used by MAS in their flight planning is based on actual fuel consumption data for that airframe and those engines. In other words, the 1.5% PDA as used by MAS is for 9M-MRO, not just for the engines. Thus the value used by MAS may actually be a fuel flow factor (as an excess % relative to the Boeing fuel flow tables).
    4. I have engineering data from ATSB from a prior 9M-MRO flight. This includes frequent readings of both fuel flow meters and both fuel quantity gages. In cruise, the ratio of R/L readings is near 1.02, but it does vary slightly between fuel quantity sensors and the fuel flow sensors, both of which have accuracies of about 1%. Using the fuel quantity sensors, the ratio is 1.021. This implies the R engine will flame out about 16 minutes prior to when the L engine would flame out if the L engine fuel flow were not increased substantially during the INOP condition after R engine flame out. This does occur in practice, and it reduces the time difference between actual R and L engine flame outs below 16 minutes.
    5. My estimate of the R engine “PDA” is +2.45% and for the L engine it is +0.45%.
    6. There is no need (or means) for us to separate engine and airframe effects for 9M-MRO, and I have used the (combined) parameter “PDA” (or FFF) in the same way as MAS uses it for flight planning.
    7. Some of the routes we analyze have predicted “PDAs” (or FFFs) substantially different than +1.5%. That indicates errors in the route parameters, assuming the aircraft is being flown in its normal configuration.
    8. Abnormal aircraft configurations are possible which can save (or waste) fuel. One can assume the bleed air is turned off, for instance, which can reduce fuel flows by about 2%. This is equivalent to a reduction in the FFF of 2% (which then gets into negative percentages).
    9. It may be less confusing to call the predicted amount of differential fuel flow a predicted fuel flow factor (FFF) rather than a predicted PDA. I don’t object to doing that. Whatever you call it, the value of that parameter, compared to the +1.5% used by MAS, is a vital part of any route feasibility determination.

  66. Victor Iannello says:

    @DrB: In the “Fuel Analysis” for flight MH370, there is a parameter called “PER FACTOR” and in the “Flight Brief” there is a parameter called “FF FACTOR”. The value of both parameters is P1.5, i.e, +1.5%. Meanwhile, in the FMC, there are two parameters called “DRAG” and “FF” (which get entered on the same line). I don’t see any specific reference to PDA. It would be the FF FACTOR (=PER FACTOR) that gets entered into the FMC.

    I don’t think it is worth the effort to rename the PDA parameter. We all understand what is meant.

  67. DrB says:

    @Victor Iannello,

    You are correct. The MH370 Flight Brief correctly uses the term “FF Factor”.

  68. DrB says:

    @Niels:

    You asked: “What I don’t understand in your fuel flow analysis / PDA estimates (and using this as one of the path selection criteria) is how you estimate the initial (19:41) fuel quantity. Can you please explain?”

    Here is a summary of my fuel/endurance model particulars:

    1. Assume Zero Fuel Weight = 174.369 tonnes.

    2. Start model at 17:06:43 with 43.8 tonnes total from last ACARS report (with 21.973 tonnes in in L and 21.827 in R tanks).

    3. Assume no fuel transfers.

    4. Assume in cruise the R engine fuel flow is 2.1% greater than the L engine fuel flow (based on my analysis of prior flights).

    5. My fuel model has three periods.

    6. Same PDA is used for all 3 periods, and it is selected for MEFE at 00:17:30.

    7. First period assumes ECON with CI = 52 at FL350 from 17:07 to 17:33.

    8. Second period assumes LRC at FL390. Climb to FL390 starts at 17:33.

    9. Third period starts at 18:34 and continues until MEFE. Speed and altitude during the third period are the same as what is being fitted (from 19:41-00:11). Climb or descent, if needed, occurs at 18:34.

    10. For the BEDAX LNAV 180 degree route, my predicted fuel weight at 19:41 is 26.8 tonnes.

    11. I compute the L and R engine fuel flows and the L and R tank fuel quantities each minute. The R tank goes dry circa 00:08. I estimate there is about 580 kg of fuel in the L tank when the R tank goes dry. Then the L engine fuel flow increases to roughly 3500 kg/hr due to single-engine INOP, and increases further during the descent, and the L engine flames out at 00:17:30. So, I estimate there is about a 9.5 minute difference in flame-out times. The aircraft decelerates beginning at 00:08 during INOP, and I predict the descent begins circa 00:14.

  69. Victor Iannello says:

    @DrB: There is another possibility related to fuel exhaustion: If either or both of the forward and aft crossfeed valves are open, both engines may draw fuel from either tank, and both engines will flameout at almost the same time without any manual intervention. Andrew once explained that the pumps from one side will dominate over the other side, and so the draw down of the tanks might not occur symmetrically. On the other hand, if the level on one side is higher than the other, that will tend to increase the pressure from the pumps drawing from that side, and could help to maintain nearly equal levels on both sides.

  70. Andrew says:

    @Victor
    @DrB

    You could call the performance degradation factor anything you please. I don’t think it matters as long as the meaning is clear. I was only trying to explain that the ‘factor’ has engine and airframe components; it is not a simple engine degradation factor, as I have seen mentioned from time to time.

  71. Richard says:

    @Niels

    You stated “What I don’t understand in your fuel flow analysis / PDA estimates (and using this as one of the path selection criteria) is how you estimate the initial (19:41) fuel quantity. Can you please explain?”

    The calculation of the fuel quantity remaining is based on a number of assumptions:

    (1) The Zero Fuel Weight is correctly stated in the Malaysian SIR as 174,369 kg.

    (2) The Departure Fuel Weight is correctly stated in the Malaysian SIR as 49,700 kg, left tank 24,900 kg and right tank 24,800 kg. All values ± 50 kg.

    (3) The ACARS Fuel Weight at 17:06:43 UTC is 43,800 kg ± 50 kg is correct at reaching the cruise altitude of 35,000 feet.

    (4) There were no fuel transfers between the tanks.

    (5) There was a climb following diversion to FL390 followed by a change to LRC speed mode.

    (6) The right engine fuel consumption is 2.1% greater than the left engine fuel consumption.

    (7) There was no load shedding, IDG switched off or Air Bleed System switched off.

    (8) The temperature deviation from ISA is correctly calculated from the GDAS data.

    At 18:28:06 UTC Bobby calculates 34,571 kg fuel remaining, I calculate 34,490 kg. This is a difference of 81 kg or 0.23%.

    We are both quite different from the Boeing analysis which shows 33,524 kg. This is a difference of 1,047 kg from Bobby’s result or 3.03%. Boeing assumes a descent to FL300, where at 210,000 kg aircraft weight the fuel flow/hr/engine increases from 3,277 kg/hr/engine to 3,762 kg/hr/engine.

    At 19:41:03 UTC Bobby calculates 26,796 kg fuel remaining, I calculate 26,709 kg. This is a difference of 87 kg or 0.32%.

    We both align within 0.5% at 18:28:06 UTC and 19:41:03 UTC.

  72. Richard says:

    @DrB

    My apologies! I did not see your comment before writing mine. Our comments to @Niels crossed in the wires.

    @Niels

    You can choose which answer you prefer 😉 The nice problem of writing “@Richard, DrB”.

  73. paul smithson says:

    @Dr B. Thanks for these additional insights on the fuel model background. I understand that your assumption for one engine inop is that the running one goes to max climb thrust per Mike E and Andrew’s reports from simulations? Would you care to speculate how much longer left engine might run if company policy set max cruise rather than max climb thrust as the default for one engine operation?

  74. DrB says:

    @paul smithson,

    You asked: “Would you care to speculate how much longer left engine might run if company policy set max cruise rather than max climb thrust as the default for one engine operation?”

    I am happy to do that calculation if someone can provide the max cruise thrust.

  75. Niels says:

    @DrB, Richard
    Many thanks for sharing the details of fuel calculations. It is good that your estimates align well, however my main worry remains, which is the unknown flight path between 18:28 and 19:41 with possible altitude and speed changes, hence with possible significant impact on 19:41 fuel quantity.
    For example, would it be possible to work with an estimated 19:41 min / max value, and compare the “back calculated” fuel quantity based on fuel exhaustion and ~1.5% “PDA” with this min – max 19:41 range?

  76. Rob Moss says:

    Just for a laugh, can we please run the CAPTIO flight path through this model to see how far out it is?

  77. Rob Moss says:

    On a more serious note, is the simulation software, or at least the algorithms used to create it, going to be made publicly available? Could it be a GPL project on GitHub?

  78. DennisW says:

    @Rob

    Just for a laugh, can we please run the CAPTIO flight path through this model to see how far out it is?

    The CAPTIO flight path is not a candidate path consistant with the assumptions baked into the model. It has course, speed, and altitude changes.

  79. paul smithson says:

    Thanks, Dr B. Unfortunately I am not able to tell you what the max cruise thrust value is. Hopefully another blog contributor may be able to.

  80. Andrew says:

    @DrB
    @PaulSmithson

    RE: “I am happy to do that calculation if someone can provide the max cruise thrust.”

    The figures are not published in the manuals used for everyday operations. Max cruise thrust is not a ‘certified’ rating and is not used for the calculation of aircraft performance. You would probably only find the figures in some obscure manual used by the performance engineers.

    The thrust limit (CLB or CRZ) used during all-engine cruise can be modified by the airline via the FMC Airline Policy page; however, the default value is CLB thrust. The FMC only uses the CRZ thrust limit if the airline decides to modify the default settings, or the crew manually selects the CRZ thrust limit after reaching top-of-climb.

  81. paul smithson says:

    @Andrew. Thanks. Are you saying that the thrust limit for one engine inop is the same as that which has been set for all-engine cruise? And only if that default (all engine cruise) has been changed would we see something different when first engine goes down?

    I don’t know whether a “thrifty” operator might prefer to reduce default max thrust available for all engine cruise? With very powerful engines and plenty of surplus thrust, would there be any *practical* performance downside to doing so?

  82. Andrew says:

    @paul smithson

    The thrust limit does not automatically change if an engine fails in the cruise. The pilots would normally activate ENG OUT on the FMC VNAV CRZ page, which changes the thrust limit to CON to provide additional thrust. Otherwise, the thrust limit remains at the setting on the airline policy page (normally CLB, but modifiable to CRZ). Sorry I didn’t make that more clear.

    The downside of modifying the cruise thrust limit to CRZ is that less thrust will initially be available if an engine fails in the cruise. Consequently, the speed will decay more rapidly, particularly at altitudes close to MAX, which could have nasty consequences if the crew becomes distracted and doesn’t immediately activate ENG OUT and start descending. If the cruise thrust limit is CLB, there’s a bit more thrust available, which gives the crew a bit more time to analyse the problem before they activate ENG OUT and start descending. That said, the difference isn’t huge.

    As far as I’m aware, the main purpose of the CRZ thrust limit relates to performance guarantees that are part of the contractual arrangements between Boeing and the aircraft operator. It has little or no use in everyday operations.

  83. paul smithson says:

    thanks, @andrew.

  84. George Tilton says:

    @DrB
    RE: “I am happy to do that calculation if someone can provide the max cruise thrust.”

    Rolls-Royce MCT: 312.3 or 343.3kN depending on engine model…

    Try pg-8 at the following link:
    https://www.easa.europa.eu/sites/default/files/dfu/EASA%20E%20047%20TCDS%20issue%2004.pdf

  85. Richard says:

    @George Tilton

    Many thanks for the MCT data!

    In the MH370 case, the engine is the Rolls Royce RB211 Trent 892-17.

    The MCT is therefore 343.3 kN from the linked document you supplied.

  86. Victor Iannello says:

    @Paul Smithson: The maximum thrust at altitude can be approximated by the sea level thrust multiplied by (Pamb/Po). For instance, at FL350, the altitude correction factor is 0.235.

    However, I’m missing the gist of this discussion. As @Andrew said, the difference between Maximum Continuous Thrust and Maximum Climb Thrust is small, and becomes relevant under conditions such as obstacle clearance with one engine inoperative. For instance, in the PMDG 777-200LR model with GE90-110B1L2 engines, at cruise altitude, the values for N1 for CRZ, CLB, and CON thrust limits are 97.4%, 101.0%, and 101.0%, respectively.

  87. Andrew says:

    @George Tilton
    @Richard

    RE: ”Rolls-Royce MCT: 312.3 or 343.3kN depending on engine model…”

    Note, the ‘MCT’ data that George provided is Max Continuous Thrust, which is a higher rating than Max Cruise Thrust. @paul smithson’s question related to Maximum Cruise Thrust.

  88. Andrew says:

    @Victor

    The discussion with @paul smithson relates to the thrust rating set by the FMC during cruise. That rating is dictated by the setting on the FMC’s AIRLINE POLICY page. It defaults to CLB thrust, but can be modified to CRZ thrust by the airline. @paul smithson originally asked @DrB ”Would you care to speculate how much longer left engine might run if company policy set max cruise rather than max climb thrust as the default for one engine operation?”.

  89. Victor Iannello says:

    @Andrew: My point is we are down in the weeds. With all the other unknowns (such as having an accurate drag model for the deceleration and having accurately modeled the fuel consumption from 17:07 to the first flameout), for the short amount of time between the first and second flameouts, it makes no material difference whether the thrust was CLB, CRZ, or CON.

  90. DrB says:

    @paul smithson,

    If I understand Andrew correctly, if no pilot action were taken upon R engine flame-out, the L engine would normally go to Max Climb Thrust (CLB), unless MAS had previously changed the default to Max Cruise Thrust (CRZ). If a pilot subsequently activated ENG OUT, then the thrust limit would change to Max Continuous Thrust (CON).

    If I understand the engine data sheet correctly, CON is about 17% less than CLB, the CON fuel flow would be reduced by a similar percentage and the duration between R and L engine flame-outs would be lengthened by that same percentage. This assumes a pilot activated ENG OUT.

    It’s not clear to me how much less CRZ is than CON for this engine, but a lower thrust limit reduces INOP fuel flow and lengthens the INOP duration.

  91. Victor Iannello says:

    @DrB said: If I understand the engine data sheet correctly, CON is about 17% less than CLB

    No. Maximum continuous thrust (CON) is slighter HIGHER than maximum climb thrust (CLB) (when there is a difference at all). Meanwhile, maximum takeoff thrust is significantly higher than maximum continuous thrust, as it has a 5-minute rating. Maximum cruise thrust (which is not a parameter that is certified) is several percent less than maximum climb thrust.

    Also, although the FMC might default to CLB during cruise by company policy, it is trivial for the crew to change the limit from CLB to CRZ (or CON).

  92. Andrew says:

    @Victor

    RE: “My point is we are down in the weeds.”

    Point taken. My post was intended to clarify @paul smithson’s question, which was related to the difference between CLB and CON thrust (you stated “…the difference between Maximum Continuous Thrust and Maximum Climb Thrust is small”). Nevertheless, I agree; the difference between the various thrust limits is small and immaterial given all the other uncertainties. My guess is the difference between the CLB and CRZ thrust fuel flows would amount to about a 30-second difference in the L engine flameout time, given @DrB’s estimated 580 kg of fuel in the L tank when the R tank goes dry.

  93. Victor Iannello says:

    @Andrew: Thank you so much for your input and your continued interest in the blog.

  94. Andrew says:

    @Victor

    You’re welcome. I just spotted an error in my previous post. I should have said @paul smithson’s question was related to the difference between CLB and CRZ thrust!

  95. paul smithson says:

    @Richard, @Victor, @Dr B. Some of you requested that I share precise path model details. I was not in a position to do so earlier as I have not had time to refine the models during the past 3 weeks. This path model that arises from my work on where (in BTO terms) an FMT “ought” to have occurred, so it starts from there rather from the 1941 ping ring.

    I have noodled around this area with various path solutions and this is the best that I can do. The start position is very close (21 microseconds) to where my models “expect” it to be.

    The parameters are:

    Start 18:32:30Z
    Long 95.252 Lat 7.113
    LNAV
    Initial track angle 190.55
    FL 340 (but geo height 35000ft for purpose of BTO calcs)
    Constant M0.8455 (initial TAS 499kts)

    The model that I am using yields an end point of -39.54, 85.22 (this is the 7th arc crossing ie BTO=18390, rather than position at 001930). BTO residuals are: 8,7,7,-50,11 and BFO residuals are -2.1, -4.3, 4.4, 4.9, 5.4, 5.6 (the last BTO, ping 6, assumes descent of 200fpm).

    Perhaps you can get an even better fit with different altitudes and more finely interpolated Wx data. I have simply used the weather array in Barry Martin’s model (0000Z, 250hPa). I did also try using Wx for higher altitudes but I think the fit was poorer and the required M number higher.

    I will be particularly interested in what your various fuel models predict for this path.

    I have focused primarily on LNAV path solutions to date because TT solutions a) have a 7th arc crossing further north, within the searched area – typically between 38.5S and 39.0S b) invariably have large residuals at ping 6 – requiring a massive slow-down. I say “massive” because it appears to exceed what looks feasible from a combination of greater headwind (or lower temp) over the last stage and longer period with one engine inop before second flameout (which was why I was interested in the Max CRZ vs CLB numbers).

    There is, of course, a solution with initial track angle of 191.0 that terminates marginally further south. But as well as requiring slightly higher speed, I have various reasons for preferring the solution mentioned above.

  96. paul smithson says:

    FYI, the 191.0 LNAV path would be something close to:-

    Start 18:31:45
    Long 95.347, Lat 7.066
    M0.8478 (initial TAS 500.5)
    Crosses 7th arc at 39.72S

    BTO residuals: 15,7,4,-54,10
    BFO residuals -2.2 (1840), -4.3,4.4,5.1,5.7,5.8 (last one descending 200fpm)

  97. Ulric says:

    I’m very pleased to see that the CMH 180 path to 34.2S is still a strong contender after all the learned investigation that has gone into this. It has long seemed the most likely, to me at least. The commentary here is generally of very good quality and still fascinating even after so much time has elapsed since the events themselves. I just wish I had the skills and knowledge to contribute more.

  98. Hank says:

    @Victor

    Your recent paper is very good. I was thinking about the difference between your approach and the particle filter model used by DTSG 5 years ago.

    It seems their particle filter generates “infinite” flight paths based on their defined maneuvering model and ranks them using the BTO/BFO data to create the PDF centered at 38 S around arc 7.

    It seems that you define each candidate profile based on selecting autopilot navigation mode, speed mode, and altitude mode and rate it based on the BTO/BFO data. The 34.4 S crossing on arc 7 provides the best fit.

    So for the case of an inactive pilot, the “infinite” random flights and large sample of reasonable routes confirms an arc 7 search of 34.4 to 38 S.

    This appears to be a very good independent confirmation of the DSTG work for an inactive pilot.

    I missed the discussion where it was determined that there could be no active pilot after 19:41.

  99. DrB says:

    @Ulric,

    The 19:41-00:11 route which appears most likely is either LNAV (geodesic) or a Constant True Track at 180 degrees. No possible Constant Magnetic Heading routes have ever been identified as a region of Interest.

    @paul smithson,

    There is no need to assume a slight descent at 00:11 in order to match the BFO then. To my knowledge, no end-of-flight model predicts a descent beginning before or at 00:11. Both my end-of-flight model and Boeing’s simulations predict the descent beginning several minutes after 00:11.

    @Hank,

    There is no “determination” (or proof) about the presence or absence of active piloting after 19:41. However, it is quite striking that there is one and only one route for which no active piloting after 19:41 is needed in order to completely match the BTOs, BFOs, their known correlations, and the endurance. In my opinion, the presence of a unique solution is a likely indicator of correctness. At the very least, it is the most likely place to search next. If that fails, then, absent new revelations, searching elsewhere is not worthwhile because the number of possible solutions with maneuvers after 19:41 is boundless.

  100. David says:

    @DrB. The “only one route” you mention would be consistent with there being an active pilot who intervened only at fuel exhaustion, pushing the nose down into a steep final dive, APU start leading to the final transmissions.

    This possibility by-passes non-replication of the timing/descent combinations of those transmissions by those unmanned Boeing simulations. Thus adding this to the unmanned possibility should enhance the search prospects of that alone. Also it can explain why there was insufficient time for IFE set-up, a piloted glide needing another explanation for that.

    Thus I think that its inclusion would strengthen the case for a search based on that route.

  101. Ulric says:

    @Dr B

    Sorry, I meant the CMT 181 route identified in the table at the top of the article. You’re quite right and that will teach me to post when I’m bleary eyed.

  102. paul smithson says:

    Thanks for your comments, Dr B. Without descent the BFO at 0011 would be 9.2Hz, which looks high – particularly in comparison to the values at ping 4 and 5.

    If it is not too much trouble – and at your convenience, I would be interested to see how he BTO/BFO/PDA for the 190.55 model referred above come out using your path modelling tool.

  103. Richard says:

    @Paul Smithson

    You stated: “@Richard, @Victor, @Dr B. Some of you requested that I share precise path model details.”

    Start 18:32:30Z
    Long 95.252 Lat 7.113
    LNAV
    Initial track angle 190.55
    FL 340 (but geo height 35000ft for purpose of BTO calcs)
    Constant M0.8455 (initial TAS 499kts)

    I will run this through my model as soon as possible.

    I am currently busy with re-running the flight paths in the ROI table (Figure 8), where you kindly pointed out an error.

  104. paul smithson says:

    Thank you, @Richard – much obliged.

  105. Victor Iannello says:

    @David: I don’t put a lot of weight on the Boeing (no pilot input) simulations failing to reproduce the accelerating rate of descent 2 minutes after fuel exhaustion. We know that a banked descent would achieve this level of downward acceleration, so the question becomes why it took 2 minutes to get to this aerodynamic state.

    I question whether the Boeing simulator accurately represents the interaction of so many complex systems that are operating beyond their rated conditions and in abnormal configurations. We also know that engine re-starts were not modeled. (In the Level D simulations that Mike Exner witnessed, an engine restart resulted in a significant roll rate.) The restart of the left engine might also reduce the available fuel for the APU, which would explain why there was no IFE log-on.

    That said, it would be unwise to completely dismiss the possibility of pilot inputs after flameout. I fully acknowledge there is a finite possibility it occurred.

    With all we know today, my best guess is the debris field was missed near 34.3S latitude, but I’d still put that guess at under 50%.

  106. DennisW says:

    @DrB

    In my opinion, the presence of a unique solution is a likely indicator of correctness.

    Well…some mitigating feedback:

    1> You picked the boundary conditions for the “unique solution” analytics.

    2> You chose to cherry pick the data and ignore Figure 5.4 in the DSTG book.

    3> Your terminus has been searched with a negative result.

  107. Hank says:

    @DrB

    I only questioned the rationale for not allowing any pilot maneuvering after 19:41 and I guess there is no basis other than it simplifies the analyses. That is OK. This is all great work. I intend no criticism.

    Your statement that “the number of possible solutions with maneuvers after 19:41 is boundless” contradicts the “unique solution is a likely indicator of correctness” statement.

    Suppose we learn someday after MH370 is found that it crossed arc 7 at 24 S at 00:11 with fuel depleted and then entered the ocean.

    Then would it be possible for someone to define one flight path with pilot engagement out of the infinite possibilities that meets all of the arc/BTO/BFO constraints. Of course the FDR would help.

  108. Victor Iannello says:

    @DennisW: I don’t disagree in principle with your three comments. The BEDAX-SouthPole route is not a slam-dunk. However, it may be the best candidate.

  109. DrB says:

    @David,

    You said: “@DrB. The “only one route” you mention would be consistent with there being an active pilot who intervened only at fuel exhaustion, pushing the nose down into a steep final dive, APU start leading to the final transmissions.

    This possibility by-passes non-replication of the timing/descent combinations of those transmissions by those unmanned Boeing simulations. Thus adding this to the unmanned possibility should enhance the search prospects of that alone. Also it can explain why there was insufficient time for IFE set-up, a piloted glide needing another explanation for that.

    Thus I think that its inclusion would strengthen the case for a search based on that route.”

    You are correct, and my coauthors and I have consistently made that point in previous comments. My personal search recommendations include redoing the search inside 25 NM first and then moving outside 25 NM if that is unsuccessful.

  110. DrB says:

    @Hank,

    You said: “I only questioned the rationale for not allowing any pilot maneuvering after 19:41 and I guess there is no basis other than it simplifies the analyses.”

    Actually, that is the only way to conduct a systematic search that will lead to one or a few solutions. Technically, our assumption is no pilot maneuvers between 19:41 and 00:11.

    You also said: “Your statement that “the number of possible solutions with maneuvers after 19:41 is boundless” contradicts the “unique solution is a likely indicator of correctness” statement.”

    Everyone has an opinion. My opinion is that the chances that any actual route with significant maneuvers after 19:41 would have a unique solution assuming no maneuvers is extremely low.

    You also said: “Suppose we learn someday after MH370 is found that it crossed arc 7 at 24 S at 00:11 with fuel depleted and then entered the ocean. Then would it be possible for someone to define one flight path with pilot engagement out of the infinite possibilities that meets all of the arc/BTO/BFO constraints. Of course the FDR would help.”

    Actually Arc 7 is at 00:19. In general, there would be multiple solutions derivable from the satellite data plus an end point. FDR data for the entire flight would be required to fix the actual route.

  111. DrB says:

    @Dennis,

    You said: “1> You picked the boundary conditions for the “unique solution” analytics.

    2> You chose to cherry pick the data and ignore Figure 5.4 in the DSTG book.

    3> Your terminus has been searched with a negative result.”

    Regarding #1, our assumptions are essentially only 2: (1) the route is flyable on 9M-MRO’s autopilot, and (2) no pilot maneuvers were performed between 19:41 and 00:11. The assumptions we used have been around since 2014. The conditions on the BTO and BFO statistics and correlations are not assumptions. They are observed and inferred characteristics. I have no clue what you mean about picking “boundary conditions”.

    Regarding #2, @airlandseaman provided significant and new information regarding Figure 5.4, which you continue to ignore. Please go back and re-read his comment. Also, as I have said repeatedly, removing the BFO statistic does not change the outcome, so your criticism is irrelevant.

    Regarding #3, yes, as we have said repeatedly, 34.3S has been searched out to 25 NM. Too bad you did not follow Ocean Infinity’s search for ARA San Juan. You would have learned something useful. No matter how good the equipment, mistakes occur wherein the debris field may be missed. People are in the classification loop, and in some places the terrain is difficult. No unsuccessful sea bottom search is 100% effective, only the successful ones are. In addition, the aircraft could be outside the previously searched area.

  112. David says:

    @Victor. Thanks.
    From our extensive earlier discussions about the gap between simulation results and the descents derived from the final BFOs, my own take is that while a pilot initiated final dive would bridge that, simulation shortfalls just might explain it.

    @DrB. Thanks for correcting my misunderstanding that the modelling supposed the aircraft to be unpiloted.

  113. Hank says:

    @DrB

    I believe you noted that there could be several flight path solutions by which an active pilot could meet all of the Arc/BTO/BFO constraints and the aircraft would cross Arc 7 at 00:19 at 24 S with fuel depleted. This is a much more complex problem than finding the best fit A/P modes, so I appreciate the focus on no active pilot. That does not mean that there was no active pilot – that would be another case which you dismiss as not very likely.

    Maybe Z programmed the A/P to fly to the South Pole. But he would know that Boeing could assume that and the endurance problem and range problem are merged. Without any consideration of SATCOM arcs, you could estimate the fuel depletion location on the route. It’s great that this also matches the BFO/BTO data. So that supports that the aircraft may have actually done it.

    But Z would know this and would know that by flying on a different course and intentionally making an occasional clearing turns or S-turns he could burn off an hour of fuel without making forward progress. It makes sense to not use direct 180 – the arc orientation would suggest any feasible solution with delays needed to be more southeast. But he would not know about arcs, so the magic path through the arcs would just be a lucky accident.

    So does Z set it up for the South Pole and enjoy the ride with no consideration of post crash simulations or does he intentionally mess with the range-endurance modelling? If the plane is at 34.3 S he went for the ride. If the plane is above 25 S he is messing with everyone.

  114. Richard says:

    @All

    I have run 14 MH370 flight paths with active pilot inputs and posted the results on Victor’s website in either the current post or the previous post.

    These flight paths with active pilot inputs have been proposed by @TBill, @Nederland, @Niels and @Hank.

    Only Niels Case C came within the tolerances set for a ROI, but this case only starts at 21:11:02 UTC and is not a full simulation from before 19:41:03UTC.

    Before anybody starts criticising a LNAV 180 LRC FL390 path without an active pilot, because there are obviously better flight paths with an active pilot, they should please describe such a flight path. I need only start time, start lat/lon, initial bearing, flight level, navigation mode, speed mode and the time(s) of any bearing, flight level, navigation mode or speed mode change.

    So far, a LNAV 180 LRC FL390 flight path is unique.

  115. TBill says:

    @Richard
    Right now I am not convinced of passive flight, but I am not going to put up any more argument here. I would rather search your area, since you could be correct, than my area, since I could be wrong. And I would feel bad if I caused the search to miss an opportunity. If your recommednation comes up empty, I got a back-up proposal where it went, and other people have other unrequited hunches such as Ed Baker for example.

    OK I may still put up some argument. But I have been relatively quiet letting IG take the lead since late March. That’s when Don hit me pretty hard on Reddit for suggesting a 20-25 South search area – I immediately asked Victor if there was change in the thinking and Victor said yes that was the new trend.

  116. Shadynuk says:

    @Hank It is of course very difficult to get into someone’s head – especially someone who would do this. And, we are not certain who’s head to get into.

    Nevertheless, the analytical approach has not produced any wreckage and any new ‘hotspots’ that may be derived analytically would be strengthened if supported by some ‘motivational rationale’. So, this is my attempt get into someone’s head.

    I cannot recall an instance where an aircraft crashed into the sea, a timely search was initiated and the wreckage was found on the bottom without first finding a floating debris field and/or the 37.5 KHz pinger signal.

    In the MH370 case neither of those has been located. The reasoning could have been that if those indicators are hidden, the wreckage would not be found. Indeed, if neither of those indicators could be found, the reasoning could have been that a sea bottom search would not be conducted. Of course, it was conducted – but it surprised many that it was – and it was not successful.

    Perhaps the best way to hide these indicators is to crash into the sea as far as possible from an operational search base. It seems to me that the latest ‘hotspot’ (LNAV 180 LRC FL390) can be ‘explained’ on a motivational basis and has analytical integrity.

  117. Tom O'Flaherty says:

    As always, this makes for engaging reading.

    I’m beginning to suspect more and more the final resting site has been scanned but missed.

  118. DennisW says:

    @Richard

    So far, a LNAV 180 LRC FL390 flight path is unique.

    I disagree. There are many possible paths satisfying the Inmarsat data. As I have said several time before, the Inmarsat data cannot be used to determine a terminus. We have seen that movie many times, and it is yet to have a happy ending.

  119. Richard says:

    @DennisW

    It does help that you disagree, without providing evidence as to why.

    To simply state there are many flight paths that satisfy the Inmarsat data, is cheap and insulting. It shows you have not bothered to read our paper or recent comments on Victor’s website:

    (1) Our current study is not just reliant on the Inmarsat data. We are also looking at the aircraft performance, engine performance and weather data in detail.

    (2) You fail to provide any alternative flight paths, despite my challenge to the MH370 community to nominate such paths. To repeat, I need only start time, start lat/lon, initial bearing, flight level, navigation mode, speed mode and the time(s) of any bearing, flight level, navigation mode or speed mode change.

  120. DennisW says:

    @Richard

    The original work done by the IG is valid, IMO. Victor’s path to the Cocos works for me as well. Your characterization of BFO errors simply does not stand up to either the previous data in the DSTG book or the industry standard math used to characterize oscillator performance. Your definition of “unique” relies on your interpretation of the data and the statistics governing that data.

    As far a navigation mode is concerned, it is possible that inputs from an active pilot make that question irrelevant.

    It would be extremely useful to have the equivalent of the DSTG book figure 5.4 for several other previous flights. The largest component BFO error is due to the random walk of the AES oscillator, and it simply cannot be characterized by mean and standard variance.

  121. Victor Iannello says:

    @DennisW said: It would be extremely useful to have the equivalent of the DSTG book figure 5.4 for several other previous flights.

    Agreed, especially if the other flights showed evidence of drift of the fixed frequency bias.

    The largest component BFO error is due to the random walk of the AES oscillator, and it simply cannot be characterized by mean and standard variance.

    If the first part of the statement is true, so is the second. I don’t think we can be sure of the first part.

    Bobby believes that we might be able to ignore the BFO statistics and arrive at the same unique path. That would be progress because it renders assumptions about the bias drift to be moot.

  122. Peter Norton says:

    DrB: “My opinion is that the chances that any actual route with significant maneuvers after 19:41 would have a unique solution assuming no maneuvers is extremely low.”

    Is there any way to determine or at least estimate/approximate mathematically this probability?

    Prima facie the probability indeed seems low.
    But doesn’t the fact that the amount of possible paths is infinite make it less improbable than it appears at first glance? (The more paths, the likelier they are to include a random match by chance, no? So if you have infinite path candidates … ?).

  123. DrB says:

    @Peter Norton,

    You said: “Is there any way to determine or at least estimate/approximate mathematically this probability?”, referring to ” the chances that any actual route with significant maneuvers after 19:41 would have a unique solution assuming no maneuvers”.

    The only way I can see to do this would be to create a number of artificial data sets from assumed routes with maneuvers. Then one can search a data set for a matching route without maneuvers. The problem is, it takes weeks to months to do a good search of a single trial set of satellite data, looking for a route with no maneuvers. Then one would have to repeat this many times, leading to many years of effort (using PCs). Even then, some might say you just didn’t pick the right route with maneuvers. So, from a practical point of view, I don’t see how to predict this probability with any useful degree of certainty.

    Here’s one way to look at it. If one takes a data set generated by a route with no maneuvers, one can always replicate that data set using a route with multiple maneuvers. But the reverse is not universally true. For instance, I can easily generate a route with maneuvers which cannot be replicated by a route with no maneuvers. All you have to do is throw in a HOLD, a large speed change, or a significant turn. In other words, a data set from a route with no maneuvers can be matched by a route with maneuvers, but most routes with maneuvers cannot be matched by a route with no maneuvers.

  124. Peter Norton says:

    @DrB: thank you for the kind explanation

    Ok, given that it seems next to impossible to determine this probability mathematically, how would you estimate this probability (of a coincidental match) is? What percentage?

    I understand that the question is both subjective and hard to answer, but what is your educated guess based on your experience (all your excellent work on this subject so far) ?

  125. Hank says:

    @Shadynuk

    I believe that Captain Zaharie (Z) hijacked his own aircraft and flew it to a point on Arc 2 west of Sumatra at 19:41. It is clearly possible that Z programmed the autopilot to fly LNAV 180 LRC FL390 and then disengaged. I have no issue with the great work done by DrB, Richard, and Victor.

    But Z is a very experienced aviator and understands that flying a long straight path until fuel depletion merges the endurance and range problems. If investigators assume a direct flight to the south pole the aircraft would run out of fuel at a predictable point in simulations.

    A much less sophisticated simulation than the one performed here could define a search zone with no use of the SATCOM data if you assume LNAV 180.

    If Z wanted to mess with the ability to perform these fixed course simulation he would burn fuel and not make any forward progress by performing a hold or just some S-turns. If he delays by holding for one hour at cruise Mach between 19:41 and 00:11 (4:30) he takes 500 or so miles out of the range.

    He could have done this on a 180 path, but that would have not matched any of the Arc parameters. If Z performed holding he would have had to have flown a more southeast route.

    The issue is why did Z program the autopilot for LNAV 180 LRC FL390 or did he mess with post incident simulations by intentionally engaging in some delay tactics? If he did delay he had no knowledge of arcs and pings and it would need to be a magic sequence of turns. But there are infinite ways for an active pilot to delay and meet all of the SATCOM constraints, but there is no question that the delays would move the Arc 7 crossing to the north from the 34.4 perfect course.

  126. Victor Iannello says:

    @Peter Norton: You are really asking for a proof of Occam’s Razor. Here’s an interesting analysis that attempts to show that more complex solutions have lower probability:

    https://www.lesswrong.com/posts/jBnX58P8J6HuTCz2g/a-proof-of-occam-s-razor

    In the DSTG study, they reported a histogram of the number of turns in the posterior distribution (Fig 10.4). It’s quite evident that as the number of turns increases, there is a much lower frequency of paths satisfying the BTO and BFO criteria.

  127. Hank says:

    @Richard

    I have no reason to criticize your work. It is very well done and it could have been exactly what Z programmed.

    You said” – Before anybody starts criticising a LNAV 180 LRC FL390 path without an active pilot, because there are obviously better flight paths with an active pilot, they should please describe such a flight path.

    I don’t have the tools to perform this analysis at home. If I did, I would try to back fit one feasible active path to at least show that it is possible.

    I would place the aircraft on Arc 7 at 00:19 at 24.0 S and set the parameters to exactly fit the SATCOM data. I would connect the Arc 7 24.0 S intercept with an assumed start position on Arc 2 at 19:41 to define a nominal great circle route. I would define the approximate time delay needed to fly between Arc 2 and Arc 6 over 4:30 hours at a cruise Mach that is consistent with the SATCOM data. I would apportion the delay between the three 1:00 hour 2-3, 3-4, and 4-5 segments and the one 1:30 5-6 segment. The aircraft arc crossing angle could vary based on whether the aircraft is engaged in an S-turn so the exact crossing angle can vary from a straight arc crossing on the reference course. It may be necessary to shift delay time between the zones to best fit the data.

    Z is could not be gaming the system in a sophisticated way because he would not know about the arcs. But there are infinite maneuver possibilities. So simple flight maneuvers are better. There could be a long circle in one spot with S-turns in other areas.

    If it is feasible to create ONE feasible flight path to 24S, it should be feasible to create ONE to 23.5S, etc.

    I think Z is smart enough to mess with any post-incident range-endurance simulations and he would not have flown a straight course.

    But it is certainly fair to believe he programmed the LNAV 180 and disengaged.

  128. Shadynuk says:

    @Hank The fact that you (and like others) do not believe that ‘he’ would have flown a straight course may be the very reason that he would fly a straight course.

    In any case, if you can conceal the floating debris and the 37.5 KHz pinger signal, you certainly make life a lot more difficult for the searchers. However, realistically, I have no idea where this wreckage could be. I continue to count the number of assumptions necessary to produce an unambiguous analytical solution – there are a lot of them.

  129. Andrew says:

    @Hank

    RE: “Z is could not be gaming the system in a sophisticated way because he would not know about the arcs. “

    I think the question that needs to be asked is, having essentially ‘disappeared’ early in the flight, would the perpetrator feel any need to further ‘game’ the system? He might reasonably have suspected the aircraft could still be tracked by primary radar after it turned back and headed up the Strait of Malacca, but he would have known the aircraft would eventually be ‘lost’ once it left radar coverage. From his perspective, nobody would have had a clue where the aircraft went after that point. It might have continued on the same track, or it might have turned, but where and in what direction? From that point on he must have felt ‘safe’ from detection, so why would he feel the need to cause further obfuscation?

  130. mash says:

    re: passive flight path + active EOF (“push down”)

    The above combination/contradiction could also be used to substantiate the hypothesis of an interchange [if not entirely impossible]: a passive flight ‘simulation’, no fuel exhaustion, a fake dive, no crash, and so on ( – just in case only! ) …

    [But still hope it is not another ‘ether’ story. ]

  131. Victor Iannello says:

    @Hank: Even if we somehow “knew” that the pilot followed a due south course, we would still need the satellite data to find the plane: we wouldn’t know the longitude because we wouldn’t know at point flying northwest the turn to the south occurred, and we couldn’t use the fuel models to predict the latitude because a crash site near 34.3S latitude requires a loiter or excursion before 19:41, and we wouldn’t know the extent of the delay. Without the satellite data, it would be almost impossible to guess where to search.

  132. Victor Iannello says:

    @Tom O’Flaherty: Welcome to the blog.

  133. Ventus45 says:

    @Hank, @Andrew

    I agree that once he had escaped the range of any possible “land based radar threat” he could quite easily just set a straight course direct to wherever he intended to go.

    But, if we assume his imperative was to “vanish”, then any “chance sighting” by anyone else, by even one, possibly, but definately if by two or more, at any point(s) in time, those sightings would (on later analysis) give his track line. If your intention is to vanish, it is a poor strategy to set a straight course for five hours plus.

    During the hours of darkness, such chance sightings ( by military ships with air search radars, or military patrol aircraft eg P3, Awacks, Hawkeye etc ) would be vanishingly slim, and thus the strategy is acceptable. But during daylight, it is definately a very risky strategy, and not acceptable, in my view. Visual detection in daylight, by other aircraft and/or vessels, would be way too easy, particularly if he was making contrails, and indeed, suspicious contrails would also show up in weather satellite images during daylight.

    This is the biggest issue I have with the northern end points on the arc, since he would have been in daylight for well over an hour, a fact, apparently ignored by most. This is the main reason why I hold to the “stay in the dark” and ditch at dawn as far south as possible strategy, i.e. (~~38S~~).

    But, if you accept northern terminuses, there are multiple possible piloted soultions, including Hank’s strategy.

  134. DrB says:

    @Peter Norton,

    You asked: “Ok, given that it seems next to impossible to determine this probability mathematically, how would you estimate this probability (of a coincidental match) is? What percentage?”

    I can’t put a number on it, but in my opinion it is very low.

    I can tell you roughly what percentage of possible combinations of the 7 route parameters happen to match the satellite data. Here are the rough numbers:

    To make it simple, let’s start the route at 19:41, so we don’t have to guess a start time. Then we have to guess the start latitude and start longitude over a region that is roughly 10 degrees in latitude and 5 degrees in longitude. To fit the data at all, one has to be within less than 1/2 degree, so that is roughly (at least) 200 possibilities. Next I have to guess the flight level, so let’s assume it could be every 500 foot interval from FL340 to FL410. That’s 15 choices. Then I need to pick a speed. Obvious choices are ECON with CI=52, MRC, LRC, M0.84, M0.83, M0.82, Holding, and 250 KCAS. That’s 8 choices. Next is lateral navigation method. Choices are LNAV, CTT, CMT, CTH, and CMH. That’s 5 choices. Finally, one must specify the initial bearing. To get any kind of a fit, one needs to be closer than 1/2 degree over a potential range of roughly 155 to 192 degrees. That’s about 75 possibilities.

    So, a very crude minimal number of “no maneuvers” routes is 200 x 15 x 8 x 5 x 75 = 9 x 10^6. So, that one unique route that matches the satellite data is one out of 9 million “no maneuvers” possibilities. So, if you want to assume the actual route had maneuvers, you must then believe that only one in 9 million “no maneuvers” routes also coincidentally matches the data. What is the probability of that happening by random chance? One in 9 million? Actually it is much lower than that, because many routes with maneuvers have zero “no maneuvers” solutions.

    To summarize, if you assume the actual route had maneuvers, then it must have been one of a very small subset that also has a “no-maneuvers” solution, and furthermore that solution is unique out of roughly 9 million possibilities. The only way to “prove” that maneuvers occurred is to demonstrate that there is no route among the 9 million “no maneuvers” which matches the data. But here we do have one solution, so we can’t “prove” maneuvers occurred, nor can we “prove” that no maneuvers occurred (from the satellite data).

    The alternative hypothesis is that the actual route had no maneuvers. Furthermore, in this case we know only a single route was flown, and therefore there should be a single “no maneuvers” solution, which there is. This is the simplest hypothesis consistent with the observables and therefore it is also the most likely.

  135. TBill says:

    @ventus45
    Those with proposed locations (pins) north of 25 South are fighting some ‘headwinds’:

    (1) Drift analysis of both Chari Pattiaratchi and David Griffin. Both influential oceanographers apparently favor high twenties to mid thirties as a crash location (as Blaine has pointed out).

    (2) Possible harder to search – too deep in some places

    (3) Violation of the Sunlight Avoidance hypothesis, poplular for some (Aussies) is the theory that the pilot might have designed the flight path to stay out of sunlight at sunrise, by going west.

    (4) Violation of the Passive Flight assumption popular for many

    (5) Violation of the Submarine Rule-of-Thumb, the striking success that Ocean Infinity apparently found the missing Argentine Sub in an area already searched implies the MH370 crash location is in an area already serached, but was missed

    So that’s why 20-25 South idea died among influential investigators.

    I am not saying those reasons above are valid, I am only saying they are popular and influential. I am also not giving the rebuttal of the nothern pin holders, who tend to be the active pilot believers.

  136. Victor Iannello says:

    @DrB: I think what would be persuasive is to show that using objective criteria, there is only one passive route that is acceptable, and that one route has a very sharp peak in the objective function. The challenge is to demonstrate that the criteria truly is objective, i.e., the criteria were not chosen in a way that biases a particular result, and that there are no other sharp peaks in the objective function. The latter is hard to prove because sharp peaks are hard to find using a search methodology. I think the current works makes progress on all fronts.

  137. DennisW says:

    @DrB

    To make it simple, let’s start the route at 19:41, so we don’t have to guess a start time. Then we have to guess the start latitude and start longitude over a region that is roughly 10 degrees in latitude and 5 degrees in longitude. To fit the data at all, one has to be within less than 1/2 degree, so that is roughly (at least) 200 possibilities. Next I have to guess the flight level, so let’s assume it could be every 500 foot interval from FL340 to FL410. That’s 15 choices. Then I need to pick a speed. Obvious choices are ECON with CI=52, MRC, LRC, M0.84, M0.83, M0.82, Holding, and 250 KCAS. That’s 8 choices. Next is lateral navigation method. Choices are LNAV, CTT, CMT, CTH, and CMH. That’s 5 choices. Finally, one must specify the initial bearing. To get any kind of a fit, one needs to be closer than 1/2 degree over a potential range of roughly 155 to 192 degrees. That’s about 75 possibilities.

    At 19:41 the aircraft lat and lon are not independent but coupled by the requirement to be on the 19:41 arc. That reduces your factor of 200 to 20.

    Likewise speed and track angle are tightly coupled. Once you pick a track angle the speed required to meet the BTO at 20:41 is determined. I prefer to think of track angle as simply a test case place holder, and not something you would include in the route probability.

    Altitude is not an independent variable either, depending on the lateral navigation mode.

    I would assign the number of “no manuever” routes to be closer to 10,000 than 9 million.

    @TBill

    Your list of “headwinds” is equally troubling. I can make no sense of 2> through 5> at all.

  138. TBill says:

    @DennisW
    Item-2 is conjecture from me, I am thinking there are some areas 19000-feet or deeper on Arc7 that could be hard to seacrh (at 20-25 South +-25 nm)

    Items 3-5 are I think clear, (3) ventus is saying he and others do not buy 20-25 South because the pilot would have been in sunlight for an hour, (4) Passive flight paths end somewhere around 26 South and…

    (5) is a big new thing. Originally we were hearing the prior search was 9x% effective…so we thought looking 20-25 South made sense…now we are hearing it is considered more likely the crash site was missed than being elsewhere

  139. paul smithson says:

    @Victor. I second your sentiments regarding “objective function” and distinct probability peak expressed above. I look forward to seeing a fuller write up of the results and interpretation in due course. If (and it’s a big if) I can follow the stats, perhaps I could be persuaded of the ability of this technique to unearth the “one true course”. As of now I remain skeptical and I continue to be amazed that it is considered possible to extract objective “truth” from among all of that noise.

  140. haxi says:

    @Victor, @DrB:

    If in the end you can find the only one acceptable route through this technique, that’ll be a huge achievement. I’m really looking forward to your upcoming paper.

  141. Victor Iannello says:

    The French investigators seem to be coming to the same conclusions as the rest of us:
    http://www.leparisien.fr/faits-divers/vol-mh370-revelations-sur-le-crash-mystere-10-07-2019-8114283.php

    They visited Boeing and left with the satellite data. They expect it will take a year to analyze the data. (Considering that the unredacted data sets for MH370 and MH371 have been in the public domain for over two years, and people have been analyzing the data from MH370 for the last five years, I’d say it is too little too late.)

  142. Ventus45 says:

    @TBill

    For the sake of covering all bases, I am going to abandon my 38S point (for now), and stick my neck out a bit and make the case for the “totally dark” scenario.

    Consider this.

    Mission Day was set by Anwar’s court appearance date, which was known by all and sundry, months in advance.
    Z knew (as did everyone else) that Anwar would be sent back to gaol.
    Z had long since made up his mind to “immediately” hit back at this legal and political corruption.
    Z had long since decided on vanishing in the SIO, and he had contingency plans ready.
    All that was needed was confirmation of the court date, the verdict (which was a foregone conclusion), and his actual next long haul flight after that.
    Z knew, weeks before, when his roster came out, that he would have MH370 on mission day.
    Thus, he had plenty of time to finalise his true flight plan (which we don’t know – yet) and to generate the sim clues.
    He wanted the Malaysian Government to know (secretly – well after the event) that it was him and that it was deliberate.
    Z knew that sooner or later, the Malaysian Government would seize and examine his simulator.
    So he decided to hide his clues there.
    But, he did not want to make it easy for them.
    We now know how he did that, by recording the flight on a separate drive, and then amateurishly deleting it, deliberately, full well knowing that it could be recovered by computer experts.

    The cryptic clues that Z provided the Malaysian Government (and later us) are contained in plain sight, in the simulator “dragged” end of flight positions S1/S2 (45S104E). The early points in the sim are not relevant to the end game.
    What is significant to the end game are only the end points S1/S2.

    Z was a very smart cookie, and no doubt very well versed in matters navigational and astronomical.
    Z gave 45 South as a freebie, (otherwise no one would ever work it out).
    Z’s determined end of flight latitude is therefore 45S.
    That part is both simple and obvious, even to a corrupt government, with a corrupt police and judiciary.
    But where was he going exactly, what longitude ?
    Be patient.

    Once his roster came out, and he knew he had MH370, he knew that he would need a little extra fuel than was normal for MH370 to actually get to 45S.
    He knew then, that on the night, he would have to conger up a credible looking normal pretext to load that extra fuel that he would need to get to 45S.
    That is why he was a bit early for the flight, and had the fuel signed off, even before Hamid clocked on.
    We know all about that extra fuel now, and he knew that we would, eventually.

    Now, to the flight itself. Z knew, that to remain hidden from visual detection, he had to stay in “the dark”. Keep that thought in mind.

    You must remember that there are two positions S1/S2.
    Both are significant.
    S1 is at cruise altitude.
    S2 is virtually at sea level (for all practical purposes).
    He deliberately “dived” the sim from top to bottom, to figuratively say, “this marks the spot”.
    So, what does this tell us about “the spot” ?

    Let us recap. What do we know at this point ?
    (1) The time of end of flight for mission day, is obviously defined by fuel exhaustion, planned to be at about 00:00 UTC.
    (2) The intended latitude we know, which is 45S.

    What we don’t know, is the longitude.
    But in point of fact we actually do.
    It is in plain sight, if you blink, twice, perhaps thrice.

    To divine the clue, you must think like a navigator, or better yet, an astronomer.
    You must remember that:-
    (a) daybreak, i.e. sunrise, is defined as when the sun is on th horizon, ie, zero degrees to the horizon.
    (b) civil twilight is defined as beginning when the sun is 6 degrees below the horizon.
    (c) nautical twilight is defined as beginning when the sun is 12 degrees below the horizon.
    (d) astronomical twilight is defined as beginning when the sun is 18 degrees below the horizon.

    Now the longitude for sunrise at 45S at 00:00 UTC on mission day the 8th of March 2014 is quite easy to calculate. It is 86.367E.
    But Z did not want to tell us outright that he was planning on going to 86.367E at 45.000S at 00:00UTC.
    He was sneaky, an evil genius, and he had a sense of humour.

    He was entering the twilight zone, but not just any old ordinary twilight zone mind.
    He was entering a very special, a very specific twilight zone, to “vanish” with “no light”, so to speak.
    He did not want civil twilight, with its glow of pre dawn light, nor did he want a nautical twilight, with it’s very faint glow of light.

    No, he was heading to the absolute boundary line between “fully dark” and light of any kind, specifically, “to the limit of night”, i.e. to the beginning of astronomical twilight.

    Now, astronomical twilight at 00:00 UTC at 45S on the 8th of March 2014 would be 18 degrees west of 86.367E, i.e. at 68.367E.
    But, that was no good as “a clue” for a number of reasons.
    Firstly, it would be dismissed by all analysts simply because:-
    (a) it was way too far for the fuel available, and,
    (b) it was way too far for the performance limited range in the time available.

    So, what did he do ?
    He had to pick a point that was theoretically reachable from Penang, and thus credible.
    Penang direct to 45S 86.367E is just under 3,100 Nm
    Penang direct to S1/S2 is just over 3,000 Nm, virtually the same distance, slightly less actually, and thus both achievable and credible.

    So, he simply dragged his simulator to 18 degrees east of his end latitude, i.e. to what should have been precisely 104.367E for S2.

    If all had have gone according to plan, theoretically, he would have reached 45S 86.367E at 00:00 UTC.
    In doing so, he would have crossed the now known 7th arc, heading 187.025 true, not at 00:19, but at 23:26 UTC.

    But that didn’t occur.
    In the event, he was going to be about 53 minutes late, and he was going to come up a bit short on range, though he had got the fuel endurance almost spot on to within 10 to 20 minutes.
    So, what does he do ?

    His imperative remember, is to “stay in the dark”.
    Precisely where he actually ends up does not matter.
    Knowing that he would be late on his plan by the time he crossed the equator (an obvious check point) forced him to recalculate, and head further west to remain in the dark.
    Thus, there must be another slight turn somewhere, probably after he had reached about 20S and had a better handle on GS made good.

    For an estimated extra 20 minutes flight time, to his now re-estimated 00:20 flame out, and since the earth is rotating, he needs to end up at 00:20 UTC a further 5 degrees west of his originally intended track line.

    From 20s90E to 45S 86.367E he would have (in the original plan) been heading about 185.8 true.
    He now has to head from about 20S90E to a recalculated position on the 86.367 – 5 = 81.367E meridian, that is reachable in the time available, and with the fuel remaining.
    That requires a slight turn, to about 197.5 true, and produces an intersection with the 7th arc at about 41S at 00:20 UTC.

    Thus, 41S 81.367E is the maximum south westerly position on the 7th arc that he could have reached.

    That is where to look next, around 41S, and if not found there, backtrack along the 7th arc from 41S to 38S.

    V45

  143. TBill says:

    @ventus45
    I agree with you philosphically, the only real flight path clue we have from ZS is the sim study. Maybe the @Freddie/Xmas rumor as an alternate flight path hint from ZS, but unlcear on the latter.

    I (and Victor) have asked readers here: what end point do we think the sim study suggests? What was the purpose of the run? You have certainly provided a counter-proposal, we’ll have to weigh it. My proposal is he flew the actual route on the simulator and the goal was Broken Ridge or Dordretch Hole. Believe the latter is a good match to BTO/BFO as active flight, merging onto the sim path at around BEBIM.

  144. Andrew says:

    @Ventus45

    RE: “We know all about that extra fuel now…”

    People keep claiming that ‘extra fuel’ was loaded. Would you care to explain why you think the fuel load was more than what was required for the flight that night?

  145. Mick Gilbert says:

    @ventus45

    Re: ‘We know all about that extra fuel now

    Yes, we do and there was no extra fuel.

    I don’t know how this nonsense still gets a run. It has been demonstrated on multiple occasions in multiple forums that the fuel load was the minimum amount required to meet regulatory and company requirements.

  146. Ventus45 says:

    @Andrew

    Simply, the fact that as I remember it, the planned alternates in the final amended flight plan were judged (by others long ago) to be much further from Beijing than other alternates that were considered to be more normally used, and were supposedly acceptable (weather wise) on the night, and thus, he took more fuel for those “further” alternates. I am assuming that he contrived to select those further away alternates as a ruse to justify the extra fuel he needed to be able to get to 45S.

  147. TBill says:

    I feel like we are missing fuel history comparison to all other MH370 flights, so as usual we are outta ammo.

  148. Ventus45 says:

    @TBill

    Other airlines also fly 777’s between KL and Beijing.
    Any flight plans with alternate details from any of them would be useful, then or now.

  149. Mick Gilbert says:

    @ventus45
    @TBill

    I’ll explain the selection of the alternate as it has been explained to me. The short version is that it is pointless selecting an alternate that you can’t land at.

    The long version is that the choice of the alternate was dictated by the weather. The TAF for Tianjin (ZBTJ) had a TEMPO for visibility of 1400m in light snow and rain. The TAF for Jinan (ZSJN) was relatively clear.

    The MAS Ops Manual (Part A) para 8.1.3.6.2 states:

    An adequate aerodrome is suitable to be used as destination alternate, 3% ERA or destination aerodrome when isolated, if the weather reports or forecasts indicate that, during a period commencing 1 hour before and ending 1 hour after the estimated time of arrival at the aerodrome, the weather conditions will be at or above the planning minima.

    In this case, the expected type of approach was a Cat 1 ILS, so the applicable planning minima for MAS Ops would have been the non-precision approach minima. The only non-precision approach available at ZBTJ is a VOR, which requires a minimum visibility of 1700 m for the approach to Rwy 34L. According to the MAS rules, the TEMPO visibility of 1400 m made ZBTJ unsuitable as a destination alternate, hence the selection of Jinan (ZSJN).

  150. Niels says:

    @Victor Iannello

    Regarding the article in Le Parisien:

    I found the following paragraph especially interesting, in the light of ongoing discussions on pilot inputs or not after 19:41. Not sure what to think of it / what is behind it:
    “Selon nos informations, le déplacement aux États-Unis a permis de révéler que le MH370 a été piloté jusqu’au bout. Les données fournies par le géant américain accréditent une piste suivie discrètement par les gendarmes français : à savoir, il y avait quelqu’un derrière le manche lorsque l’avion s’est abîmé dans l’Océan indien.« Les juges nous ont indiqué que rien ne permet de dire que le pilote est impliqué », prévient Me Marie Dosé.”

  151. Andrew says:

    @Ventus45

    RE: “…the fact that as I remember it, the planned alternates in the final amended flight plan were judged (by others long ago) to be much further from Beijing than other alternates that were considered to be more normally used, and were supposedly acceptable (weather wise) on the night, and thus, he took more fuel for those “further” alternates.”

    Here’s why I think claims of ‘extra’ fuel are wrong:

    The two alternates for ZBAA (Beijing) that were nominated on the ATC filed flight plan were ZBTJ (Tianjin) and ZBSJ (Shijiazhuang). The ATC flight plan was filed about 12 hours before the flight’s ETD. The Operational Flight Plan (OFP – the document used by the pilots) that was issued 90 minutes before departure nominated a single alternate, ZSJN (Jinan), which, as you said, is further away from ZBAA than ZBTJ or ZBSJ.

    At the time the ATC plan was filed, the forecasts for ZBTJ and ZBSJ showed they were both suitable to be nominated as alternates. However, as Mick outlined above, the later forecast for ZBTJ that was issued at 070930Z had a TEMPO for visibility of 1400m in light snow and rain, which is below the required planning minima, making it unsuitable to be nominated as an alternate. The forecast for ZBSJ that was issued at 070705Z had similar conditions to the earlier forecast, however, the METARS show that the actual visibility was as low as 1800m during the day, which is only just above the required minima. In my view, it was sensible for the MAS dispatcher and/or aircraft commander to select another alternate with a better forecast.

    I should also note that the MAS operations manual lists the following as alternates for ZBAA: ZBTJ, ZSSS (Shanghai) and ZSPD (Shanghai-Pudong). ZSSS and ZSPD are significantly further away from ZBAA than ZSJN, so the choice of ZSJN as the alternate actually resulted in MH370 carrying less fuel than it might have done.

    @Mick Gilbert

    RE: “The only non-precision approach available at ZBTJ is a VOR…”

    The LOC/DME approaches are also non-precision approaches! Nevertheless, the TEMPO was also below the required visibility for those approaches.

  152. Andrew says:

    @Ventus45

    Further to my comments above, the following map shows the relative positions of the alternate airfields in relation to Beijing:
    http://www.gcmap.com/mapui?P=ZBAA-ZSJNZBAA-ZSPDZBAA-ZBTJZBAA-ZBSJ&MS=wls&DU=nm

  153. Victor Iannello says:

    @Niels: Yes, I also saw that paragraph. I don’t know what to think of it. Are they implying there were pilot inputs before and after fuel exhaustion? My guess is they are leaning heavily on the French analysis of the flaperon, which suggested that the flaperon was damaged as it was dragged along the water, and not so much on the ATSB’s evaluation of the flap, which the ATSB believe was retracted.

  154. Andrew says:

    Some of the characters were dropped from the URL I pasted above. Try the following link:
    ZBAA Alternates

  155. Victor Iannello says:

    @Andrew said: so the choice of ZSJN as the alternate actually resulted in MH370 carrying less fuel than it might have done.

    That’s a great point that I hadn’t seen before. If the intention was to carry extra fuel, the captain could have easily justified more fuel by designating an airport further from ZBAA as the alternate.

  156. Andrew says:

    @Victor

    Thanks, but on reflection it possibly wasn’t such a great point after all. A captain would have a hard time trying to justify an alternate so far away from the destination unless the destination weather was so bad that a diversion was likely. In such cases it would be preferable to divert to an ‘online’ port where the airline’s own ground staff would be available to deal with the passengers. In this case, however, the weather at Beijing was pretty good, so a closer alternate would be more reasonable because the likelihood of a diversion was low. Nevertheless, I think the map I posted clearly shows that ZSJN isn’t a whole lot further away from Beijing than ZBSJ.

  157. TBill says:

    @Andrew
    Re: fuel
    I agree if there was extra fuel it was just a small amount extra. Obviously the pilot’s job is to add a little extra fuel to get to alternate airport, and my understanding for China you want a little margin over the normal amount extra fuel. As a conservative engineer I can appreciate that is exactly what I would do too.

    Therefore we have “no leg to stand on” re: extra fuel, although Malaysia could prove that by giving us other MH370 flight fuel loads. My concern is actually that we are not getting any fuel quality data (density etc), which I think could indicate Malaysia knew MH370 was still flying so fuel was not the issue. But normally fuel check is made, is my understanding. MAS COO Dunleavy said they assunmed the flight was still flying, so I do not know if they interpreted the SATCOM sat calls as a live plane, but sounds like they had a clue.

    Guess we might have MH371 fuel load on the in-bound but that does not help much I don’t think.

  158. TBill says:

    @Victor
    That was a good article from France.
    I wanted to know what type of engineer is Ghislain Wattrelos?
    A number of serious discussion points, but I liked where it said we have no info to place ZS as the pilot. In other words, what I would say, nothing about the design of the aircraft captures the identity of the pilot, except the voice data recorder maybe, but that is on 2-hr loop and can be manipulated (as per your write up on JW some years ago). So even if we find the data black box, which could be emnpty, which if so would be further circumstantial evidence of tampering, we never have “proof” of ZS.

    I feel in America we have the Duck Rule, if it quacks like a duck etc. but that rule does not hold up in all cultures, is my thought.

  159. Andrew says:

    @TBill

    Fuel planning is a fairly rigorous exercise. The fuel analysis in MH370’s OFP clearly shows the aircraft was not carrying any more fuel than was required by law or the company’s own fuel policy. The only bone of contention is the choice of alternate, but I hope I have shown that ZSJN was justified given the forecast and actual conditions at the closer alternates.

    Every flight is different, so I’m not sure that other MH370 fuel loads would prove anything. In most cases you would probably find that other Beijing flights held ZBTJ or ZBSJ as the alternate, simply because they are the alternates that are regularly nominated for Beijing in suitable weather conditions. In some cases, however, you would find that a more distant alternate was nominated, due to the conditions at Beijing or the close-in alternates.

  160. Niels says:

    @VictorI
    You wrote:” My guess is they are leaning heavily on the French analysis of the flaperon, which suggested that the flaperon was damaged as it was dragged along the water, and not so much on the ATSB’s evaluation of the flap, which the ATSB believe was retracted.”

    Yes, possibly. Though the article imo suggests that Boeing provided information that “accredited” the investigative track / hypothesis followed by the French (that someone was at the controls when the aircraft crashed into the SIO)

  161. Richard says:

    @Niels

    I have worked in France for a number of years. The French authorities only take evidence from French sources.

    The ATSB analysis of the Flaperon is therefore automatically discredited.

    The statement from the French newspaper article is made without any evidence: “The data provided by the American giant substantiates a track discreetly followed by the French gendarmes: namely, there was someone behind the control column when the plane crashed into the Indian Ocean.”

    In my view, Victor is correct, when he says “My guess is they are leaning heavily on the French analysis of the flaperon”.

  162. Victor Iannello says:

    @Niels: I don’t put a lot of weight on a statement like that. I’d be very surprised if Boeing today believes the plane was piloted to the end. My guess is that Boeing provided data, and the French investigators interpreted that data in a way that supported their theory. But without seeing the data, it’s hard to know.

    I also wonder how current Boeing’s data and analyses are. My impression is that Boeing’s involvement ramped down in the first year after the disappearance.

  163. Niels says:

    @VictorI, Richard
    I agree that without seeing the data / evidence we should be careful with attributing weight on such statement.

    I may discuss with some of my good French contacts to get a better feel for their investigative “culture”.

    @VictorI
    I’m pondering about your impression that “Boeing’s involvement ramped down in the first year after the disappearance”. I had a similar impression, and it doesn’t feel well, to put it mildly.

  164. Nederland says:

    @Richard
    @TBill

    In the flight path via BEBIM, I personally wouldn’t worry much about fuel. It is probably true enough that a level path would come up with a lot of fuel remaining.

    However, my argument was that this flight path does make sense only when one also assume a wild (and unknown) change in altitude (a lot of fuel burned) during the turn around Sumatra.

    https://www.docdroid.net/GvlrLaV/mh370-waypoint-30.pdf#page=7

  165. Mick Gilbert says:

    @TBill

    Re: ‘Guess we might have MH371 fuel load on the in-bound but that does not help much I don’t think.

    Actually Bill, it possibly does help a bit. We know that if MH370 had landed in Beijing without touching any of its fuel reserves it would have had about 11,900 kg of fuel left in its tanks, comprised of:

    Contingency fuel: 1,200 kg
    Alternate fuel: 4,800 kg
    Final reserve fuel: 2,900 kg
    Company fuel: 3,000 kg

    We know that MH371 had returned from Beijing and landed with 8,200 kg of fuel in its tanks. We know that a large part of the 3,700 kg difference would have been the 3,000 kg of company fuel required for Beijing.

    We don’t know what the alternate for Kuala Lumpur was but I’m guessing that it was probably Penang. KL-Penang is about 20 nm shorter than Beijing-Jinan which would roughly account for the other 700 kg. So, you’re there or thereabouts.

  166. Mick Gilbert says:

    @Andrew

    Re: ‘The LOC/DME approaches are also non-precision approaches! Nevertheless, the TEMPO was also below the required visibility for those approaches.

    Yes, noted, thank you.

  167. TBill says:

    @Nederland
    Great! thank you for checking in.

    @Mick
    Thank you for that fuel analysis, makes sense.

  168. DrB says:

    @DennisW,

    You said: “I would assign the number of “no manuever” routes to be closer to 10,000 than 9 million.”

    My statement about the number of “no maneuver” routes being 9 x 10^6 should have emphasized that this was without considering the satellite data, which can then be applied to reduce the number of “no maneuver” routes which can lead to independent “Regions of Interest” (ROI).

    Your estimate of the number of ROI is a bit low when you include the uncertainties in the 19:41 and 20:41 locations due to the BTO reading noise. At +/- 1 sigma this changes the speed by 8 knots and the bearing by a degree. Thus I would say that the speed setting, with an uncertainty of +/- 8 knots or more, requires 3 values to be investigated. This is just at one altitude, and other combinations of speed setting and flight level can match the observed ground speed, but not at the same endurance. So one needs to explore about 3 flight levels at each speed setting to see if any of them have the correct endurance. In addition, the winds vary with altitude, so the ground speed will vary even more than the air speed. I could quibble about the number of lat/lons, but whether it’s 20 or 30 is not important. I would estimate the number of ROIs to be about 1×10^5.

    So, in summary, the number of independent MH370 routes is about 9×10^6, and the number of ROIs which are independent when applying the satellite data is about 1×10^5. So,if a route with major maneuvers was actually flown, its satellite data has to match only a single one out of 10^5 independent ROI with no maneuvers. That’s a pretty slim chance in my book, and especially so when the pilot did not know when the SDU communicated with the ground station, although he would probably know this during the phone calls if he was aware of them.

  169. DennisW says:

    @DrB

    OK, I understand. I was using the satellite data as a “prefilter” of candidate routes without actually doing any detailed BTO and BFO qualifications.

    Your conclusion is valid in that a peak in the detailed stats would be very significant. Frankly, I had not even thought about your approach.

  170. Victor Iannello says:

    @Niels: To be honest, my main interest is if there are additional data sets that were supplied to the French that we don’t have.

  171. TBill says:

    @Victor
    argh…one article says Ghislain Wattrelos still believes in his heart MH370 was a shootdown somewhere by someone

    …recall last month I was pointing out to @haxi the one key China NoK believes it was mechanical issue since that NoK is not buying into the pilot has ability to shut off comms without an alert signal

  172. Niels says:

    @VictorI
    Not entirely sure how to read your comment, however I would like to add

    Regarding Boeing’s attitude: an unconditional, strong commitment by the aviation industry to clarifying this terrible mystery would imo be appropriate and important.

    About the data supplied, the article says: “Grâce à la coopération de Boeing, les enquêteurs sont repartis avec une masse considérable d’informations à traiter, dont les données satellites de l’avion. « Un an de travail leur sera nécessaire »”

    It doesn’t help much , but it suggests more than just the satellite data.

  173. Victor Iannello says:

    @Niels: I would bet that there is no material information in that pile of data that we don’t already have, and I would also bet that the satellite data comprises much of it what was provided. I’m also willing to bet that the satellite data is identical to the unredacted data that was shared with me.

  174. CanisMagnusRufus says:

    VictorI
    I have closely followed NoK Wattrelos’ comments these few years and he had maintained an attitude of scepticism towards US, UK, and Australia’s involvement. It was this which led him to hold the ‘shootdown over S.China sea’ theory. The latest developments suggest that he has discarded that theory and now has to look closely at the ‘crash in the SIO theory.’
    The French word ‘abattu’ could be loosely translated as ‘brought down intentionally’ (as opposed to landed safely). The French have clearly rejected the pilot suicide theory, but are holding to the intentional diversion / ditching theory.

  175. Victor Iannello says:

    @CanisMR said: The French have clearly rejected the pilot suicide theory…

    Why do you believe that?

  176. DennisW says:

    @CMR

    The French have clearly rejected the pilot suicide theory…

    I don’t read that into the French remarks, however, the pilot suicide theory has been weak from the beginning. As has the mechanical and shoot down theories.

    It is clear (to me) that MH370 was diverted for a reason related to the political situation in Malaysia. The Malaysian early response (even criticized by ICAO), is indicative of that.

  177. Hank says:

    @DennisW

    I think that a piloted flight that ends with the death of the pilot could be considered to be a suicide or a the result of a mission. The result is the same for the pilot. But the thought given to hiding the aircraft at sea could be much higher for a mission with that specific objective. I have always believed ZS had a higher purpose and this would not have been a classic suicide. Maybe that is what some people mean by claiming this is not a suicide? Others use it to suggest a mechanical cause followed by a ghost flight (which could be reasonable) or a hijack to Kazakhstan.

    @DrB

    While one out of 9 X 10^6 autopilot mode options is very impressive, it is infinitely less than the infinite possible flight paths with an active pilot, which includes the LNAV 180 LRC where the pilot may have not done anything more. It is statistically meaningless to infer that any of the infinite flight path possibilities with active pilot would have less than 10^7 solutions. Infinite plus 9 x 10^6 is Infinite.

    You have clearly demonstrated that for no active pilot you have identified the best autopilot settings. This improves on the DSTG particle filter (which also assumed no active pilot) and it is worth re-looking at the search data around 34.5 S.

  178. Victor Iannello says:

    @DennisW: It depends on your definition of suicide. I think you would agree that the captain was ultimately responsible for navigating the plane to the SIO, killing him, the crew, and the passengers, whether or not there was a preferred plan not taken in which everybody survived. Some would call that murder/suicide. I prefer to not use terminology like “suicide”, “murder”, and “controlled to the end” because the meanings are different to different people.

  179. Niels says:

    @VictorI
    Independent of what exactly was obtained by the French:
    Their progress indeed seems to be slow. However, I find it important that such party performs a relative independent investigation. To me it appears that more than Five Eyes have been closed on 7/8 March 2014 and several instances afterwards. How much progress has there really been made in the past years in understanding what happened? The “pilot suicide” story which recently gained momentum is based on little or no hard evidence. At best it is the result of excluding other possibilities. How can one exclude other possibilities if the process of collecting facts has been so merky and perhaps even selective? I’m afraid much more patience and persistence is needed for this case.

  180. Victor Iannello says:

    @Niels: By the time the French finish the investigation, it won’t matter. Based on what I’ve seen to date, I have little hope that anything they do will be helpful. Certainly, releasing bits of information without any underlying data just causes confusion.

    I also disagree with your statement that theories based on a deliberate diversion by the captain has little or no hard evidence.

  181. DennisW says:

    @Victor,

    I agree. My reference is only to the intention of the diversion.

  182. CanisMagnusRufus says:

    my effort at translation (with help from Google)
    Flight MH370: revelations about the mystery crash
    by Timothée Boutry, Jérémie Pham-Lê et Vincent Gautronneau
    10 july 2019 à 21h45, modified 11 july 2019 à 08h13

    It’s the biggest mystery in civil aviation. On the night of March 8, 2014, flight MH370 connecting Kuala Lumpur (Malaysia) to Beijing (China) disappeared from radar screens with 239 people on board. Despite an exhaustive search in the Indian Ocean, apart from some debris, the wreck of the Boeing 777 of Malaysia Airlines has never been found.
    In July 2018, Malaysia’s international investigation team issued its final report stating that the investigation had failed to establish the true cause of the disappearance. Juridically, there is now only one country in the world to investigate this disaster: France. A judicial investigation conducted by two Parisian investigating judges is still ongoing.
    The French justice system has jurisdictional authority because of the presence on board of the plane of three of its nationals. Ghyslain Wattrelos, a French engineer, lost his wife and two of his three children that night. This Wednesday, he was received at the Paris court with his lawyer Marie Dosé to review with the magistrates and investigators mobilized to solve this enigma. The meeting lasted more than two and a half hours, and which included some of his in-laws, and during which he was able to appreciate the collosal amount of work done. And it’s far from over.

    French investigators visit Boeing
    The investigation have made a real advance. Between May 22nd and 24th, the judges and investigators of the Gendarmerie Air Transport Research Section (SR GTA) were finally able to travel to the United States at Boeing Headquarters in Seattle. The visit, anticipated for more than a year and a half, had been repeatedly postponed, the aeronautics giant not showing any inclination to cooperate. On site, the French visitors had to sign a confidentiality agreement. In concrete terms, all the data collected have been placed under seal and can not be added to the file. On the other hand, the investigators will be able to make use of them in writing their reports. Which will satisfy the civil parties. “The judges have worked hard to make this visit possible that we have been waiting for a long time. It’s satisfying,” says Dosé.
    Thanks to Boeing’s cooperation, the investigators left with a considerable amount of information to process, including satellite data from the aircraft. “A year of work will be necessary,” says the lawyer who also welcomes the involvement of Judge Carole Ramet, who called to other functions. The meeting originally scheduled with investigators FBI, who conducted their own investigations, however, could not take place. It could take place soon in France.

    The (investigation) track of pilot suicide (theory)
    According to our information, the trip to the United States revealed that MH370 had been piloted till the end. The data provided by the American giant supports a (investigation) track discreetly followed by the French gendarmes: namely, there was someone behind the (controls) when the plane crashed in the Indian Ocean. (But) “The judges told us that nothing allows us to say that the pilot is involved,” warns Marie Dosé.
    However, among those very close to investigations, it is considered that the captain’s suicide thesis is the most plausible. “Some abnormal turns made by the 777 can only be done manually. So someone was at the helm,” said one close to the investigation. Logically, this major data point raises the question of a voluntary crash. “It is too early to state categorically, says an investigator. But nothing substantiates that anyone else could have entered the cockpit … ”

    The two faces of the pilot
    Zaharie Ahmad Shah’s profile has been closely examined. On the one side, the 53-year-old rider does not show any roughness. Experienced – more than 18,000 flying hours – Zaharie Ahmad Shah was the father of three children, married, he does not seem to have a suicidal profile. On the other side, however, Zaharie Ahmad Shah was not so smooth. The pilot of Malaysia Airlines was a strong political opponent of those in power.The day before the disappearance of the MH370, he planned to attend the trial of Anwar Ibrahim, the main leader of the center-left Malaysian opposition. From there to imagine a vengeance against those in power? “It’s very complex and intimate, the causes of suicide,” says a good connoisseur of the file. In the final report, however, the Malaysian authorities appeared to exhonerate the pilot. “It’s not surprising. Malaysia has no interest in incriminating the national airline … ”

    The terrorism track investigated
    The Anti-Terrorist Branch of the National Police (SDAT) and the Directorate General of Homeland Security (DGSI) are participating in the investigation to explore the path of external interference. Without having been able to substantiate it so far. These specialized police officers, for example, used facial recognition software from the video recordings of the passengers boarding but the investigation was not very conclusive.The police also looked at the possibility of terrorist action by Uyghur or Tamil groups, but found no evidence to implicate them. Finally, they also dismissed the any action against engineers from the American company Freescale, present on board the aircraft.

    Sabotage investigated
    At the beginning of the investigation, the presence of a Malaysian aeronautical engineer sitting near Satcom, the satellite system of the aircraft, allowed for imagining the hijacking of the aircraft. “This assumption seems unlikely because he would have been intercepted by other passengers and the crew when he began to rummage with a screwdriver over him,” says a someone familiar with the file. This Wednesday, the judges have also explained to Ghyslain Wattrelos that the unit to be hacked was 3.60 m in height and was very difficult to access. The gendarmes of the SR GTA have also recently received a report from the Satcom that excludes the possibility of remote hacking.

    Inconsistencies that are raised
    A report on the passengers and baggage brought on board has just been given to the investigating judges. Its author was present in court at the meeting. “We realize that there are several contradictory passenger lists, for example in the seating of passengers, says Ghyslain Wattrelos. It was also learned that a mysterious load of 89 kg had been added to the flight list after take-off. A container was also overloaded, without anyone knowing why. The expert draws no conclusion. It may be incompetence or manipulation. Everything is possible. This will be part of the questions for Malaysians.”

    http://www.leparisien.fr/faits-divers/vol-mh370-revelations-sur-le-crash-mystere-10-07-2019-8114283.php

  183. TBill says:

    @CanisMagnusRufus
    They say: “However, among those very close to investigations, it is considered that the captain’s suicide thesis is the most plausible.”

    My reading is the pilot suicide thesis is their current leading theory. However, they point out they have no proof the pilot was ZS. In other words, (my words) no radio transmission or anything to provide exact identify of the pilot in the cockpit. So we must extrapolate from the evidence we have, which to me certainly appears like ZS at the controls.

  184. Nederland says:

    @TBill
    @Richard

    Although I have not been participating much recently, I’m still following the case and try to have a look at this blog site once a week or so.

    Evidence is piling up that MH370 recovered from a steep descent (e.g. French flaperon analysis, possible new input from Boeing). The crash site can therefore easily be further away than 25 nm from the seventh arc. A route BEBIM – ~McMurdo is still the only route that works exclusively with waypoint navigation. We know that MH370 was following a waypoint route after the diversion and up to the point it was turning south. Why not assume the same pattern for the rest of the flight?

    This route implies that MH370 burned a lot of fuel while circling around Sumatra, trying to avoid radar detection. I remain unconvinced that any pilot (presumably a non-Indonesian national) could predict with any confidence that the Indonesian radar was turned off after midnight (when Indonesia said it was operational at that time and did not detect MH370 in own airspace), or would have taken that risk with so much fuel left. The 18:40 phone call means that MH370 must have descended temporarily if it continued on to BEBIM later. The reason for this, then, was to reduce radar line of sight.

    @Richard

    Thanks for doublechecking!

  185. Victor Iannello says:

    @Nederland said: Evidence is piling up that MH370 recovered from a steep descent (e.g. French flaperon analysis, possible new input from Boeing).

    The French analysis of the flaperon is not new, and occurred before the ATSB’s analysis of the flap, which indicated the flaps were retracted. My guess is the majority of the new input from Boeing is the unredacted Inmarsat data set, which we’ve had now for two years. Sharing the data might have required a confidentiality agreement, as that data was shared with Boeing under similar terms. Perhaps the end-of-flight simulation results were also shared.

    Does anybody really think that Boeing would have additional information regarding the end-of-flight that was not shared with the ATSB?

  186. Nederland says:

    @Victor

    The French flaperon analysis excluded the possibility that the flaperon came off before impact because of flutter. This means that the aircraft can not have hit the water at very high speed (as indicated in the two final BFOs). Otherwise the flaperon (and other parts) could not be in the state they are now, i.e. relatively well-preserved.

    The French flaperon analysis also stated that the French investigators were unable to receive final confirmation from Boeing that their analysis of the damage was correct.

    One possibility is that the new cooperation with Boeing means that their analysis of the flaperon damage has been confirmed.

  187. Victor Iannello says:

    @Nederland: If we’ve learned anything, it’s to not trust anything but hard data. I’ve seen so much bad reporting and incorrect interpretation of data that I put little weight on what journalists say, and even less weight on what a lawyer, i.e., a paid advocate, says.

  188. TBill says:

    @Nederland
    I am still in general agreement with you. I assume you have checked out MH370-Captio.net flight path which is a well-thought out radar avoidance path near Sumatra. Maybe however the pilot did suspect radar was off, and nonetheless made a decoy flight path northwest before FMT south.

    Overall I would say there are divided opinions on how the flight to the SIO was conducted, how the crash happened etc. The only thing really matters is, if someone decides to search, it is up to them to decide where they want to search.

    I think if I was advising Malaysia, the assumption should be Arc7 +/-25, even though I agree it may well have gone further. I think they should probably check 20-25 South even though that is only a secondary pin for me. Then they can probably call it quits in good faith.

    As far as if we can pinpoint a location, that would be great but there is so much divisiveness, I have limited confidence in the suggestions. So then if it is a private search, they can check whatever pin they want to. If they take suggestions, I’ll maybe put one in, perhaps jointly with you.

  189. Richard says:

    @Nederland

    You stated “One possibility is that the new cooperation with Boeing means that their analysis of the flaperon damage has been confirmed.”

    That is conjecture based on incomplete information from a newspaper article.

    The Australian ATSB report in 2016 on the confirmed MH370 debris item of the Right Outboard Flap, manufactured in Italy by the Airbus company Alenia, is based on a physical analysis and concludes the flaps were not extended. (“Damage examination on the recovered part of the right outboard flap (Item 19), together with the damage found on the right flaperon (Item 1) indicates that the right outboard flap was most likely in the retracted position and the right flaperon was probably at, or close to, the neutral position, at the time they separated from the wing.”)

    The French DGA report in 2015 on the confirmed MH370 debris item of the Right Flaperon, manufactured in Spain by the Airbus company CASA, is based on physical analysis and rejects flutter as a possible cause of damage, but comes to no firm conclusion, only an unconfirmed hypothesis that the damage was due to impact with water. (“The little data supplied by Boeing did not enable the examination to be progressed by making calculations that would have made it possible to confirm or reject the proposed hypothesis.”

    The Malaysian SIR concludes “It should be noted that the DGA/TA, after examining the flaperon soon after it was found in July 2015, had concluded that the flaperon was likely to be deflected at the time of impact. This was primarily based on the damage observed on the trailing edge of the flaperon. However, this scenario was considered a hypothesis only due to lack of corroborating information, and more importantly, it was done without the benefit of the damage information available from the right outboard flap which was found much later.”

    Quite clearly the analysis by the DGA was thwart by the ongoing rivalry between Boeing and Airbus.

    Quite clearly the analysis by the ATSB and DGA was thwart by the lack of co-operation between Australia and France.

    Nevertheless I prefer the more recent ATSB analysis, above the earlier DGA analysis. I prefer both analyses above conjecture based on a newspaper article.

  190. Tom O'Flaherty says:

    I think the course of action taken fits well with a pilot expecting a delayed response having gone missing. I believe that expectation could have been formed following years of night flying in the area concerned.

    What would be interesting to know is whether at any time in the past ZS had intentionally delayed handover or gone dark for a brief period to observe what (if any) reaction had followed.

    I remain convinced that MH370 was chosen well in advance due to a combination of route, time of day and having a pliant co pilot aboard.

    I see the Jailing of Anwar Ibrahim as a coincidence, timing wise. I expect it was a foregone conclusion for some time well before the day of the flight, although it could well have been a factor.

  191. Victor Iannello says:

    For you French speakers, here’s an interview with Ghyslain Wattrelos and Xavier Tytelman. There is not much in the way of new information. GW continues to believe the plane was downed in the South China Sea and the Americans know, the Chinese know, and some French know. GW claims the theory that there was a suicidal pilot that was in control until the end was proposed by a single Boeing engineer, and does not seem to be the official position of Boeing.

  192. TBill says:

    @Victor
    I call denial…if GW found a Boeing engineer admitting pilot suicide, then he might be getting to the heart of the matter.

  193. Richard says:

    @Victor

    Many thanks for making the interview with Ghyslain Wattrelos (NOK) and Xavier Tytelman (Aviation Consultant and Co-Author of the newspaper article) available.

    Both agree in the interview, that the visit to Boeing brought nothing new from a technical perspective.

    Ghyslain Wattrelos continues to believe he is being given the run around by the authorities. He is certain that the US, French, UK, Australian, Chinese and Malaysian Governments know the truth, but are keeping the truth hidden. He hopes, that one day someone will blow the whistle.

  194. Nederland says:

    @Victor
    @Richard
    @TBill

    I couldn’t give less on the newspaper article in Le Parisien, which is probably a concoction of something already known since years.

    My point was that there is increasing evidene that MH370 recovered from the steep descent indicated by the two final BFOs.

    The more debris parts turned up, which have in common that they are relatively well preserved, the less likely it became that ALL of these came off in flight because of flutter: e.g. the flaperon, the outboard flap, some wing panels, flap track fairing and so on (not least the vertical stabiliser).

    I agree that there was a communication problem between France and Australia. As a result, the French tentative assumption that the flaperon indicates a controlled ditching is probably wrong. Their report isn’t very specific on this, this was just a hunch at the time. The report, however, is very affirmative in that the flaperon did not come off because of flutter (I’d say this is hard data because they personally inspected the item, not just a photograph). But if it didn’t come off because of flutter, and this was a high speed impact, extrapolated from the two final BFOs, then how can the flaperon (and all the other parts) be so well preserved?

    (“t appears possible to exclude in-flight loss of the flaperon since its weight is concentrated forwards, which would a priori lead to a fall with the leading edge forwards and the probable destruction of the latter. The damage to the trailing edge would also likely be different.”)
    It is also worth noting that the ATSB didn’t see any evidence for flutter damage for any of the parts they examined either – at least they never made any positive statement in that direction.

    One alternative explanation I could think off is an in-flight explosion in which an entire wing came off and went down separately, but this sounds unlikely, too, and there are also parts from both wings, not just the one.

    Most importantly, the area around the arc has already been searched to widths of 25/40 nm. If the wreckage is outside that area, then the aircraft probably recovered from the initial steep descent. There are very few other options left.

    (btw, the end of flight scenario is entirely different from the route, other than that routes south of 25S are back in contention).

  195. ventus45 says:

    @Nederland

    Re:
    “Most importantly, the area around the arc has already been searched to widths of 25/40 nm. If the wreckage is outside that area, then the aircraft probably recovered from the initial steep descent. There are very few other options left.”

    There is no evidence that there was a “steep descent”, let alone a “dive”. The so called “Eight Second Dive Evidence” (00:19:29 to 00:19:37) is not “evidence” at all.

    This has been an unjustified assertion of the “ghost flight brigade” from the earliest days, simply to discredit the “ditching brigade”. It is entirely predicated on accepting the 00:19:37 BFO, which Ashton et al clearly said was “”iffy””

    TO QUOTE DIRECTLY FROM:
    The Search for MH370 – Chris Ashton, Alan Shuster Bruce, Gary Colledge and Mark Dickinson
    (Pages 15 – 16)
    (HIGHLIGHTS MINE)
    5.3. Refinement of BFO Samples.
    Detailed analysis of BFO samples taken from other flights showed a high degree of consistency for the signalling message frequencies, with the exception of those that were performed immediately after the initial logon process.
    This called into question the BFO measurements after the log-on sequences at 18:25 and 00:19.
    However it was also determined (by the same method) that the first message transmitted by the aircraft in the logon sequence, the Logon Request message, did provide a consistent and accurate BFO measurement.
    This means that we can use the Logon Request message information from 18:25:27 and 00:19:29, but it is prudent to discount the measurements between 18:25:34 and 18:28:15 inclusive, and the one at 00:19:37.

    (NOTE THAT LAST STATEMENT)

    The “ditch” is a real possibility.
    It is about time that this is admitted by the ghost flight brigade.

  196. Victor Iannello says:

    @ventus45: You are quoting from Ashton’s paper from Sept 2014 and ignoring the later, more extensive work on the BFO by Ian Holland of the DSTG from January 2018. (That more recent work was performed with the assistance of Inmarsat and Thales.) Whether or not there was a controlled descent and ditching after the steep descent suggested by the final two BFO values, there certainly is evidence that a steep descent occurred, and I strongly suggest you read and understand Holland’s paper before accusing others of “unjustified assertions” in order to “discredit the ditching brigade”, which frankly is a load of horse manure.

  197. CanisMagnusRufus says:

    @VictorI
    I think you misunderstood GW’s interview.
    GW continues to believe the plane was downed in the South China Sea
    He didn’t say this anywhere in the interview.
    …and the Americans know, the Chinese know, and some French know.
    This is true, he did say that.
    GW claims the theory that there was a suicidal pilot that was in control until the end was proposed by a single Boeing engineer, and does not seem to be the official position of Boeing.
    Not exactly. What GW said was the following: I think, very quickly, a link was drawn between the claim of one Boeing engineer that the plane was ‘piloted’ till the end, and the idea that the pilot in charge, ZS, was suicidal. I’ll tell you exactly what the judge told me. We don’t have ANYTHING about the pilot to support this theory, and much to support against this theory. (see 5:15 onwards)

    I have closely followed GW’s interviews and I have never seen him this happy till now. Clearly, the outcomes from the French investigators visit to the US has improved his mood. I have great sympathy towards him, and towards all the other NoK’s.

  198. Nederland says:

    @ventus45

    Even the early article by Ashton et al. acknowledges that the penultimate (and ultimate) BFO indicate a steep descent:

    “The errors at 16:42,16:55 and 00:19 are thought to be due to vertical movement of the aircraft”, p. 21

  199. Nederland says:

    Wattrelos also says right in the beginning that he met “judges” (a group that includes aeronautical experts) who recently came back from Boeing “with a pile of documents” but said that these are confidential.

    The information that MH370 was in control until the end (as reported from Le Parisien) came from a Boeing engineer (~5:30 min). Right in the beginning (~2:10 min) he seems to indicate (difficult to understand) that this is based on a single piece of debris (presumably the flaperon). He goes on to say that the judges have told him they have no evidence to incriminate Zaharie, but a lot to exonerate him.

    W. talks about the Inmarsat data only in general terms, nothing new here.

    @TBill

    Yes, I did check MH370-captio.net (which came out later) as well.

  200. Victor Iannello says:

    @CanisMR said: I think you misunderstood GW’s interview. He didn’t say this anywhere in the interview [that the plane was downed in the South China Sea].

    Of course he did! There was an entire discussion about this starting around 18:00. First he claims that the Americans know what happened because the area was surveilled by AWACS and ships. Then around 18:11 he said the plane was shot down (“Cet avion etait abattu”) and the only question is for what reason. He says if it was shot down because it was hijacked, he could accept that. In various parts of the interview, he completely rejects that the captain committed suicide, the military radar data because of the anomalous altitude measurements, and the SIO crash site because no debris was found in Australia.

  201. TBill says:

    Re: GW
    I agree with GW that Boeing, Malaysia, FBI, CIA, France etc etc know something about MH370 that they are hiding. But when GW gets access to priviledged info to say it looks like pilot diviersion, he cannot accept that. He will only accept the inside info if they say the cause is shoot down. He “knows” it was a shoot down and when the authorities finally agree to his theory he will be happy, otherwise he knows they are lying.

    Seems to me….

  202. Sid Bennett says:

    It has been some time since I have last made a public comment on the various options, primarily as it seemed appropriate to let Ricahrd’s drift analysis proceed without too many distractions. However, as the discussion has reverted to the interpretation of the satellite data, I have read some of the recent work by Dr.B and others and would like to offer some comments.

    With regard to who did it, I prefer the slang term “the perp” which de-personalizes the discussion and focuses on the objective analysis of the satellite an other quantifiable data.

    I spent a lot of time trying to match up the pre and post FMT candidate paths, and experimented with some of the proposed diversion paths prior to and after the FMT with only moderately interesting results.

    But, at the time,I returned to my earlier analysis using Barry’s spread sheets and assuming the minimum number of turns in the flight. I added in the offset path in the leg leading to the FMT, which, amongst others Barry, Geoff Victor and I studied. My own thinking was confused by a statement in one manual that the offset would be cancelled if a waypoint turn of greater that 90 deg was performed, which is certainly true at the FMT, but I have recently seen a statement in a blog on flight simulators that the criterion is 135deg. I have not revisited this issue as yet.

    The most satisfactory result I had obtained with a FMT 18:41:30 where the azimuth would be 196T and continuing the turn to 182 deg for the remainder of the flight and a constant M=0.82 The end point at 00:11 would be 92.45013 -34.49622. To do this, I had to modify Barry’s spread sheet to enter the detailed track azimuths for the offset and for the FMT so as to give smooth transitions in azimuth consistent with standard turns.

    The path runs very close to ISBIX. So, instead of a arbitrary turn to 182 deg, the next waypoint after IGOGU would be ISBIX. Alternatively, a path IGOGU-BEDAX_ISBUX also results in a 182 deg heading. All of these may differ by a fraction of a degree, but seem good enough given the sparsity of the data. (I believe that a turn to 180 at some point after the FMT would also yield a good result, but I have not studied the scenario. Dr.B’s recent work suggests it might, but the M would need to be reduced so as to permit continuity through the FMT without assumed maneuvers.)

    At this juncture I do not see an objective way to differentiate between the various alternatives. It would be helpful to be able to do so in order to find a definitive path so as to make a stronger case for the search location, but since the location is the same, is it a distinction without a difference?

    Whether the plane crashed near to the end point or performed a glide maneuver may not be relevant as we would be searching a small length along the 7th arc and could easily search that area to a greater transverse distance if the initial search did not locate the plane.

    To a certain extent the above results are constrained by the design of the spread sheet, which assumed a constant M for the entire path and which used an earlier version of the met data. In addition to the above assumptions, the 182T is constrained to be a GCP, so that cross-track errors are ignored.

    I am happy to share the actual; spread sheet with anyone who has previously posed a detailed study of the route. The actual base spread sheet has previously been published by Barry.

    Sid

    I have recited some history from memory and I apologize to any IG member whose work I have mis-characterized or omitted to mention.

  203. Nederland says:

    Re end of flight

    Also note the wording Foley used when questioned about the ATSB search strategy by the Australian Proof Committe:

    “Today we have an analysis of the flap that tells us it is probably not deployed.

    “We have an analysis of the final two transmissions that say the aeroplane was in a high rate of descent.

    “We have 30 pieces of debris, some from inside the fuselage, that says there was significant energy at impact.”

    “We haven’t ever ruled out someone intervening at the end.”

    Note the deliberate change between “high rate of descent” and “significant energy at impact” (rather than *high* energy at impact). This implies that the impact was probably somewhere between controlled ditching and steep high-speed descent (whether controlled or uncontrolled).

    Perhaps this was the reason why the ATSB decided in late 2015 to double down on the width of the southern search area rather than to cover northern areas.

  204. Victor Iannello says:

    @Nederland said: Note the deliberate change between “high rate of descent” and “significant energy at impact” (rather than *high* energy at impact). This implies that the impact was probably somewhere between controlled ditching and steep high-speed descent (whether controlled or uncontrolled).

    So you think a “steep high-speed descent” does NOT result in “significant energy at impact”?

    You are putting too much emphasis on Peter’s choice of words. He could have just as easily said “high energy impact”. It means the same thing.

  205. Nederland says:

    @Victor

    I have given my explanation in the above. The largest piece found from the Germanwings 4U 9525 crash is just a fraction of the size of the outboard flap, for example. The outboard flap, flaperon, and other parts are too well preserved to have come off as a result of a high speed impact as indicated by the final BFOs. Most parts from MH370 show no interior (core) damage, they simply broke off.

    I therefore think that Peter Foley chose his words wisely, yes. But it doesn’t matter at the end of the day.

  206. Richard says:

    @Nederland

    Here are 3 items of debris from the German Wings crash.

    They are not just a fraction of the size of the MH370 debris items as you claim.

    https://www.dropbox.com/s/6nfempknr9arxxj/German%20Wings%201.png?dl=0

    https://www.dropbox.com/s/e20s01p9fcxgywr/German%20Wings%202.png?dl=0

    https://www.dropbox.com/s/w060whn0scgazvs/German%20Wings%203.png?dl=0

  207. Nederland says:

    @Richard,

    OK, thanks for checking.

    I’d say #1 is badly damaged/crushed, much unlike the outboard flap and flaperon, which look basically intact except for the bits that were cleanly separated (presumably at first contact with water).

    #3 is far smaller than the outboard flap.

    I cannot comment on the dimension of #2. Is this a part of the stabiliser? If so, it was very well protected.

    Are there any wings part from the Germanwings crash or similar incidents? This would yield a useful comparison.

    (https://www.atsb.gov.au/media/5771520/debris-report-3.pdf, p. 4 gives a good impression of the dimension and overall state of the outboard flap).

  208. Brian Anderson says:

    @Nederland,
    The angle of descent, even at 15 or 16,000 ft/min, is not extreme. It is certainly not steep, nor vertical. These terms have been used extensively by the popular press. Assuming the aircraft speed is, say 500 knots, then the descent angle would be about 16 deg below the horizon. However the speed would be increasing rapidly, as would the potential for structural damage to the airframe. With the likely acceleration at that time, and with no engine power (another assumption) the aerodynamic loads on the control surfaces may make it impossible for one person on the control column to recover from the descent.

    The simulations done by ALSM and others suggest that the descent through 15,000 ft/min was during the latter stages of an accelerating spiral dive. Even if a successful recovery was possible, there is no way to determine the likely heading of the subsequent “glide”.

    The structural damage, as evidenced by some of the fuselage interior debris, strongly suggests a catastrophic impact, although that could still be the result at the end a glide scenario.

  209. Don Thompson says:

    @Nederland

    The part depected in Richard’s #2 item is the lower end of D-AIPX’s rudder, apparently still attached to something of the vertical stabiliser.

    You wrote, the “outboard flap, flaperon, and other parts are too well preserved…“. The outboard flap found at Pemba, and inspected by the ATSB, is only one third of the full span of the flap, part of its trailing edge wedge is gone, the remnant of the aluminium alloy flap carrier shows a significant fracture, the remnant of the pivot link shows a facture consistent with a spanwise bend. Preserved is not a word that is in any way consistent with the state of the outboard flap segment, nor with any other part that I have observed.

  210. Hank says:

    @Brian Anderson

    I think the Ram Air Turbine would deploy and would provide hydraulic power for the flight controls. I do not believe the 777 has servo tabs to provide manual reversion, but I do not know for sure. The RAT is there to provide emergency electric power and primary flight control hydraulics.

  211. Nederland says:

    Yes, the right outboard flap is around a third of the original piece. It is cleanly fractured with no apparent core damage which you would expect if the aircraft hit the water at around 500 knots and gets squeezed as a result. The largest parts from Germanwings (#1 above) is heavily crushed (and appears still smaller than the outboard flap). The flaperon, on the other hand, is in one piece, except that a part of the trailing edge came off (presumably on impact with water), nothing like this survives from the Germanwings crash. Almost all the other parts found from MH370, too, are fractured around the edges with no crush/squeeze (core) damage. The damage to the right outborad flap (to me) indicates that the aircraft came down near level but with the left wing first. Therefore there are almost no parts found from the left wing, but a disproportionate number from the right wing, and the left section of the trailing edge of the right outboard flap came off, too.

    In the case of the Germanwings crash, the crash site was found immediately with around a million pieces. We only have around 30 pieces from MH370 and amongst these are many still larger (and in a much better state) than the largest pieces from Germanwings. It is well possible that even larger parts, e.g. from the airframe, were once floating on the ocean but sunk over time.

    Ethopian 961 came down in a similar way (attempted ditching with no fuel), left wing first, and the fuselage broke up on impact). I would at any rate expect a worse outcome in the case of MH370 since the first stage of the descent was a high-speed descent rather than a glide.

  212. Don Thompson says:

    @nederland,

    D-AIPX made a controlled descent into a mountainside. Before it impacted with the ground there were no aerodynamic conditions that presented any risk to its structure. The condition of recovered debris from 9M-MRO, and the most likely interpretation of its end of flight profile, does not exclude the possibility that it experienced some aerodynamically induced structural failure before impact.

    The area over which the debris of D-AIPX is strewn is more worthy of consideration/comparison than the debris itself. That terrain, the vertical profile, in deep seafloor bathymetry could be a challenge to making a successful side scan survey.

  213. TBill says:

    @Nederland
    Interesting perspectives above…I was not paying attention to MH370 when NTSB expanded the orig search zone wider.

    The overall picture is that ATSB was saying then, nominal 38 South Bayesean hot spot if we assume 18:40 sat call was FMT. What IG is now saying, my words, essentially IG believe that hot spot moves to 34.5 South if we relax 18:40 to not being the FMT. In both cases passive flight after 19:41 is prioritized.

    Re: final crash scenario, IG holds passive, rapid descent per BFO, whereas William Langewiesche is saying active pilot, rapid descent. In both cases we end up near Arc7. Of course I fear it may be active, not-near-Arc7 end point descent. It could also I suppose be passive not-ending-near-Arc7 descent.

  214. Victor Iannello says:

    @TBill: If there was no loiter, no ongoing descent at 18:40, infinite fuel, some BFO bias drift after 19:41, and faster-than-predicted debris drift, 38S looks like a hotspot. In consideration of our best estimates of fuel consumption and debris drift, and a more rigorous treatment of some of the statistical parameters, a crossing near 34.3S looks more likely.

    If there is a single route (i.e., a single, sharp peak of the objective function) with no pilot inputs after 19:41, that makes it very unlikely that the route was flown with significant pilot inputs after 19:41. This is an important point to understand. If this condition is satisfied, then “passive flight” would not be an assumption. In fact, it would be proven with a high level of confidence. The question is then whether the condition is satisfied.

    I have no idea why you believe “William Langewiesche is saying active pilot, rapid descent”.

  215. TBill says:

    @Victor
    Without re-reading William Langewiesche, my interpretation is he is willing to go along with passive flight assumption (at some risk that is incorrect, he says) but I thought he was saying the descent itself seemed to be active, that the pilot pushed the nose down for a catastrophic crash. I am not in total agreement with all of William Langewiesche’s interpretations, but overall of course I am in general agreement with him on the basic scenario.

  216. Nederland says:

    Take for example Egypt Air 990 (suspected pilot suicide, impact on water):

    “The small size of most of the recovered pieces of wreckage was consistent with the airplane impacting the water at a high speed.”, p. 34

    There was a second debris field associated with the left engine which came off in flight and this included some wing parts.

    If, however, the French report is right to say that the flaperon probably didn’t come off in flight, but only when hitting the ocean, then the impact does not seem to be at a very high speed.

  217. Nederland says:

    Link to the accident report of Egypt Air 990:
    https://www.ntsb.gov/investigations/AccidentReports/Reports/AAB0201.pdf

    Silk Air 185 (suspected pilot suicide/crash in river):

    “The aircraft parts found in the river were highly fragmented and mangled on impact, see Figure 8. As a result, all parts of the aircraft were mixed together on recovery, making sorting and identification of the many small pieces difficult.”, p. 8

    https://reports.aviation-safety.net/1997/19971219-0_B733_9V-TRF.pdf

    Some parts came off in flight and were therefore found on land, including a wing panel around two metres long (p. 9 and figure 7).

    Pictures of the debris found in the river: p. 229-232 in the pdf (figure 8)

  218. Richard says:

    @Nederland

    When we find MH370, you will find 10’s or 100’s of thousands of small pieces on the ocean floor, as long as you sift through the sediment, far enough around the crash site.

  219. Nederland says:

    @Richard

    Again, my argument is that none of the recovered pieces in the incidents quoted above is comparable in terms of size and failure mode to the outboard flap, flaperon etc (if they didn’t come off in flight).

  220. Victor Iannello says:

    @Nederland: The debris fields from Egypt Air 990 and Silk Air 185 provide evidence that in a high speed descent leading to a crash, control surfaces may separate in-flight and be recovered as relatively large pieces, which is consistent with the recovered flaperon and flap of MH370.

    The French hypothesis that the flaperon did not separate in-flight is based on the lack of dents on the leading edge, which they believe would have been first to hit the sea because the center-of-mass is shifted towards the leading edge. It’s also possible (and many of us think more probable) that the damaged flaperon, which is asymmetric and light-weight, would have spun on the way down, perhaps more like a falling sycamore seed.

  221. Nederland says:

    @Victor

    Sorry, but I presume the DGA team was aware that a good chunk (a third?) of the trailing edge is missing when they came up with their assessment.

    Also, their other argument is that the damage that occurred at the point the flaperon separated from the rest of the wing can only be explained with water impact, not with flutter or similar aerodynamic failure scenarios:

    “The loads generated, pushing from the lower surface towards the upper surface locally (unlike uniform aerodynamic loading), resulted in a bending load from the rear towards the front as well as of inboard towards outboard. This caused torsion on the flaperon.” (p. 26, There are more technical explanations like this in the report.)

    The previous working assumption was that the damage to the hinges indicates flutter damage, which seems not to be true.

    (This is unrelated to speculations about whether or not the flaperon was deflected).

    I remain unconvinced…

  222. Richard says:

    @Nederland

    Again, my argument is that three recovered pieces from the German Wings incident, that I picked off the internet are comparable in terms of size to the MH370 outboard flap and flaperon.

    Here is a picture of what appears to be a section of the vertical stabiliser, which is even longer than the MH370 outboard flap or flaperon.

    https://www.dropbox.com/s/adfaqewm1dez5xu/German%20Wings%204.png?dl=0

    The facts contradict your continued unfounded conjecture.

  223. Nederland says:

    @Richard

    How can you tell the length of that piece?

    No surprise that the vertical stabiliser is the least damaged in the Germanwings crash.

  224. Andrew says:

    @Nederland

    RE: “If, however, the French report is right to say that the flaperon probably didn’t come off in flight, but only when hitting the ocean…”

    The French DGA report states (my emphasis) ‘..it appears possible to exclude in-flight loss of the flaperon since its weight is concentrated forwards…’. It also states, in relation to in-flight separation: ‘A simulation of a flaperon fall with an initial speed corresponding to that of an aeroplane in flight would enable this to be definitively eliminated.’. Given that no such simulation was conducted, it is by no means certain the flaperon could only have separated on impact with the water. As Victor and others have mentioned, the flaperon might well have separated earlier if the aircraft broke up in-flight. The flaperon’s aerofoil shape might then have caused it to fall like a sycamore seed.

    Further, the DGA’s separation hypothesis relies on the flaperon being deflected at the time of impact with the water. However, the ATSB’s more recent analysis shows the flaps were ‘most likely in the retracted position’ and the right flaperon was ‘probably at, or close to, the neutral position’ at the time the surfaces separated from the aircraft. If that was the case, the justification for saying the surfaces could only have separated on impact with the water seems even less certain.

  225. Ventus45 says:

    The fall characteristics of the flaperon could probably be determined, if one was tested in a vertical wind tunnel, like the ones skydivers use. There are plenty of them around.

  226. Ventus45 says:

    Add, personally, I have a hunch, that due to it’s shape, it would be more likely to continuously pitch down about the quarter chord line, that is, rotate about the quarter chord line, that is, “tumble”.

    If it hit the water at the instant when the trailing edge was rotating though the bottom position, that might explain the ripping off of the trailing edge.

  227. Ventus45 says:

    Why not get a flaperon from a decommissioned 777 in a boneyard, and test it ?

  228. Nederland says:

    The hypothesis that the flaperon was perhaps deflected was derived from the lower third of the trailing edge, which probably came off on impact with water. Perhaps the aircraft was in a tail down attitude and this could also explain the similar damage to the outboard flap. How else can the damage to the trailing edges of both parts be explained other than with impact on water (i.e. if they came off in flight)?

    At any rate, this is unrelated to the torsion damage on the leading edge (which independently makes it unlikely that the flaperon came off in flight) as well as to the aerodynamic considerations on how the damaged flaperon goes down. Neither the SIR nor any of the ATSB reports came to a different conclusion here and/or made a case for flutter damage/in flight separation for any of the parts found from MH370.

    I agree there is no 100% certainty right now, perhaps something new will come out from the recent collaboration with Boeing (other than a second hand report).

  229. David says:

    @Nederland. The torsion damage might have resulted from the flaperon outboard end striking the adjacent outboard flap, either by that end of the flap being forced down and inboard or the flaperon forced up and outboard, though there is some conflicting evidence.

    I covered this, the French report and other aspects of flaperon and flap damage in the below, which also refers to earlier work. It might be useful background.

    https://www.dropbox.com/s/kqgi5tjcnjzc25e/Flaperon%20Separation%2C%20MH370.docx?dl=0

  230. Nederland says:

    @David

    Have you received comments from the DGA?

    Some parts of your paper seem to support the above mentioned scenario:

    “Whereas the DGA ditching hypothesis does not support a ditching, deployed or not and clearly this lateral displacement hypothesis does not support a flaps-down ditching, it might support one flaps-up. The right wing striking the sea might generate both the lift and drag to break the wing as described above”

    “a separation initiated by flutter should would portray more ambiguity in the character of failures of the upper and lower skins than the DGA examination describes”

    p. 14

  231. Andrew says:

    @Nederland

    RE: “The hypothesis that the flaperon was perhaps deflected was derived from the lower third of the trailing edge, which probably came off on impact with water.”

    My issue is that you keep saying ‘probable’. The French DGA report hypothesises a possible scenario, which they were unable to confirm. The report clearly states: ‘Beyond the possible in-flight loss, which is not entirely excluded, this scenario can only be considered as a hypothesis…’.

  232. Richard says:

    @Nederland

    You asked “How can you tell the length of that piece?”

    (1) By comparing its length to the height of the 3 crash site workers in red jackets in the same picture to the left and the 1 crash site worker in a red jacket in the picture below to the debris item.

    https://www.dropbox.com/s/vjx40q4mbuchyk0/German%20Wings%205.png?dl=0

    (2) By comparing the debris item to a picture of D-AIPX and the known dimensions of an Airbus 320-200.

    https://www.dropbox.com/s/kkwndeup6uuuz6o/German%20Wings%206.png?dl=0

  233. David says:

    @Nederland. “Have you received comments from the DGA?”
    No. Not that I expected any.

  234. Nederland says:

    @Andrew

    My understanding of the French report is the following:

    The “hypothesis” that the aircraft was ditched/flaps down was seen as *possible* at the time (this is now seen as unlikely because of additional evidence).

    The scenario that the flaperon was lost in flight is “not entirely excluded”, but “it appears possible to exclude in-flight loss” (i.e. unlikely)

    The scenario that the flaperon came off on impact with water is seen as *probable* and not just because of the damage to the lower third of the trailing edge (e.g. “it appears that the flaperon impacted the water while still attached to the aeroplane”, p. 29 and “Local impact on the lower side trailing edge generated torsion on the flaperon (bending upwards from inboard to outboard And from rear to front)”, p. 26)

    Clearly, the report argues in favour of separation on impact, not in-flight loss (and not 50/50).

    http://mh370.mot.gov.my/Appendix-1.12A-2-Item1Flaperon(Main).pdf

  235. Richard says:

    @Nederland

    I estimate the length of the D-AIPX Vertical Stabiliser debris item is around 5.5 m.

    The length of the 9M-MRO Flaperon is 2.315 m.

    https://www.dropbox.com/s/j233ffuj7rd4zeh/D-AIPX%20VS%20Debris%20Dimensions.pdf?dl=0

  236. Nederland says:

    @Richard

    OK, I can see that the vertical stabiliser of the Germanwings flight is of considerable size.

    However, as has been pointed out before by Don (“D-AIPX made a controlled descent into a mountainside”), the Germanwings flight is perhaps not the ideal point of comparison.

    I also drew attention to Egpypt Air 990 and Silk Air 185, which seem to match the presumed descent rate of MH370 better than Germanwings.

    How about the photos in figure 8 (p. 229-232 of the pdf) from Silk Air 185? How would a piece the size of the outboard flap or flaperon fit in here?

    https://reports.aviation-safety.net/1997/19971219-0_B733_9V-TRF.pdf

    (We have two separate line of arguments: the one is that the flaps etc can still be in the shape they are despite high speed impact on water, the other is that, while the former seems unlikely, it is still possible that these pieces all came off in flight).

  237. Don Thompson says:

    @nederland, others

    A recurring call at this blog has been a call for ‘raw’ data, that is, the most fundamental, minimally processed, data pertaining to a particular topic concerned with the loss of 9M-MRO. Civil ATC radar surveillance data: the trope of wildly fluctuating altitudes, and so on, now refuted. 25 hours of logged SATCOM data: e.g. correlation of satcom data vs otherwise known position for prior MH371 leg. etc, etc.

    ICYMI, the objective is to make an independent review and, if necessary, challenge prior conclusions.

    In the case of the flaperon an early deduction that became widely held set out that an upward force, acting on the trailing edge of the flaperon, caused fracture and departure of the t/e wedge.

    That no other debris was recovered during the subsequent 9 months helped cement a notion that the flaperon t/e broke off by the action of it being dragged along the ocean surface. Then, in July 2018, the DGA-TA report published as an appendix to the Malaysian SIR.

    That DGA-TA report contains much detail of the fundamental analyses made by DGA-TA. For example, planche 50-53, 55 & 56: SEM images of portions of the remnant composite upper and lower flaperon surfaces to assess fibre fractures within the inner and outer skins. Findings from the SEM analysis set out that fibres of the upper surface, inner and outer skins, fractured in traction whereas the lower surface, inner and outer skins, predominantly showed compression fractures. This composite materials analysis suggests that the force was acting downwards on the flaperon t/e. A downward force acting on the t/e contradicts any notion that the t/e damage resulted from ‘dragging along the water’.

    Above, @Nederland mentions the MSR990 crash. Some significant facts, relevant to the case of MH370, are presented by the Structures Group and Performance Group of the NTSB’s investigation. These include: 1) rate of descent progressing through -17K,-29K,-16K, to -25K/fpm during the final 75 seconds (radar derived by three ARSRs) before impact; 2), the flight path angle exceeded -70º during the final 60 secs of descent; 3), that the two debris fields were some 400m apart; 4), that the area of larger debris field was described as only 85m x 75m, and 5) #1 engine detached during descent to be located in the smaller west debris field.

    My assessment is that separation of the flaperon, and the adjacent flap, is a consequence of an event during descent but before impact of the greater part of the aircraft.

  238. Nederland says:

    @Don Thompson

    We have to agree to disagree then: the French conclusion from the same evidence is not that the flaperon was “dragging along the water” but that it came off violently on impact with water (not in flight).

    (Btw, I still don’t understand why the lower third of the flaperon would come off simply as a result of aerodynamic failure).

    The purpose of this blog is also to discuss the possible crash site, and the failure mode of the flaperon and other pieces is certainly important to do that, as the discussion and responses to it clearly show.

    I’m not trying to proselytise, just explaining what I think happenend at the end of flight based on the present evidence as I understand it.

  239. Richard says:

    @Nederland

    Again you state “the French conclusion”.

    The French did NOT come to a conclusion! There was only an UNCONFIRMED HYPOTHESIS, that the damage was due to impact with water.

    Again I quote the DGA report to you: “The little data supplied by Boeing did not enable the examination to be progressed by making calculations that would have made it possible to confirm or reject the proposed hypothesis.”

    You are being obtuse.

  240. Nederland says:

    @Richard

    We are going in circles and you keep on quoting the same line without any context.

    The “hypothesis” was a controlled ditching. The information derived immediately from the failure analysis included the obervation that the flaperon likely (not certainly) came off when hitting the ocean, e.g. “Considering an impact with water as likely,” (same page). All of theses considerations then combined to come to the “conclusion” (heading of that section) that this could have been a ditch. If it wasn’t a ditch, then the underlying assumptions (i.e. came off on impact) are not therefore wrong.

    Again, this is my view, you are free not to share this view and have your own view. Fair enough.

  241. Richard says:

    @Nederland

    I and others, object to your unfounded conjecture, your obviously false statements, your bogus attempts to twist unconfirmed hypothesis into a firm conclusion, your changing the word “possible” into “probable” or “definitely”, your quibbling about the semantics of “high energy impact”, your repeated misleading attempts to describe the status of MH370 debris as intact, well preserved and much larger, when compared to other crashes.

    I do not object that you disagree.

    I and others, object that you are deliberately trying to mislead the investigation.

  242. Nederland says:

    @Richard

    That’s a very strong verdict when all I was trying to do is to explain why I think a glide after a steep descent is a likely end of flight scenario.

    At no point was I deliberately trying to mislead anyone, as you make it out. Your criticism is over the top.

    I stand corrected on the size of the Germanwings vertical stabiliser (an honest mistake which I have admitted), but that’s it.

  243. TBill says:

    @Nederland
    Here is a prior “skipping” crash scenario by me on flight sim that could explain the flaperon. It is a high speed/catastrophic glide scenario with one wing touching first. Victor might also have a video like this, if so his will be way better than mine.

    https://www.youtube.com/watch?v=YDn9B0fFqRU

  244. Nederland says:

    @TBill

    OK, thank you. Yes, something like that perhaps. Would you say this is controlled or uncontrolled (or can it be either one – the angle does not appear to be very steep, possibly/probably no match to the descent rates derived from the two BFOs…)?

  245. ArthurC says:

    A thought – in order for an aircraft to break up before reaching the surface of the ocean, its speed must highly exceed design limits.
    Would that not cause it to break up into a lot of pieces, which would, in turn, create a larger debris field with a higher chance discovery?

  246. Don Thompson says:

    @ArthurC wrote “in order for an aircraft to break up before reaching the surface of the ocean, its speed must highly exceed design limits.

    Not necessarily, if the descent was uncontrolled/out-of-control the airframe may be experiencing loads that it simply wasn’t designed to withstand. I referred above to the case of MSR990, an Egyptair Boeing 767. During its out-of-control descent an engine, and possibly more, separated from the airframe yet the two debris fields were both described as quite compact.

  247. Nederland says:

    On the other hand, Egypt Air 990 came down with full engine power and near supersonic speed, and this is why the left engine and associated parts came off in flight.

  248. TBill says:

    @Nederland
    The above skipping crash is just showing a high speed (flaps up) smooth glide to the water surface at 5% bank, after fuel exhaustion so the aircraft is gradually losing atitude and speed maybe at about 250 knots at the end. But it is perhaps illustrative, if the descent to the water was not so steep. Flight sim programs tend to not show a steep descent at fuel exhaustion, especially the simple PSS777 model used here gives a kind and gentle glide.

    Brings up a question in my mind if we have the pictoral view of Boeing’s official flight sim cases hitting the water? If so I think it might make a visual case for wing hitting first. It is little hard for me to “unsee” the video I posted without realizing it sort of might explain the flaperon.

  249. Nederland says:

    @TBill

    Ethopian 961 is also an example of an aircraft gliding with one of its wings hitting the water first, flaps up, with no fuel/hydraulics left:

    https://www.youtube.com/watch?v=WE2Yn0cipTY

  250. Andrew says:

    @Nederland

    RE: ”On the other hand, Egypt Air 990 came down with full engine power…”

    No, the thrust levers were moved to idle before the aircraft started to descend. The engine start levers were subsequently moved to cutoff and the thrust levers were firewalled, but the engines had already shut down (Eng 2 increased thrust for a very short time before it was shut down).

  251. Asian says:

    My question is: Military radars surely must have tracked this plane…australia, singapore, indonesia. thailand. maybe even a destroyer, aircraft carrier or awacs radar plane?

  252. Nederland says:

    @Andrew

    OK, thanks.

  253. Don Thompson says:

    @Nederland,

    Silly me, I actually thought you’d done some research rather than simply throwing out specious grijnions.

  254. Nederland says:

    @Don Thompson

    You may be aware that I had discussed these two incidents (Egypt Air 990 and Silk Air 185) in greater detail a couple of years ago on this blog. As far as I am aware I was the first to draw these comparisons. At which time I had studied both at some length. I have now mistaken Egypt Air 990 with Silk Air 185 (they are very similar).

  255. ArthurC says:

    @Don Thompson

    Thank you very much.

  256. Nederland says:

    @Don Thompson

    And btw, the main point is that Egypt Air 990 (according to the report) came down at near “transonic speed”. Much of the report therefore is a somewhat apologetic discussion of why the left engine and some other parts came off at this rate (“The reason is that airplane performance and aerodynamic forces change in radical ways as the airplane’s speed approaches 1.0 Mach as it did here.”, p. 42). Whatever the exact reason for this, I doubt that MH370 came down at such a speed in the unpiloted end of flight scenario.

  257. Victor Iannello says:

    @Nederland said: You may be aware that I had discussed these two incidents (Egypt Air 990 and Silk Air 185) in greater detail a couple of years ago on this blog. As far as I am aware I was the first to draw these comparisons.

    I did a quick search of the blog comments to see when the comparison with Silk Air was first discussed. The earliest comment was this one from me on May 8, 2017:

    When considering the possibility of separation of parts from MH370 due to overspeed in a steep descent, it is useful to consider the recovered parts from SilkAir 185 in Dec 1997, which was flying from Jakarta to Singapore, and descended at high speed into the Musi River in Southern Sumatra. The US NTSB concluded that the captain deliberately lowered the nose and crashed the plane. Indonesian investigators disagreed, saying the evidence was inconclusive.

    Prior to the crash into the river, parts separated due to the high speed, and were found on land some distance from the crash the site. Therefore, we can compare the type and condition of parts expected from a high speed impact versus separation from high speed prior to impact.

    These pages, taken from the investigative report, show pictures of the recovered parts. Fragments of composite honeycomb panels recovered from the river resemble similar panels from MH370, and are shown in Fig 8b. By contrast, parts of the horizontal and vertical stabilizer recovered on land are shown graphically in Fig 10a and photographically in Fig 11. The size and condition of these parts resemble the flaperon and flap track fairing recovered from MH370.

  258. Nederland says:

    @Victor

    I don’t want to claim any credit in this regard. A quick search reveals that I had also posted on this (not on your blog, but at that time on the Jeff Wise blog), also in May 2017. I can honestly not recall having read it on your blog first, I’m actually quite sure that I had posted on this long before (maybe not in English).

    http://jeffwise.net/2017/05/23/reading-the-secrets-of-mh370-debris/comment-page-1/#comment-206404

    I was just saying that I had studied these and other cases long ago in some detail.

    The comparison to these incidents is nothing new anyway, the new element at that time was the possible failure mode of the flaperon.

  259. Nederland says:

    Here is a post from 13 March 2017 (7:52 am, and the following posts), it seems to have disappeared now, but is cached in google:

    http://webcache.googleusercontent.com/search?q=cache:UKH6hLxLK8cJ:jeffwise.net/2017/02/23/first-mh17-perpetrator-identified/comment-page-3/+&cd=1&hl=en&ct=clnk&gl=de&client=firefox-b-d

    “Nederland says:
    March 13, 2017 at 7:52 am

    @Jeff Wise

    In this case (Egypt Air 990):

    http://www.webcitation.org/5zlFg31jj?url=http://www.ntsb.gov/doclib/reports/2002/AAB0201.pdf

    parts of the left wing came off, were found in a separate debris field and in better condition (p. 34-36)”

    and so on.

  260. Victor Iannello says:

    @Nederland: The comparison of MH370 to Silk Air is obvious, and not just because of the condition of the debris. Others certainly had made the comparison before my comment on May 5, 2017. I just thought it was odd for you to make a point of saying that this was a new insight that you brought to this blog.

  261. Nederland says:

    @Victor

    I thought it was odd that Don Thompson wrote I hadn’t studied these cases.

  262. TBill says:

    @Nederland
    Ethiopian 961 is reasonable comparison for a glide scenario at 175 knots though I am thinking MH370 would have hit >200 knots on a hard glide.

  263. Don Thompson says:

    @Nederland,

    In the case of an honest error concerning the state of MSR990’s engines, I’m happy to disown my sarky reply above.

    However, I continue to hold the position I have set out in dialogue above. For my part, I will park discussion for now.

  264. Nederland says:

    @Don Thompson

    Not to worry. We all have stressful daytime jobs, and honest mistakes can happen.

  265. TBill says:

    @Asian
    “My question is: Military radars surely must have tracked this plane…australia, singapore, indonesia. thailand. maybe even a destroyer, aircraft carrier or awacs radar plane?”

    We have a variety of radar sources up to 18:22 just past waypoint MEKAR. After that time, the only active radar the public is aware of is Singapore airborne radar operating in the vicinity. The public has not been advised if the Singapore radar data provides any additional info about where MH370 was after 18:22, or where it was not seen. Indonesia military radar is generally presumed to have shut down for the weekend.

  266. Nederland says:

    Actually, at that time in early 2017 I was looking systematically for incidents where parts of the aircraft came off before crashing. This coincided with incidents of pilot suicide (Egypt Air 990, Silk Air 185), The only other incident I could find was China Airlines 006. I can not recall that another incident came up in that discussion.

    Egypt Air and Silk Air have in common that they both came down at near supersonic speed (either because of engine power or because of an arcane manoeuvre that goes beyond me, e.g. Silk Air: “The evidence showed that the airplane had exceeded its flight envelope during its high-speed transonic descent.”).

    China Airlines 006 was a case of dive/recovery, again with engine power. It experienced excessive g forces, and the landing gear door and parts of the empennage came off, perhaps during the recovery manoeuvre. The aircrafts were manufactured in 1997 (Silk Air), 1989 (Egypt Air), 1982 (China), MH370 in 2002.

    (There is one incident where a flaperon came off during a normal landing in Frankfurt)

    In the case of MH370, on the other hand, the common assumption is that the engines flamed out one after another, and the speed was then reduced, and there was probably no sudden event initiating the descent, which occurred at some point after both engines flamed out.

    All I want to say is that in-flight separation really is the exception. It is not immediately clear (to me) why MH30 experienced similar forces.

  267. Don Thompson says:

    @Nederland wrote: “in-flight separation really is the exception.

    Getting into a situation where both engines flame-out through fuel exhaustion would certainly be regarded as an exception. No matter what might be added to that situation, pilot input or none, the outcome is no more or no less exceptional. MH370 is right out there on the extremity of exceptional. Separation of flight control surfaces is consistent with exceptional.

    Rather than a “sudden event” you might consider a rapid and cascading sequence of conditions precipitating its descent.

  268. Nederland says:

    @Don Thompson

    I agree this is an exceptional event.

    However, the accident reports above are clear in their assessment that aerodynamic failure/flutter can happen but only at supersonic or near supersonic speeds.

    I don’t know if the maximum velocity of an unpowered 777 can reach that stage, regardless of angle of descent. (and if that is beyond flight envelope by design).

    In the above (post by Brian Anderson) and elsewhere I read about an (upper?) estimate for the Boeing design of 500 knots, which is not near supersonic speed.

  269. Victor Iannello says:

    @Nederland: If you are arguing that a no-thrust steep descent would not damage the airframe, that is wrong. Without a doubt it is possible to reach transonic speeds while descending with no thrust (whether or not the speed goes supersonic). Both Mmo=M0.87 and Vmo=330 KIAS would be far exceeded, and airframe damage would certainly occur at steep descent angles. The combined thrust of the engines doesn’t come close to the weight of the plane. Gravity is a very strong force.

  270. Sid says:

    Hi,

    I would like to withdraw the post I made on 14 July. I spoke without fully reviewing my old spread sheets and misstated what should be my current position.

    The best results that I got were for a start point at 18:22 at the last radar position with an offset cancelled (or perhaps not) at the FMT at 18:41 at IGOGU and then at 186T (overflying ISBIX) and continuing till the 7th arc at a constant M 0.82.

    This is a discrete path that is nominally tangent to the innermost arc.
    I have not considered fuel consumption as the model has never been updated to take into account all of the work that has been done by others.

    I believe that this resets my thinking to the state before the debris was found and before the first search program failed to discover the crash site.

    I thus remain in the camp that would be currently described as believing that either the site was missed during the search, or the plane glided after fuel exhaustion.

    I apologize for any confusion or consternation that I have caused.

    Sid

  271. Nederland says:

    @Victor

    OK, thanks. Basically a question. Would you say, though, this is likely? It sounds it can occur but only in certain conditions (steep descent angle).

  272. Victor Iannello says:

    @Nederland: The final BFO values show an increasing descent rate that is an average downward acceleration of 0.67g over 8 seconds. If there were no pilot inputs, the aircraft was banked, as I described in this post. If the wings were level, there likely was a nose-down pilot input. Either way, very high speeds and airframe damage would likely occur without pilot inputs to recover.

  273. Nederland says:

    @Victor

    Yes, I had read that post. But the figures are only derived from Microsoft FSX? Is this reliable enough to come to firm conclusions?

    (Sorry, not trying to annoy, just a genuine question)

  274. Victor Iannello says:

    @Nederland: The aerodynamic model is accurate enough to state with confidence that the 0.67g downward acceleration occurs during a banked descent OR with pilot nose-down inputs. Also, in the Boeing simulations that reached this level of downward acceleration, the descent was banked.

  275. Nederland says:

    @Victor

    And would that level of acceleration be sufficient proof that the final speed could be in the area of ~0.9M?

    “Also, in the Boeing simulations that reached this level of downward acceleration, the descent was banked.”

    So that would mean unpiloted?

  276. Victor Iannello says:

    @Nederland: Depending on the altitude, either Mmo or Vmo would first be exceeded during the accelerating descent. Either way, the speed limit would be likely exceeded and airframe damage would result.

    Yes, all the Boeing end-of-flight simulations that we have were for no pilot inputs.

  277. Nederland says:

    @Victor

    OK, thanks.

    I have to say I remain a bit sceptical on how reliable the FSX results are. I have also read about the Boeing simulations on the accelaration/BFOs. That seems clear enough, but I’m not sure if Boeing would want to put information on aircraft design, final speed in relation to free fall, and aerodynamic failure (an obvious safety issue) at any rate out in the public.

    Even in the reports above there is a lot of doubt on how reliable Boeing’s own high-end simulators are to come to conclusions like these (e.g. Egypt Air 990, p. 40: The simulator data are based on wind tunnel tests and updated with available flight test data. The maximum Mach number for which the simulator is programmed (Mach 0.91) corresponds to the airplane’s never-exceed airspeed. The maximum speed calculated for the accident airplane during the accident sequence was Mach 0.99 at 0150:23)

  278. Victor Iannello says:

    @Nederland: Of course all simulators lose accuracy at the performance limits of the aircraft, i.e., at high speed, high angle of attack, etc. That said, if you want to believe that a B777 in a steep descent won’t experience damage to the airframe, I’m done trying to persuade you otherwise. Perhaps others will want to continue with you. There is a reason why aircraft have speed limits, and those limits are exceeded in a steep descent.

  279. Nederland says:

    I may be wrong, but as far as I’m aware the aircraft is designed aerodynamically so that its final speed in free fall (air resistance) should not exceed the airframe envelope for obvious safety reasons.

    I’m not saying that damage to the airframe is impossible, but perhaps not something to expect all too readily in the probable end of flight scenario (i.e. engine flame out one after another, and a period of continuous glide before the descent).

    Other opinions are of course welcome.

  280. Don Thompson says:

    @Nederland

    You wrote that you had researched the case of MSR990, EgyptAir reg SU-GAP.

    Did your reading of NTSB sources on the crash delve any further than the Air Accident Brief NTSB/AAB-02/01 (PB2002-910401)?

  281. Nederland says:

    @Don Thompson

    Can you please come to your point. Thank you.

  282. Nederland says:

    Only material in the public domain of course.

  283. Hank says:

    @Nederland

    There is no such design constraint. An aircraft can break up. This happens.

    Unpiloted flights end two ways.

    They bank and the nose drops and this steepens into a high speed, ever tighter spiral dive. Aircraft could break up at transonic speeds.

    Alternatively if the nose pitches up the plane could stall and the aircraft would spin at a more steady vertical path.

    More likely the nose dropped and a classic spiral dive happened for MH370. Unless an active pilot flew it down. Without any fuel the RAT provide hydraulic power for the primary flight controls so it is possible to pull out of a graveyard spiral and glide.

    Options are breakup in spiral dive, dive into the water, spin into the water, or be flown into the water.

  284. Don Thompson says:

    @Nederland wrote “as far as I’m aware the aircraft is designed aerodynamically so that its final speed in free fall (air resistance) should not exceed the airframe envelope for obvious safety reasons.

    The Air Accident Brief sets out “after the airplane’s FDR and CVR stopped recording, the airplane descended to an altitude of about 16,000 feet msl, then climbed to about 25,000 feet msl and changed heading from 80º to 140º before it began its second descent, which continued until it impacted the ocean“.

    It was during the second descent that parts of SU-GAP’s structure separated.

    The FDR and CVR stopped recording after the engines were shutdown.

    Putting those events in order through an elapsed period of 128 seconds: engines shutdown, recorders ceased & aircraft reached 16,000ft, regained 9,000ft to reach 25,000ft, began final descent (unpowered and during which parts of airframe separated), impact.

    And the spoilers were deployed three seconds into that sequence, presumably, to arrest some acceleration.

    My point is that your research appears to have missed some detail.

  285. Victor Iannello says:

    @Nederland: I may be wrong.

    There is no ambiguity. You are definitely wrong. Read how Vd is defined, for instance. That’s not the maximum achievable speed in a dive. It’s the dive speed beyond which things break up.

  286. Nederland says:

    @Hank, Don, Victor

    OK, got it.

  287. TBill says:

    @Hank
    I do not agree steep spiral dive is automatic outcome of unpiloted fuel exhaustion. That’s one of the big questions for MH370. My understanding, and my own limited FSX sim cases, the aircraft will tend to fly in “lazy” circles to the water surface, and *not* go into a tight spiral dive.

    Victor has previously shown in FSX that a tight spiral could be forced to happen by holding the control surfaces at a bank, so when the power goes out, the bank is preset. But Boeing’s cases (which did not force a bank) do not show this immediate hard bank behavior.

    Alternately, DonT has suggested previously, maybe the pilot made some other less standard settings changes, that forced a spiral dive to happen at the end. Also Victor has suggested (correctly) that the sequence of events at fuel exhaustion is so complex, that maybe the Boeing simulator just cannot capture why a spiral dive developed. The other possibility is the pilot intentionally nosed the aircraft down for a violent crash around Arc7.

    But all of they above is trying to argue that a fast spiral dive did develop unexpectedly for some reason, when a less violent descent would tend to be the expected base case. So the other possiblity I think is that there was not a tight steep, spiral descent, and the flaperon maybe came off when the banking jet hit the water.

    This my summary of the debate, as I understand it. What I want to do is go back to the Boeing simulation cases and see the max speeds that developed.

  288. Andrew says:

    @Nederland

    In case you’re wondering why an aircraft might break up at very high speed, it’s largely due the extreme buffeting caused by shock-induced airflow separation. At transonic speeds, the airflow is already supersonic over parts of the wings and fuselage, even though the aircraft is still subsonic. Under the right conditions, it’s possible to see the shadow of the shockwave that forms on top of the wing of an airliner cruising faster than about M0.80. As the speed increases further, the shock wave becomes stronger and eventually the airflow behind the shockwave separates and becomes very turbulent. The empennage suffers most of the buffeting and is typically the first part of the aircraft to become damaged and separate in a high-speed upset event.

    The Adam Air PK-KKW accident investigation report is another interesting case where the aircraft began to break up at high speed and subsequently impacted the water (http://knkt.dephub.go.id/knkt/ntsc_aviation/baru/Final_Report_PK-KKW_Release.pdf). Have a look at the photos of some of the debris on pages 71-74.

  289. Hank says:

    @TBill

    You are correct.

    There are two lateral-directional stability modes at play – Spiral Mode and Dutch Roll Mode. Both are always present. The spiral mode is more dominant in light planes – and I know this mode well from experience.

    The Dutch roll dominates in jet transports because of the highly swept wings. The wing sweep significantly increases the effective dihedral. It is this mode that you see in a hand thrown model glider with very large dihedral.

    In Dutch Roll if the aircraft is disturbed in yaw or roll it rolls back the other way and “wazzles” around (scientific term) This is so uncomfortable all transports provide a yaw damper to prevent this from happening – makes passengers sick in the back.

    But with yaw damper off it is possible that the 777 just flies around in lazy descending circles.

    So maybe the 777 “wazzled” down in a Dutch Roll and hit the water at low angle. Or maybe it fell into the spiral mode at some point and became a “lawn dart.” At some bank angle the “roll into the turn” forces may offset the “roll out of the turn” forces and then the spiral then dominates. I suspect it is not possible to model how such a transition happens, if at all.

    You are right that the aircraft may not have rolled into spiral dive and may have “wazzled” down in a Dutch Roll or transitioned into a spiral dive at some point based on some disturbance that increased bank angle rather than rolled it back,=.

  290. Nederland says:

    Thanks for all the further explanations! I appreciate it.

    @Andrew
    Thanks for pointing me to the Adam Air report.
    I find it interesting that the elevator (p. 72) does not show damage at its lower end, like the outboard flap and flaperon. The flight spoiler (p. 73f) does show damage like that, but you can clearly see (I think) that this is fatigue damage rather than a one-off impact. It looks far more like it came off in flight than a lot of parts from MH370, not just the two.

    @Don Thompson
    Egypt Air 990 reached its highest speed at 0.99M just before the engines were shut down. I couldn’t take from the report the exact time when the parts associated with the left engine came off other than that this happened still in flight.

  291. TBill says:

    @Nederland
    The other thing I just noticed reviewing Egypt Air 990 is that the voice and data recorders stopped when the engine/power was cut. This is what I am suggesting for MH370, that the pilot may have cut all power at IGARI to stop recording what was a about to happen (depressuring).

  292. Nederland says:

    @TBill

    Yes, I noticed that too. The question is if that was the reason to shut down the engines at that time?

  293. Don Thompson says:

    @TBill suggests “that the pilot may have cut all power at IGARI to stop recording what was a about to happen (depressuring)

    All power? Isolating the Transfer Buses from supply?

    That’s a very big increase in pilot workload for very little gain (if any).

  294. Don Thompson says:

    @Andrew,

    Thank you for the reminder about PK-KKW.

    An observation: in the case of both PK-KKW/DHI574 and SU=GAP/MSR990 the major structural failures occurred some time after altitude rate of change and recorded mach had peaked.

    In the case of PK-KKW, some 15 sec after RoD+mach peaked, the DFDR recorded a positive to negative load-factor transition coincident with the CVR ‘thump-thump’ event that is assumed by the NTSC to indicate structural failure.

    The SU-GAP/MSR990 timeline is longer and a structural failure in the air is assumed by the separate ‘west’ debris field where the #1 engine was located. Time from peak rate of change in altitude + peak mach until the final descent and impact is in the order of 80s. There is no recorder data for that 80s of SU-GAP’s descent, only radar data. I’d expect the separation of the #1 engine to have occurred late in the sequence as the two debris fields were only 400m apart. Mach and RoD is certainly much reduced after the peak and during that final 80s.

  295. Victor Iannello says:

    @TBill: To follow up on @Don Thompson’s comment, I can’t imagine that any pilot would choose to fly an hour on RAT power, if that’s what you are asking. It is possible that the CVR was unpowered by isolating the left main and transfer busses by flipping the bus tie, the IDG, and the backup generator switches, leaving the right busses (main and transfer) functional. This would keep the flight control mode as NORMAL and also allow the autopilot to be engaged.

  296. Nederland says:

    Andrew wrote:

    I’d expect the separation of the #1 engine to have occurred late in the sequence as the two debris fields were only 400m apart.

    An inference, although plausible (I feel that shutting down the engines at transonic speed is the kind of “disruptive event” that I feel is missing in the presumed sequence for MH370 and could explain why one of the engines came off at some point.)

  297. Don Thompson says:

    @Nederland

    Again, you are not paying attention.

  298. Richard says:

    @Andrew

    Many thanks for the Adam Air PK-KKW accident investigation report.

    As you say an interesting case where the aircraft began to break up at high speed and subsequently impacted the water (http://knkt.dephub.go.id/knkt/ntsc_aviation/baru/Final_Report_PK-KKW_Release.pdf).

    As you recommended, I took a look at the photos of some of the debris on pages 71-74. I find the photos very revealing. The pictures are certainly damning evidence, for those who believe in conjecture and complain at Don for making an inference from the short distance between the debris fields.

  299. Nederland says:

    @Richard

    I don’t feel anyone was complaining about Don’s inference from the distance.

  300. Nederland says:

    The parts on p. 71-74 don’t look like they came off on impact with water. They look like they came off in flight (with engine power). The accident report confirms that. (The aircraft disintegrated and was destroyed when it impacted the water at high speed and a steep descent angle.)

    This seems to confirm the possibility that the flaperon, outboard flap and other parts of MH370 did not come off on impact with water if this was also a high-speed steep descent (similar to AdamAir).

    My concern was the following:

    I find it interesting that the elevator (p. 72) does not show damage at its lower end, like the outboard flap and flaperon. The flight spoiler (p. 73f) does show damage like that, but you can clearly see (I think) that this is fatigue damage rather than a one-off impact. It looks far more like it came off in flight than a lot of parts from MH370, not just the two.

  301. Victor Iannello says:

    @Nederland: I think you now agree that the flaperon could have been damaged and separated from MH370 in a high speed descent. Correct?

  302. Nederland says:

    @Victor

    Yes, this is correct. This has never been the point I was trying to convey.

    In the above post I was arguing that I think it is unlikely that the flaperon and outboard flap (and other parts) came off on impact with water if indeed this was a high speed descent (that is, say >500 kts).

  303. Barry Carlson says:

    @Victor,

    “…the flaperon could have been damaged and separated from MH370 in a high speed descent”

    As the French hold the Flaperon and have apparently extensively investigated its failure mode, I presume that the cause of the trailing edge composite failure was investigated. That being the case, one can only assume that a high frequency flutter was ruled out due to the tell-tale carbon fibre break/shear evidence not supporting it.

    For the Flaperon to have left the wing while the aircraft was airborne, the wing must have failed independently outboard of the No.2 engine, with the Flaperon being ejected as part of that process. This could mean that the trailing edge collided with something else, otherwise the damage was secondary and occurred in the marine environment.

  304. Andrew says:

    @Don Thompson

    RE: “I’d expect the separation of the #1 engine to have occurred late in the sequence as the two debris fields were only 400m apart. Mach and RoD is certainly much reduced after the peak and during that final 80s.”

    I agree. The fact the aircraft levelled off and then climbed before its final descent seems to show there was a struggle between the two pilots for control of the aircraft and I assume that struggle continued until the end. My guess is the aircraft structure was weakened during the first descent and began to break up during the second descent, shortly before it impacted the water. It seems clear the Mach and RoD were lower during the second descent (as you mentioned), but I suspect the indicated speed was very high and probably well over the design limit.

  305. airlandseaman says:

    Regardless of the exact sequence of events prior to the main impact, the evidence points to a very high speed descent and in-flight separation of at least the flap segment and flaperon, and probably other external pieces. There is literally zero evidence supporting an attempted water landing.

  306. Barry Carlson says:

    Re: July 18, 2019 at 7:52 pm.

    The reference to “carbon fibre” was accidental. It should read as ‘tell-tale composite break/shear’. My bad!

  307. Nederland says:

    Presuming that the ATSB outboard flap analysis is correct that the flaps were most likely not down (because there was no fuel left), but in neutral position, then two possibilities remain, right?

    1) Impact at very high (transonic?) speed. In this case, the flaps and other parts must have seperated in flight. They can not have come off as a direct result of impact with water because otherwise they would be far more damaged.

    2) Impact at relatively low speed following an extended glide (~200 kts?). In that case, no parts came off in flight, the aircraft broke up on impact, but in a less violent way.

    A mixed scenario where one half of the aircraft came down at very high speed, the other at moderate speed can be excluded (unless the entire aircraft broke up in flight).

    Are we on the same page here?

  308. Don Thompson says:

    @BarryC

    The DGA-TA stated that they had not received detailed spec of the assembly from Boeing so as to construct CAE models & run tests. DGA-TA did undertake SEM inspections using samples from along the ripped/fractured composite skin edges.

    I’m perplexed that so little attention is given to the ‘metalwork’. The damage to the actuator attachment lugs and the adjacent ribs, and the missing ‘hinge plate’ portion, could yet provide important clues. Certainly corrosion has occurred on the aluminium alloy that limits the possible inspection techniques but characteristics of the fracture are evident.

    The same applies to the outboard flap segment. That flap, spanwise, comprises four composite torsion boxes with a two piece trailing edge wedge.

  309. Barry Carlson says:

    Thanks Don.

    Yes, and with work undertaken with remote analysis by @David, and @Kenyon and others, I am inclined to believe the right wing broke up in flight and the Flaperon trailing edge composite damage wasn’t directly caused by over-speed but indirectly after the event.

    Whereas, the “damage to the actuator attachment lugs and the adjacent ribs, and the missing ‘hinge plate’ portion” was most likely concurrent with the outer right wing breaking-up.

    My impression has been over the past 2 years or more, that the aircraft impacted the sea 30° AND with RH side low, i.e. the drag creating parts, tail and left wing trailing high.

  310. Nederland says:

    @Barry

    What do you think could be the cause that the Flaperon trailing edge composite damage wasn’t directly caused by over-speed but indirectly after the event.?

  311. Don Thompson says:

    @BarryC

    […] and with work undertaken with remote analysis by @David, and @Kenyon and others, I am inclined to believe the right wing broke up in flight

    It’s my long held opinion that what you describe is a likely occurence. However, I would emphasize ‘in descent‘ rather than ‘in flight‘. A consequence of an uncontrolled descent.

    The most significant consequence, in the context of the seafloor search, is the implication for target recognition in the side scan/SAS images.

  312. Richard says:

    @Nederland

    You obviously recognise the importance of the MH370 end of flight analysis. This has consequences for the search area definition and the post FMT flight in general. The analysis of the MH370 debris is an important part of the end of flight analysis.

    Your discussion started with the premise, MH370 likely recovered from a steep descent and is likely > 40 NM beyond the 7th Arc. You pointed out the DGA analysis excluded the Flaperon parted from the aircraft due to flutter before impact with water. You argued debris is well preserved implying a low speed impact.

    You continued your discussion with comparisons to the debris from German Wings, Ethiopian 961, Egypt Air 990, Silk Air 185 and China Airlines 006. You maintained repeatedly Flutter is not a possible cause of separation in the descent of MH370. You argued Flutter only occurs at supersonic or near supersonic speeds, despite the Adam Air 574 report.

    Now you agree that the MH370 Flaperon could have separated in a steep descent, but this was never the point you were trying to convey. You are now saying either the Flaperon separated in a steep descent or there was a recovery from the descent and a low speed ditching. A high speed impact with water can be excluded.

    You could have spared us the run around in the 12 comments to the contrary.

    I disagree with your new conclusion. A high speed impact with water is possible and possibly in combination with Flaperon separation due to Flutter.

  313. Nederland says:

    @Richard

    I think you misunderstood some of my posts.

    Yes, my point is that the DGA hypothesis is that the flaperon did not separate in flight but on impact with water (not yet confirmed, but this is a carefully worded text because of its implications and because it was written as part of a judicial enquiry).

    There are other reasons which I think could support the above scenario.

    I am aware that the end of flight is important. I’m posting about this in good faith.

    A separate argument was that an unpowered descent is less likely to cause in flight break-up than a descent with engine thrust. This has been partly refuted (perhaps less likely but still possible) and I am happy to agree that at least in the uncontrolled end of flight scenario separation due to flutter is a possibility.

    You wrote:

    I disagree with your new conclusion. A high speed impact with water is possible and possibly in combination with Flaperon separation due to Flutter.

    This is not a new conclusion. On the contrary, this is what I started with. My point in the above was that *IF* this was a high speed impact (e.g. uncontrolled spiral descent), then the flaperon (and outboard flap) must have come off still in flight/during the descent but before impact. I have never changed the overall argument.

    Is this clearer now?

  314. Richard says:

    @Nederland

    You are so contradictory and cannot even remember what you previously stated.

    When Victor, Don, Mike, Andrew and others, all disagree with you, then it is pathetic to blame me for not understanding you.

    I give up!

  315. Nederland says:

    @Richard

    I don’t think we are that far away, other than in terms of assigning possibility/probability estimates to certain scenarios.

  316. TBill says:

    @Victor
    @Don Thompson
    Yes I propose all LGEN RGEN power off at IGARI. Two reasons:(1) I propose a hypothesis that plausible deniability was a prime objective (below) and (2) I think it may help explain what we see at IGARI and elsewhere.

    Yes possibly, the RAT extended at IGARI (but I tend to prefer not so) or APU was put on at IGARI and Overhead panel was used to cut power to the L and R buses to keep the digital data recorder OFF. Perferably the potential visual evidence at Penang of “spaghetti” of the drop down O2 mask could be stopped, but that is optional, and of course the B777 now has lots of extra power for higher altitude with all GENS off and bleed air off too. I do not know if he we might actually disengage one or both Generators, which is irreversible, or cut them out electronically. Of course I am not claiming to be an expert on how the electric system works in such unsual mode, rather I submitted a hypothesis. Via FSX/PMDG I see I can still seem to fly manually though.

    How to conduct a deniable hijacking was my MH370 5th anniversary essay-
    https://twitter.com/HDTBill/status/1104562266383548417?s=20

  317. Nederland says:

    ”Now you agree that the MH370 Flaperon could have separated in a steep descent”

    OK, I see, this is the “look out for contradictions” game.

    In the above, I agreed that it is generally possible that parts of an aircraft can come off in a high speed descent, even if unpowered. In theory, this can also be true for MH370.

    I never meant to say that I believe that the MH370 flaperon (likely) came off in flight, for the reasons I mentioned.

    I’m not entirely excluding this possibility, just saying that because of the DGA analysis, and other reasons too, especially the new finds since then, I now think this is not very likely.

  318. ArthurC says:

    If this is ultimately about manned flight at the end or not, I have an argument:
    Would the perpetrator try to land on water and risk dying of starvation or rather get it over with and crash hard?

  319. Nederland says:

    @ArthurC

    One possibility: The message that MH370 reconnected to the satellite at 0:19 came up on the screen, right?

    So, whoever was at the controls could have felt that this allows some tracking (which it did) and decided to fly a bit further, not in order to survive but in order not te be found.

    This could possibly explain the sequence steep descent/possible glide after the final satellite transmission.

  320. airlandseaman says:

    Nederland: In that event, it would have come up on the screen at 18:25 too, 6 hours earlier. So, that would not be news.

  321. Nederland says:

    @airlandseam

    I believe that powering up the SDU was a manual event, i.e. intentional (for whatever reason, this is still much debated) because the SDU was shut down first (presumably intentionally too).

    I don’t think anyone could reasonably expect that the satellite reconnects in the event that the aircraft runs out of fuel. This was likely unexpected/unintentional.

  322. Victor Iannello says:

    There are three EICAS messages related to the SATCOM:

    (1) SATCOM
    (2) SATVOICE LOST
    (3) SATCOM DATALINK

    When the SATCOM is powered down, all three messages appear. When the SATCOM is powered up and logged into the Inmarsat network but not logged into an ACARS server, only (3) appears. At both the log-on at 18:25 and 00:19, (1) and (2) would appear before the log-on and disappear after the log-on, leaving only (3). It could be that the disappearance of (1) and (2) precipitated a pilot-initiated steep descent, but as Mike said, that condition also occurred at 18:25.

  323. airlandseaman says:

    Re “I don’t think anyone could reasonably expect that the satellite reconnects in the event that the aircraft runs out of fuel. This was likely unexpected/unintentional.”

    That statement is 100% false. Nearly everyone does expect it.

    It is virtually guaranteed that the the AES will reboot following main engine fuel exhaustion (MEFE). It is a natural and predictable event, caused by the momentary loss of Left Main AC Bus 115VAC power (to the AES). It was explained in detail by ATSB and Boeing. I observed it (APU auto start) in all my Nov 2, 2014 simulations. After MEFE, there is more than enough fuel in the line to power the APU for up to 13 minutes…but at the very least 3 minutes… which is enough to explain the BFO and BTO data we do have at 00:19, and lack of IFE Logon data at 00:21.

  324. TBill says:

    @Nederland
    Yes maybe it was mistake the SATCOM got powered at the end, or maybe someone was playing with us.

  325. Nederland says:

    @Victor

    Thank you for clarification.

    I think 0:19 was an incomplete log on, does that change anything?.

    How much time do you think would pass between the messages at 0:19 or what would be the exact sequence?

    I’d say it’s more logical that a pilot who perhaps had the intention to hide the plane/evidence was concerned that (3) SATCOM DATALINK could give away his position still after so many hours of getting away from the intended destination. And that this could cause a change of plans. That means recover and glide to get further away from the possible zone of detection rather than continue the steep descent downwards and end close to that zone.

    For whatever reason the pilot had decided to go online again at 18:25 no doubt remains a mystery.

    But I think he was obviously not very concerned about this, perhaps because he was planning to continue to fly for a few more hours (with or without destination).

  326. Nederland says:

    @airlandseaman

    Say, if you had asked a pilot BEFORE MH370: “Will the SATCOM reconnect in case you run out of fuel”

    would his answer have been:

    It is virtually guaranteed that the the AES will reboot following main engine fuel exhaustion (MEFE). It is a natural and predictable event, caused by the momentary loss of Left Main AC Bus 115VAC power (to the AES).

    I’d doubt that. I thought most pilots were unaware how to shut down an SDU as this is not part of normal training. It is always easy to find out these things after they had occurred.

    Also, it is not entirely clear to me that the crash in the SIO was the intended aim of the diversion, so I wouldn’t necessarily presume pre-planning for that event.

    Or it could have been all intentional to obfuscate the search as @TBill thinks it is possible. If you assume that degree of planning.

  327. Victor Iannello says:

    @Nederland: It can be confusing, but when the EICAS message for a system appears, it means the system is having problems. So messages (1), (2), and (3) all indicate problems. When (3) appears, the pilot can be assured that an ACARS message will NOT reveal his position.

  328. Nederland says:

    @Victor

    I think the pilot turned off ACARS early in the flight, so he never worried about that.

    Nevertheless, you would think that a SATCOM connection (log on) that occurred accidentally tells the operator something about location (like distance to satellite).

    Better safe than sorry.

  329. airlandseaman says:

    Nederland: The AES logon at 00:19 was not incomplete. It was in fact a complete and successful, normal logon. The only thing that was not complete was the subsequent IFE logon expected at ~00:21. That did not happen at all, probably due to impact prior to that time.

    There is zero evidence the pilot flying had any awareness the restoration of power to the left AC bus would cause the AES to logon again at 18:25 (or 00:19). The AES was only one of many loads on the left main AC bus, so it is very unlikely the left bus was de-powered to affect the operation of the AES. It is far more likely it was de-powered in connection with the decompression circa 17:21 to 18:23.

  330. Nederland says:

    @airlandseaman

    I don’t think the possible reasons for why the SDU was shut down/went on first time are very important to understand the end of flight.

    The message came up on the screen, and I think there was some reasonable apprehension that this could give away some positional clues.

    The IFE could also have been shut off manually from the cockpit. In fact, this seems to be a reasonable possibility if the aim was to not allow passenger ground communication.

  331. Victor Iannello says:

    @Nederland said: I thought most pilots were unaware how to shut down an SDU as this is not part of normal training.

    Don’t believe that poppycock you read from somebody that thinks the plane is buried at Baikonur in Kazakhstan. First, a B777 pilot might not know what an SDU is, but he or she would be very familiar with the SATCOM (of which the SDU is one subsystem). That the SATCOM is unpowered if the IDGs are lost is stated in the FCOM. Also, in the PMDG 777 simulator (which the captain installed on his home computer), isolating the left bus will produce the three EICAS messages related to problems with the SATCOM. Whether or not the pilot isolated the left bus to depower the SATCOM, that is certainly a possibility.

  332. Nederland says:

    @Victor

    Does the simulator also show that SATCOM comes back on after running out of fuel?

  333. Victor Iannello says:

    @Nederland: No. With no fuel in the left tank, the APU does not start.

  334. airlandseaman says:

    But the SATCOM does come back on after MEFE in the level D simulator. The fuel in the APU line powers the APU for several minutes after MRFE.

  335. Nederland says:

    @airlandseaman

    Not sure if anyone would want to practice things like these on a work simulator and leave a lot of data for investigators to evaluate…

  336. Victor Iannello says:

    @airlandseaman: In some Level D simulators, the fuel in the line to the APU is not modeled. Boeing claims it is not modeled in the engineering simulator they used for the end-of-flight studies. There were also engine re-starts in the simulations you observed. This also did not occur in the Boeing simulations.

    @Nederland: I should add that the PMDG 777 simulator would be a poor choice to model what occurs after fuel exhaustion. For instance, in that model, at high enough airspeeds, the engine speed from windmilling was sufficient for the IDGs and backup generators to supply bus power. The transition from NORMAL to SECONDARY flight control mode is also not properly modeled. The same for the automatic start up of the APU using the fuel in the supply line. I was able to use the simulator for my end-of-flight studies only by manually forcing certain systems to behave in the way I believe they actually do behave.

  337. ArthurC says:

    @ALSM

    Wow, I always thought FS was a computer game. No sarcasm or disrespect intended, a sincere evaluation.
    A serious and realistically accurate application, but a “game” nevertheless.

    It seems that it actually replicates everything in great detail, if it even deals with aircraft behaviour in flame out or power out situations.
    Impressive!

  338. Don Thompson says:

    @TBill wrote “possibly, the RAT extended at IGARI (but I tend to prefer not so)

    If both Transfer Buses are isolated from supply, as you propose, there is no discretion for deployment of the RAT: it will happen.

    Neither is there any discretion for release of the cabin passenger O2 masks. The overhead panel unlock relays are supplied off the Standby AC Bus, the failsafe supply for the AC Standby Bus is the main battery via a DC-AC inverter.

    @Nederland,

    As Victor has described, SATCOM effects are notified on the EICAS. In the case of the short term power recovery following loss of engine driven generation, there will be many more messages than just those related to SATCOM scrolling up the EICAS display.

    The SATCOM system is not something that is unknown to 777 crews. The VOICE and DATA functions it provides to the Flight Deck, and the related EICAS messages, are well described in the Flight Crew Operating Manual.

  339. Victor Iannello says:

    @ArthurC: Mike was describing the simulations on a Level D simulator used by a major airline with B777s. That’s certainly not a game.

    As for MSFS just being a game, it is far more than that. Essentially, the PMDG 777 add-on uses the B777 FCOM is its design specification, and the FCOM is supplied as the “user’s guide” for the software. Of course it does not have the fidelity of a Level D simulator, but if you understand its limitations and use the tool appropriately, it can do a superb job in modeling aspects of a flight.

  340. Niels says:

    Re Cocos Keeling Islands:

    Would anyone know if Cocos Island airport would have been available at night for emergency landing around 7/8 March 2014?

    Another question: is/was there a radar facility around the airport? I have been looking with google maps and street view but haven’t been able to find one.

  341. Nederland says:

    Thanks, Don, no more questions at the moment

  342. Peter Norton says:

    Victor Iannello: “@TBill: To follow up on @Don Thompson’s comment, I can’t imagine that any pilot would choose to fly an hour on RAT power, if that’s what you are asking. It is possible that the CVR was unpowered by isolating the left main and transfer busses by flipping the bus tie, the IDG, and the backup generator switches, leaving the right busses (main and transfer) functional. This would keep the flight control mode as NORMAL and also allow the autopilot to be engaged.”

    Ok, but can the pilot be reasonably expected to know this?

  343. mash says:

    re: Hypoxia Physics/Physiology Question

    According to The Atlantic Magazine (WILLIAM LANGEWIESCHE):
    “None of those cabin masks was intended for more than about 15 minutes of use during emergency descents to altitudes below 13,000 feet; they would have been of no value at all cruising at 40,000 feet. The cabin occupants would have become incapacitated within a couple of minutes, … ”

    Thus the following re-consideration (numbo jumbo) :-

    What is the [practical] difference between a FL350-hypoxia and a, say, FL420-hypoxia?

    Is it comparable to the difference between :-
    a. A 35-floor and a 42-floor ‘jump down’ (free fall),
    b. A fifth and a sixth (*1/7),
    c. A 2.5th and a 3rd (*1/2),
    d. And so on … ?
    [Especially if the ‘jump down’ (to ground) actually started before the ‘climb up’ (to higher floors), so to speak. ]

    If not much difference, what is the [real] purpose of “the climb”, if any?

  344. Andrew says:

    @Niels

    RE: “Would anyone know if Cocos Island airport would have been available at night for emergency landing around 7/8 March 2014?”

    Put it this way, I don’t know of any reason why it wouldn’t have been available. It’s an ‘uncontrolled’ aerodrome (in the ATC sense), with pilot activated lighting and instrument approaches to runways 15/33.

    RE: … is/was there a radar facility around the airport? I have been looking with google maps and street view but haven’t been able to find one.”

    There is no ATC radar at Cocos Island. A radar was installed at Christmas Island in 2011 as part of an Australian DSTO initiative. That radar was intended for maritime surveillance, primarily to detect small boats used by illegal immigrants (https://www.securityinformed.com/news/co-10726-ga.12678.html). I don’t know if it was still operating in 2014. As far as I’m aware, the only other radar asset that provided coverage of the area was Australia’s JORN, operated by the Royal Australian Air Force. However, JORN does not operate continuously in peacetime and the RAAF claimed it was not operating the night MH370 disappeared.

  345. Andrew says:

    @Peter Norton

    RE: “Ok, but can the pilot be reasonably expected to know this?”

    Most B777 pilots would not know the power source for the CVR as it is not mentioned in the manuals that pilots normally use and it is not covered during training. Nevertheless, the information is accessible if a pilot chooses to do some background research. Having discovered the power source, a B777 pilot should have enough systems knowledge to de-power the right-side electrical buses. The ramifications of doing so should also be apparent to any pilot who chooses to do a bit of research.

  346. Ventus45 says:

    @Niels

    Re Cocos Keeling Islands:

    Would anyone know if Cocos Island airport would have been available at night for emergency landing around 7/8 March 2014 ?

    One could only assume yes, in the sense that the pilot can activate runway lighting by PAL on 118.1Mhz, but there is no tower or radar or ATIS.

    Ground services are provided by TOLL, see:
    http://christmasislandairport.com.au/wp-content/uploads/2017/10/REM-O-AIR-FOR-IOT-047-Airport-Fees-and-Charges.pdf

    See the AIP below.
    https://www.crc.id.au/xplane/charts/ERSA-2019-MAY-23/Cocos%20(Keeling)%20Island%20(YPCC)%20FAC.pdf
    Note:
    “All inbound and outbound ACFT arebrequired to activate RWY LGT by PAL + AFRU 118.1 to suspend use of RWS for non aviation activities and to allow personnel to vacate RWS.

    As for the “non aviation activities”- the video in the link below is worth a look.
    https://www.abc.net.au/news/2019-06-23/cocos-islands-golf-course-on-international-runway/11222500

  347. Peter Norton says:

    thanks Andrew

  348. George G says:

    @drB
    @Richard
    @Victor

    (and All),

    May I please “rephrase” your proposed means of determination of likely or potential areas (Regions Of Interest) for future search or re-search for the remains of MH370 so that I may then ask some questions.

    1. You propose construct of a multi-dimensional matrix.
    2. You limit that matrix to “arcs” 2 through 6 to so that end effects are possibly minimised in the analysis.
    3. Please note that (as I see it) the arcs are simply one dimension of the matrix.
    4.You chose to include Arc 6 within the matrix.
    5. Arc 7 was not included in the matrix due to it being clearly and distinctly different to the previous arcs as evidence points to it being part of the final descent and thus not part of the previous flight southwards, or generally southwards, to somewhere above the Southern Indian Ocean.
    6. It would be possible to also limit the matrix to within the bounds of only Arcs 2 through 5 if it was argued that potentially Arc 6 may also be part of the final descent, even if only at the start of that descent.
    7. The use of the matrix is to define a number (a large number) of possible paths generally southward which would define a potential path for MH370.
    8. To limit one dimension to a reasonable definition you have limited the “actual” path to a singular path started prior to Arc 2 and uninterrupted until Main Engine Fuel Exhaustion. This dimension by this limitation becomes the navigational path mode.
    9. Note that the “singular” path mentioned above in 8. is assumed to be uninterrupted as mentioned above.
    10. As mentioned above in 6. it would be possible to limit the matrix to within the bounds of only Arcs 2 through 5 to eliminate any questions, if any, that Main Engine Fuel Exhaustion may have occurred prior to Arc 6.
    11. It is theoretically conceivable that an alternate path or alternate paths other than a singular uninterrupted path could be added to the matrix provided that such alternate path or paths were mathematically definable. For example, a flight towards a possible land destination subsequently followed by another generally southwards could be added if such a path was reasonably proposed. Note the use of “theoretically conceivable”. This would be extension into another dimension, as I understand it. This is the dimension referred to above in 8. as the the dimension limited “to a reasonable definition”.

    AND, Finally: (I almost forgot):

    12. For each possible path through the matrix given mathematical constraints upon the possible values, or known values, for each dimension at each Arc Point along the aircraft flight path, THEN the derived theoretical values of the expected BTO (Burst Time Offset) from the satellite communications to and from the aircraft are compared to those actually recorderd by Inmarsat. This comparison thus determines the likelihood, or otherwise that this point in the matrix is a point of the actual flight path. Similarly the BFO (Burst Frequency Offset) values are analysed, or compared.

    13. Concerning “possible values, or known values” as mentioned above in 12. :-
    A: The time at each Arc crossing is a known value. Although, re-reading your method summary, it may appear that you have included this as a general variable. I suspect that the values for this dimension are limited to the known recorded times.
    B: You have used the initial bearing (at the time of crossing Arc 2) as another dimension. You have systematically and incrementally “scanned” this variable or dimension between an reasonable set of limits. You have chosen 155 degrees True through to 195 degrees True as the reasonable range, or set of limits.
    C: At subsequent arc crossings you have updated the aircraft or flight bearing dependent upon the Lateral Navigation Method.
    D: Other variables or dimensions are similarly systematically and incrementally “scanned” and updated as the ‘flight” progresses, resulting in a set of systematic and incremental flight paths as defined by elements of the matrix.

    Questions:

    Q1: For presentation purposes, specifically to aid in comprehension, are the “solutions” or the set of flight paths easily broken down into sets of (say) Extremely Unlikely, Possible, Probable, Likely, Most likely ? For example, my chosen variable, or dimension, above in 13.B., being initial bearing, might be a simple case. I suspect that your results (set of solutions or flight paths) for this parameter may be very distinct. Or there may be one distinct peak in likely results, perhaps even two or three peaks. For example, I would suggest that the likelyhood of a 155 degree result is minimal, in consideration of all the other parameters which contribute to the potential solution likelihood, and thus would be relatively classifiable.
    Perhaps a two dimensional graph for this parameter, variable, dimension, alone, may assist in comprehension.
    Similarly, other parameters, such as Flight Level or Speed Control Mode may have clear peaks ?

    Q2: You may or may not have already said this earlier. If so, please accept my apologies for asking now. You have already determined and/or declared that the BFO results can be “ignored” and utilising the BTO results alone leads to the same solution as when the BFO results are included in the analysis. Have you also constructed an analysis of only Arc 2 through Arc 5 and how conclusive is this ?

  349. airlandseaman says:

    mash: The climb after IGARI was from 37200 feet (FL350) to ~40000 ft. The difference in UTOC, as you suggest, is not much. In either case, OTOO 1 minute, after complete decompression…maybe 2-3 minutes if the constant flow masks were used. So, while the climb would decrease the time by a few seconds, it is unlikely that was the main reason for the climb. The stepped speed profile approaching KB (increasing in 3 steps from 460 at 17:31 to 530 kts at 17:38) suggests the altitude was higher than 40,000 at the top of the turn, and the PF was seeking to find the optimum altitude to maximize the ground speed.

  350. Niels says:

    @Andrew, Ventus45
    Thank you for your explanations about Cocos Keeling airport. @Ventus45: the video indeed is clarifying regarding non-aviation activities; the atmosphere at the islands seems rather laid back.

    The reason I’m interested is that my MH370 flight path calculation results typically indicate a settling of fixed cruise conditions in the 21:00 – 22:00 time window, in the vicinity of Cocos Keeling.

  351. Nederland says:

    Two more thoughts on the end of flight scenario:

    Leaving aside the flaperon for now (I have yet to hear of a reason why the lower third of the trailing edge came off if not on impact with water: eaten by a shark?, bumped into a goldfish?, same for the outboard flap)

    then we still have the other ~30 pieces mentioned in the Debris Examination Report.

    Obviously, these are badly damaged. But do they really look like parts of an aircraft that came down with such high speed that it completely disintegrated on impact? Such debris is shown in the Silk Air 185 report, 229-232, figure 8a-h. Maybe some, but not all.

    One could therefore hypothetise that some came off in flight, others on impact. But description of the damage is almost always the same along the line: tension damage/pulled off, core of structure not crushed, fractured along the side. To me, this does indicate that they separated because of a one-off event (suddenly on impact) rather than due to aerodynamic overload (slowly/gradually); and that the impact was not as violent as in the case of Silk Air (no core/crush damage).

    Take for example item #22 (above, p. 52-7), the vertical stabiliser. This appears to be the only item that is squashed internally, probably due to the impact with water (it did not come off in flight), although it would be the part best protected. On the other hand, aircraft parts, like the right fan cowl (#6, p. 1-4) show no (signifcant) crush damage on the core and can obviously not have separated in flight either.

    The third interim report therefore came to the conclusion: Recovery of the cabin interior debris suggests that the aircraft was likely to have broken up. However, there is insufficient information to determine if the aircraft broke up in the air or during impact with the ocean.

    To me, this sounds more like all recovered parts broke up at the same time, when the entire aircraft broke up, rather than that some came off in flight, others on impact.

  352. airlandseaman says:

    Nederland wrote: “I have yet to hear of a reason why the lower third of the trailing edge came off [of the flaperon] if not on impact with water: eaten by a shark?, bumped into a goldfish?, same for the outboard flap”

    What a ridiculous, 100% false statement. The subject has been discussed in depth for the last 4 years by scientists, engineers and others, here and elsewhere. It has been addressed by ATSB and the French. I published the first paper on the subject only 2 days after the flaperon was found 4 years ago. You are free to disagree with the assessment, but to state “I have yet to hear of a reason…” is pure fake news. http://bit.ly/2DKBxuf

  353. Nederland says:

    @airlandseaman

    Thanks for your reply.

    I have to say I can’t remember all drop box files, posted 4 years ago, but I had repeatedly asked in the above about the reasons why that part came off.

    Also I’m afraid I can’t open your file.

    From the French report I could only find the information that it came off on impact with water. I can’t remember an ATSB report on this, as this was part of the French, not the ATSB investigation.

    I’d thereful be grateful if you could point me to those reports (French/ATSB).

    Thank you in advance.

  354. TBill says:

    @Victor
    Victor Iannello: “@TBill: To follow up on @Don Thompson’s comment, I can’t imagine that any pilot would choose to fly an hour on RAT power..”

    How about just no AutoPilot to Penang or even less time? Right now I agree with you that A/P was probably back on by Penang. I am suggesting proactive actions to obfuscate the intentional nature of the apparent intentional depressurization near IGARI. I have cited various possible ways to do that, but temporarily or permanently cutting off DFDR power from cockpit or MEC Bay is one possible technique.

    I would not automatically rule out flying after IGARI with RAT deployed, but it is certainly untidy based on our current assumptions saying pilot wanted a completely normal aircraft to fly for 7+ hours. If you say BFO values were off by a constant 10 units (say due to upsets and 18:25 restart) there might be an argument for a slow curved path roughly ISBIX to 22 South…not sure how the N571 OFFSET looks if BFO is +10.

    @mash
    The greatest unknown in the hypoxia severity is cabin altitude, because it takes a while to get all the pressure out of the aircraft. I presume if the pilot was being deliberate/fast he would cut bleed air and open outflow valves. That would probably be fast, but that’s a weakness of MS flight sim programs, in my experience they show depressurization being fairly slow. If you could really get the air out fast, then at FL400 you probably could get to a point of rapid loss of consciousness depite O2 masks.

    @Nederland
    In addition to what Victor said, I never say the word “SDU” anymore. Instead I say “SATCOM” which is the actual terminology. This because JW says pilots do not know the “SDU” exists…but that is word-play, because pilots do know all about the “SATCOM”.

    As far as your idea the pilot not knowing his specific nefarious destination, that is of course possible but it seems many experienced pilot authors (Ed Baker, William L., Ewan Wilson, ALSM) feel the facts point logically to intentional depressuring at IGARI. The purpose of that action would presumably be to allow the planned mission to continue without interruption. I am certainly open to the negotiation theory, but it seems most likely any negotiation was over by the ZS cell phone call on the KLIA 32R runway.

  355. airlandseaman says:

    The link works fine for me. Here is a new link: http://bit.ly/2PdLoe6

    You should also read this: http://bit.ly/2XYJCqA

  356. Nederland says:

    @TBill

    I entirely agree that pilots generally know about SATCOM. I think that pilots would not normally know that SATCOM would come online if the aircraft runs out of fuel. Not even a high end simulator (like the ones used by MAS) seems to accurately show that.

    It would certainly be somehow possible to get this information in advance if one looks very closely into this. But it is obviously not guaranteed if the MH370 pilot did this. Even if he did, he could have missed that detail.

    (If he looked into these details so accurately, then he may deliberately have mislead with a possible steep descent, followed by a glide.)

  357. Nederland says:

    @airlandseaman

    Again, thank you.

    Your conclusion at the time:

    Part of the flaperon along the trailing edge is missing and the edge is ragged (looks to be “torn”, not broken off due to bending

    This is the exact opposite of what the French DGA wrote:

    The damage being greater on the trailing edge on the inboard side, notably on the lower surface side, it appears that the contact occurred first in this zone. The loads generated, pushing from the lower surface towards the upper surface locally (unlike uniform aerodynamic loading), resulted in a bending load from the rear towards the front as well as of inboard towards outboard. This caused torsion on the flaperon. etc

    I am aware of the tail strike hypothesis in the file by Kenyon. At the very least, this is an added assumption. If viewed in combination with the trailing edge damage on the outboard flap, it would be an amazing coincident if both hit the tail of the aircraft, and that latter kind of damage does really not look like a tailstrike.

    I apologise for the silly wording in the above and I appreciate that this is a sensitive topic, but I have to say that I would give more weight to the DGA analysis who have personally inspected that part, and this is a government body. As far as I’m aware the ATSB has never stated anything to the contrary either in publication or press statements.

  358. airlandseaman says:

    Nederland: OK. I’ll take that as an admission your statement above (“I have yet to hear of a reason why the lower third of the trailing edge came off …”) was not true.

    The French flaperon analysis is flawed in several aspects. I won’t repeat all of it again.

  359. TBill says:

    @Nederland
    Not sure you noticed you got an indirect response from @Gybreght on JW. if we look at the Boeing simulations, we could infer crash severity for a more level flying crash scenario, and I am thinking it can be quite severe depending on vertical and horizontal speed at water impect. Obviously the internal debris seat monitor frame and bulkhead semi-definitely disprove a Sully-style soft landing. But those debris items could possibly be explained by a notch more severe (ie flaps up fast glide).

    It is generally believed that IFE would reboot automatically even if the pilot had it turned off for the PAX. So for loss of IFE at end of flight reboot, I am rusty, but I think we have to invoke fast crash pulling breakers in the MEC Bay. The other reason for pulling the IFE breakers in my view is possibly to prevent the IFE flight computer from recording info such as flight path.

  360. Nederland says:

    @TBill

    Thank you for pointing me to the Boeing simulations.

    On the IFE, in the ATSB report, p. 10 one assumption was: The fact that the expected IFE system transmission was not received could be due to: • the IFE system being selected off from the cockpit overhead panel at some point after the 18:25 logon

    They also point to the possibility of an unusual aircraft attitude. So, it’s not just that the plan crashed (according to this report).

    @airlandseaman

    I was simply saying that I repeatedly asked for a reason why the trailing edge broke apart.

  361. airlandseaman says:

    If the reason for the missing IFE logon was due to an inverted aircraft, clearly it was within seconds of impact.

  362. Nederland says:

    True enough, but the possibility that the IFE was switched off seems plausible to me.

  363. Victor Iannello says:

    @TBill: If you want to interact with Gysbreght, do it elsewhere, but not here. There was a reason he was banned from this blog.

  364. Victor Iannello says:

    @All: Here’s a really interesting video showing Apollo 11’s descent and landing on the moon. The annotations are great.

    https://www.youtube.com/watch?v=RONIax0_1ec

  365. Andrew says:

    @Victor

    Thanks for the video. It was a stunning achievement, especially given the technology of that era. I was 6-years old at the time and remember being glued to the television as it all unfolded.

  366. David says:

    @Victor. Gripping. Aldrin had a lot to do.

  367. mash says:

    @airlandseaman

    Thank you for the reply – so there is not much difference … for “the climb”.

    But then another possible question is:

    What is the (practical) purpose of the accelerations if it is not the ‘shortest distance’ path [to ‘escape’], assuming the detour around Penang is just a waste of time?

    [… without concluding prematurely that all these ‘meaningless’ moves is just a big deception show, say to create the impression of an emergency return/landing (at the ‘first’/’contingency’ stage), assuming all planned. And perhaps consequently to challenge the “decompression” theory, and so on … ]

  368. airlandseaman says:

    Re speed steps at KB: I’m not sure if I ever posted the final results from my KB PSR analysis that show the speed steps. So, here is one example assuming the altitude is 40,000 ft. As in the past, adjusting the assumed altitude moves the CoS average (single dot here) higher or lower. 40,000 is a good fit to the speeds before and after the CoS.

    Paul’s azimuth smoothing idea turned out to be the key to removing most of the radar position noise, which cleaned up the instantaneous speeds a lot. It makes the speed steps very obvious. http://bit.ly/2M2mbFu

  369. Brian Anderson says:

    @Nederland
    “To me, this does indicate that they separated because of a one-off event (suddenly on impact) rather than due to aerodynamic overload (slowly/gradually)”

    This statement suggests that you are not familiar with how sudden and catastrophic aerodynamic flutter can be.

    I suggested, in a post days after the flaperon was found, that flutter could explain the TE damage. Note that the TE separated along the line of fastenings which attach the top and bottom skins to the rear spar. There is no other internal structure to the rear of that spar. This is also the line of maximum bending moment, i.e. maximum stress in the skins, in the event of flutter. It’s the place that I would expect the separation to occur.

    There are a number of factors which can contribute to flutter occurring, not just the aircraft speed. Any play in the flaperon attachment points and linkages, inevitable as the airframe ages, will increase the possibility. What state were the actuators in at this point, and will they provide any damping?

    The onset of flutter can be very rapid, and within a couple of seconds the oscillations can reach destructive amplitudes and loads. I suggest that the TE damage could easily occur within 10 seconds, with simultaneous over-stressing of hinges and linkages, followed very rapidly by the departure of the remainder of the flaperon.

    If you doubt this, have a read of the reports of the Reno P51 accident in 2011, and other examples.

    Note too, that the B777 FCOM makes reference to the possibility of flaperon flutter while testing engines on the ground.

  370. Nederland says:

    Thanks, Brian

    My quote was in relation to the other ~30 parts. It looked more reasonable at that time to conclude that the flaperon came off because of flutter. Yes, 10 seconds could be right, that is still a marked difference from a sudden event at impact, and I believe the DAG is able to tell that difference.

    If the line where the trailing edge broke is the most vulnerable, then it is still the most vunerable if it broke on impact.

    The Reno P51 (‘Gallopping Ghost’) accident was an airshow gone wrong? Involving a modified WW II military jet? (During these events, the vertical acceleration peaked at 17.3 G … The accident airplane had undergone many structural and flight control modifications that were undocumented and for which no flight testing or analysis had been performed to assess their effects on the airplane’s structural strength)

  371. David says:

    @Don Thompson. Since the Boeing simulations take no account of the effects of APU start or of relight attempts, that suggests the data on level D simulators that do are not accurate enough for Boeing and begs the question as to where that data came from for others. For example, the availability of left tank residual fuel to the APU has been established by Boeing only since the crash.

    I assume that manufacturers of airline simulators would write their own software but if Boeing cannot simulate APU starts and relights what does that say about the fidelity of those that can for pilot training? Is it that their accuracy requirements are less? In that case they may be accurate enough for end of flight analysis here? If so, why would Boeing not have used one of these for their simulations?

    Assuming there to be accuracy requirements generated by the simulator industry or a regulatory body, such that airline simulators yield a common and sufficiently accurate outcome for like aircraft configurations, can you throw any light on whether they provide reliable answers for our purposes here when Boeing declines to use them?

  372. David says:

    @Dr B. Adoption of the ‘alternate’ electrical generator loading configuration, whereby the SDU/SATCOM re-boot initiation would occur at right engine failure, would result in;
    • a range reduction so shift north of the 7th arc crossing if right engine fuel quantity remains as in your modelling,
    • a slowing of speed to match 7th arc timing over the shorter distance and,
    • a maintenance of that speed until re-boot initiation, whereas in the ‘normal’ configuration there is a drop from its higher speed after that failure.

    That loading configuration offers a higher possibility of meeting BFO descent criteria.

    I think you commented on this before to the effect that this falls within PDA tolerances; though the change to usable fuel and time over the last leg do now look to be substantial. In effect some 580 kg of residual left engine fuel is foregone and the speed of the last leg would need to reduce by some 10 minutes as I understand it, the time between engine failures. There would be lower speed and less time between arcs generally but also less distance.

    There would be some small iteration with the effect of reduced speed on fuel consumption also.

    Should an increased electrical load have been carried by the right engine for some time, that would reduce its fuel, amplifying the first two dot point effects above.

    If not swept up by tolerances, conceivably there might be another persuasive course solution that would accommodate this change to assumptions?

    Should it be that it does fall within tolerances, the crash site might be expressed as a band along the arc rather than a point?

  373. airlandseaman says:

    David: Re: “Since the Boeing simulations take no account of the effects of APU start or of relight attempts, that suggests the data on level D simulators that do are not accurate enough for Boeing and begs the question as to where that data came from for others. For example, the availability of left tank residual fuel to the APU has been established by Boeing only since the crash.” This is nonsense. The level d full motion sim we used is incredibly accurate. The “data” comes from a high resolution dynamic math model that remains very accurate up to the point of speeds very close to mach 1. The APU auto start function was accurately modelled. Stop trying to explain away results you don’t like by making totally unfounded assumptions about the fidelity of the level d sim.

  374. Brian Anderson says:

    @Nederland,
    Yes, Galloping Ghost was an accident waiting to happen, but I reference it only to illustrate how quickly flutter can develop, and how catastrophic the result. In this case the trim tab was torn off within a second or two of the flutter developing.

    Guarding against flutter, especially of moving surfaces, such as ailerons and elevator and rudder, is the reason for critically balancing these surfaces. Improper balance, play in hinges and linkages etc will increase the possibility of flutter occurring. Even in a surface as large as the flaperon it still would not be surprising, with flutter, to have an oscillating frequency of say 10 to 15 Hz with diverging amplitude. Catastrophic.

  375. David says:

    @ALSM. If so accurate why Boeing not use it?
    Please be more open minded.

  376. Victor Iannello says:

    @ALSM, @David: I’ve also found it odd that the simulations on the airline simulator included behavior (engine relight, APU auto-start) that were not included in the Boeing simulator.

    I remain suspicious of the fidelity of ALL the Level D simulators in their ability to accurately model behaviors after fuel exhaustion with no pilot in the loop. Some of the limitations of the Boeing simulator are known. While the airline simulator included effects that were not in the Boeing simulator, it too had some unexplained anomalies. For instance, for the case in which two engines reached fuel exhaustion at the same time, the relight of the left engine caused a very steep bank to the right (as much as 55 deg) and a rapid increase in descent rate. Then, for some unexplained reason, the wings suddenly leveled, which arrested the rapid descent. @Andrew also observed an anomaly using his company’s simulator in which the left throttle did not fully advance (without his input) after the right engine reached fuel exhaustion.

    After reviewing all the simulations, my main takeaway is that if the aircraft reaches an increasingly steep descent (downward acceleration of 0.67g) as suggested by the final BFOs, and there are no pilot inputs, then the aircraft is banked and will impact the sea not far (<10 NM) from that point. On the other hand, a pilot may provide nose-down inputs to cause the same downward acceleration with level wings. With no recovery, the aircraft will again impact the sea not far (<10 NM) from that point. It is possible that the pilot executed a dive-glide-dive maneuver, which would satisfy the BFO evidence and the debris evidence, although that would be a strange sequence.

  377. airlandseaman says:

    David: The Boeing engineers used an engineering tool (with limitations), not a full motion, high fidelity training simulator. You will have to ask them why. Probably because of the large number of sims they conducted, and the convenience of the engineering simulator to their work. Your pure speculation that they did it because the engineering simulator was superior is baseless nonsense. It is worth noting that Paul Matson (777 pilot that arranged for the UAL sim time) was well aware that the APU would start automatically from his years of training.

    Don: Please inject some factual information into this conversation.

  378. airlandseaman says:

    It should be noted that, regardless of simulations, Boeing informed us that the 777 APU is designed to auto start when electrical power is lost. Thus, fhe simulations were consistent with what should be expected. Why, then, not trust the sim’s?

  379. Victor Iannello says:

    @airlandseaman: We know that engine relights and an APU autostart occurs after fuel exhaustion. I’m just curious why Boeing chose to use a simulator which did not model these known effects, regardless of whether or not the simulator was full motion (which is not required if there are no pilot inputs).

    In the simulation you witnessed with the double fuel exhaustion, the rapid wing leveling after the left engine relight (and steep bank to the right) remains unexplained. My guess is it’s a simulator error or unintended simulator configuration.

  380. airlandseaman says:

    Victor:

    I know the restart anomaly you are referring to, and we discussed it in real time. The trainer that was controlling the simulator was also puzzled. It looked to us like the computer reset the simulation due to encountering some unexpected state. But let’s be clear. That anomaly was not a common event in the 4 hours of sim’s we conducted, and importantly, when it did occur, it was obviously a computer glitch, not a simulation of how things really work. You have seen all the videos. You should be able to attest to the realistic behavior of nearly everything in those sim’s.

  381. Victor Iannello says:

    @airlandseaman: I focus on the engine relight because it is an event that can explain the downward acceleration level and timing suggested by the final two BFOs, and also it is behavior not modeled by the Boeing simulations. It is unfortunate that a glitch occurred which leveled the wings, because I think that this simulation would have clearly shown that the crash occurred soon after and not far from the position of the final log-on acknowledge.

  382. airlandseaman says:

    Victor:

    The engine relight attempts are designed in to the 777 computer algorithms, thus normal. Success depends on accessible residual fuel. There were other engine relight attempts in the simulations that did not cause a glitch. I can’t recall now if any of the others were momentarily successful, as in the case of the one associated with the glitch.

    Anyway, in the simulator, successful relight attempts were far less common than momentarily successful attempts. After all, they are only intended to help out when fuel is available, and some other problem caused the flame outs (like heavy rain, hail, etc.). In the case of MH370, it is more likely than not that there was not a successful attempt. Either way, the result would still be impact near the 7th arc.

  383. Don Thompson says:

    @David

    The requirements of airline training simulators are defined by the Code of Federal Regulations, Title 14, Part 60 (and equivalent regulations set out by national regulators). While we typically use the term ‘level D’ to qualify the fidelity of a Full Flight Simulator (FFS), the core requirements for a Level D FFS enhance visual and motion cues for the synthetic flight deck environment. The fidelity of a Part 60 FFS is demonstrated by replicating a real aicraft flight test program: provide the inputs, demonstrate the outputs conform. This involves systems, engines, flight dynamics, and so on.

    The primary purpose of the FFS is to provide a testing environment in which all the training for published tasks and procedures can be tested in a realistic, yet synthetic, environment. A situation where the aircraft has no fuel would be a consequence of ignoring so many non-normal events, and reactive procedures, that the nuance of a few litres of fuel in a line is quite irrelevant.

    An ‘iron bird’ engineering simulator is a different machine: importantly, it will use much original aircraft avionics, whereas a training device will use simulated panels, displays, emulate avionics in software, etc. The engineering rig may iterate the simulation at a higher rate than a FFS.

    I think it very unlikely that an airline training simulator will deal with the situation where fuel dregs are scavenged from the left main tank and APU fuel line. Unlikely, also, that the engineering simulator would have demonstrated that nuance prior to March 2014. However, the engineering simulator should provide the capability to readily add conditions and parameters, given appropriate conditions and a timeline, so as to demonstrate what could be expected to occur in this example where an APU start is attempted when the L Main Tank is drained. But it will not do that on the basis of some real-time, first principles, computation of the volume of the main tank, the aircraft attitude, the suction of the APU inlet pump, etc, it will involve a model defined by some logic conditions and a timeline deduced by the relevant engineering team.

    A Part 60 FFS must simulate a specific aircraft, the ATD docs will reference a real tail number (or numbers, where the FFS can support different configurations such as engine fit, e.g. RR, GE, PW). The UAL FFS experienced by ALSM was, I believe, a PW engined -200 series a/c. Any simulation tasks undertaken by the investigating team and their technical advisors would only be valid with R-R Trent engines. I have a recollection that the Trent 892 and PW4000 engines have differing auto-relight requirements. The Boeing engineering simulator would permit the appropriate customer configuration to be modelled. I have not had sight of specific detail for any of those tasks.

    Further, to the remarks concerning electrical loading. The maximum avionics load is approximately 22kVA. While an IDG can deliver up to 120kVA. To reach anything approaching a 120kVA load one must assume that the galleys are operating, the IFE system, business class seats are moving, the demand driven electric hydraulic pumps, fuel pumps, and so on, are all operating.

  384. Don Thompson says:

    Concerning the APU start.

    The APU start is commanded as a consequence of both Transfer Buses losing power. While the engines may windmill, turn the accessory gearbox and the generator shafts, a generator control unit will de-energise its generator when the engines are ‘unlit’.

    Just to dig into that…

  385. Victor Iannello says:

    @Don Thompson: I had forgotten that the airline simulations were for P&W engines.

    Here’s what Boeing says about it’s simulations:

    The simulations were completed in the manufacturer’s engineering simulator. The engineering simulator uses the same aerodynamic model as a Level D simulator used by the airlines. The simulator is not a full motion simulator but instead is used when a high level of system fidelity is required. The appropriate firmware and software applicable to the accident aircraft can be loaded.

    The simulated scenarios do not represent all possible scenarios, nor do they represent the exact response of the accident aircraft. Rather, they provide an indication as to what response the accident aircraft may have exhibited in a particular scenario. As such, the results are treated with caution, and necessary error margins (or safety factors) should be added to the results.

    It was not possible to simulate all likely scenario conditions due to the limitations of the simulator. Specifically, flight simulators are unable to accurately model the dynamics of the aircraft’s fuel tanks. In the simulator, when the fuel tank is empty, zero fuel is available to all systems fed from the tank. However, in a real aircraft, various aircraft attitudes may result in unusable fuel (usually below engine/APU inlets) becoming available to the fuel inlets for the APU/engines. If this resulted in APU start-up, it would re-energise the AC buses and some hydraulic systems. This could affect the trajectory of the aircraft. Similarly, the left and right engines may also briefly restart, affecting the trajectory.

  386. TBill says:

    @Nederland
    I am thinking the ASTB comment that IFE reboot could be shut from cockpit switch is point later questioned here by the experts. The end flight sequence is very complex however. If active pilot option, and he saw the SATCOM came back on unexpectedly he could have fiddled with L BUS settings to turn it off again of course.

  387. airlandseaman says:

    Regarding the simulator…

    Manufacturer: Thomson
    Model: 700-200
    Year of manufacture: June 1997
    Level: D

    Configuration

    AutoThrottle: Under control of FMS in ECON Mode.
    CI=50
    ZFW (Zero Fuel Weight): 395,000 lbs
    Fuel at 1707 (ACARS reported): 96,500 lbs
    Fuel at end of flight: 200-800 lbs per tank, depending on scenario
    CG: 28%
    Altitude: 35,000 (for most of the simulations)
    TAS @ IGARI: 471 kts
    Wind Velocity at 38S: 57 kts
    Met Direction at 38S: 240 degrees (direction from which the wind is blowing)
    Heading at 38S: 186 degrees
    Track at 38S: 179 degrees

    I cannot find a record of which engines were selected for the sim’s we conducted. UAL has GE and P&W engines in their fleet. (Two General Electric GE90 or two Pratt & Whitney PW4070/4090 turbofan engines). But the simulator is used to train pilots from dozens of airlines all around the world, so it is configurable for RR engines as well.

  388. Nederland says:

    @Brian

    I’m not arguing that in principle the lower end of the trailing edge could have come off because of flutter. But it sounds relatively implausible to me that the lower end broke up first, and then the whole thing, in both cases because of flutter. More importantly, I think the DGA can give a reasonable estimate on the damage from inspecting the flaperon, even without receiving full information on the material from Boeing. To my mind, this should not be easily dismissed because the final conclusion/hypothesis that the flaperon was deflected now seems wrong, with new evidence turning up later.

    Also, am I correct to say that in case of #22 of the Debris Examination Report (vertical stabiliser), the outer skin looks relatively undamaged on the right hand side, but that the outer skin came off on the left hand side? (Or did they dismantle the whole thing, without saying so?) At least, the inner skin is severely fractured on the left hand side. To me, this indicates that the stabiliser fell flat on the water with its left hand side first. This would in turn mean that at least the tail section of the aircraft could have rotated to the left on impact, and did not disintegrate suddenly at very high speed.

    Victor wrote:
    It is possible that the pilot executed a dive-glide-dive maneuver, which would satisfy the BFO evidence and the debris evidence, although that would be a strange sequence.

    I am of course not saying that the aim of a possible glide was to survive. The overall aim of the descent was probably the opposite. But one of the aims surely was to make sure that the aircraft is difficult to find. An unexpected APU start and the system coming online again could possibly be seen as detrimental to this latter aim.

    @TBill

    Another possible explanation for the lack of an IFE logon could be that the plane turned level again soon after the final transmission. That would rebalance the level of fuel left in the pipe, too. The APU could therefore have flamed out before the next handshake was to be expected. But again, I still haven’t seen evidence (or can not recall any) that overrides the clear statement in the ATSB report that you can switch off the IFE in the cockpit and this also means no IFE logon.

  389. Don Thompson says:

    @TBill

    IFE switch: with sight of the work instructions for the AD that implemented the P5 switch for IFE control we confirmed that the switch did not remove power from the MEC located components of the IFE system. It’s these components in the MEC that provide the comms functions. The AD was concerned with removing power from cabin located devices for safety reasons (IFE seat position equipment, bulkhead displays, powered seats, and so on).

    Prior to implementation of that AD, the cabin crew/purser had an ability remove power from the seat position equipment using a switch at the purser station. Implementation of the AD gave control of power to ALL cabin located equipment from the flight compartment.

  390. Andrew says:

    @ALSM

    RE: ”I cannot find a record of which engines were selected for the sim’s we conducted.”

    The EICAS display in the simulator videos shows that it was a PW-powered aircraft. The upper EICAS has EPR indications, which are only found on PW and RR powered aircraft. The lower EICAS has N2 indications, but no N3. That narrows it down further to a PW-powered aircraft.

  391. airlandseaman says:

    Andrew: Thanks for those observations. I’ll add that to the configuration notes.

  392. Nederland says:

    @Don

    Do you also have clear confirmation that the IFE would still log on if it was switched off from the cockpit (against the ATSB statement)?

  393. Richard says:

    @Don Thompson

    Are you saying that the IFE connection transmission expected at 00:21:06 UTC, should have occurred irrespective of the IFE switch position on P5?

    In other words, the “P5 switch for IFE control did not remove power from the MEC located components of the IFE system. It’s these components in the MEC that provide the comms functions.”

    This implies that MH370 crashed before 00:21:06 UTC.

    This implies in turn, a steep descent after 00:19:37 UTC and not a long glide.

  394. Victor Iannello says:

    @Richard said: This implies that MH370 crashed before 00:21:06 UTC.

    Or, the APU was out of fuel.

  395. Richard says:

    @Victor

    Are you saying that MH370 recovered from the steep descent at 00:19:37 UTC and there was a long glide after the APU was out of fuel, using only the RAT?

  396. Don Thompson says:

    @Nederland,

    Is it not apparent that the debris fragment catalogued as Item 22 in the “Summary of Possible MH370 Debris Recovered” is a small piece of a starboard side panel of the vertical stabiliser, the fragment is the central part of the panel, the top and bottom having been torn/sheared/broken away? The damage could be due to bursting, tearing, torsion, but I can be very confident that the least likely cause is falling “flat on the water”.

  397. TBill says:

    @Don Thompson
    “If both Transfer Buses are isolated from supply, as you propose, there is no discretion for deployment of the RAT: it will happen.”

    You got me back in the cockpit.
    No, I am not proposing what you say above. I am proposing at IGARI:

    L GEN OFF
    L TIE ISLN…cuts SATCOM
    R GEN OFF
    R XFER OFF
    R TIE OFF…cuts DFDR
    but….L XFER ON…prevents RAT deployment
    APU ON or OFF (optional)

    I recognize PMDG is not guaranteed accurate, but this is my proposal.
    CVR could be handled later…I assume no absolute need to cut CVR on L XFER at IGARI, just DFDR for the next phase (intentional depressurization).

  398. Nederland says:

    @Don

    Sure, it is obvious that this is only part of the vertical stabiliser and came off from the rest. The observation was that it is badly damaged on the left side (the one you see on p. 53 of the pdf), and not very damaged on the right side (the one you see on p. 57).

  399. Nederland says:

    Edit: the other way round of course, p. 53 and 55 show the right side, which looks not so bad, and p. 57 the left side, which looks bad.

  400. Victor Iannello says:

    @Richard: I’m not a fan of that scenario, because it suggests a dive-glide-dive, which seems pointless. I’m just saying if the APU ran out of fuel, it could explain the missing IFE log-on.

  401. Don Thompson says:

    @Richard, and others.

    I envisage three possibilities why the IFE connection was not initiated (it’s not a log on)

    1) the APU’s fuel had expired before the interval to the IFE connex had passed;
    2) had the APU continued to generate power, an extreme aircraft attitude caused loss of LOS to the satellite. Loss of link (a condition immediately unknown to the transmitter) implies a re-transmit or attempt to re-establish the datalink after a defined period;
    3) the aircraft impacted the ocean before the interval to the IFE connex event had passed.

    Two of three options suggest impact occurred within a very few minutes after 00:19.

    IFE and power: the intent for removing IFE power is to “remove power
    from the cabin systems if smoke or flames occur”. The intent is not to break the integrations between IFE and other cabin/passenger service systems, the integrations provided by the units located in the MEC.

  402. Don Thompson says:

    @Nederland,

    The vertical stab piece is a panel, it has an inner surface and an outer surface. The panel is not a consistent thickness across its span, that is not immediately evident from the images. The profile is flat on the outer face whereas contoured on the inner face. Review the description of items 9 and 15, the profile of the inner face of these (similar) panels is much clearer. It’s likely that the destructive force flexed the panel, stretching the inner face to fracture.

  403. airlandseaman says:

    And to remind everyone, ATSB/Boeing reported virtually the same 00:21 IFE anomaly reasons Dec 3, 2015, to wit:

    “The fact that the expected IFE system transmission was not received could be due to:

    1. the IFE system being selected off from the cockpit overhead panel at some point after the 18:25 logon [now dismissed; cockpit switch does not stop the IFE logon, only power to seats]

    2. the IFE and/or SDU unit losing power (APU flame-out)

    3. the IFE and/or SDU becoming inoperative (due to impact with the water) before the connections could be set up

    4. an unusual aircraft attitude breaking the line-of-sight to the satellite (aircraft transmission not received by satellite).”

    So, #1 has been dismissed as incorrect, #2 is unlikely given 13 minutes of fuel in the APU line, #3 is the most likely, and #4 has the same implications as #2 as far as where to search.

    NOTE: We also know the San Juan was missed the first pass.

  404. airlandseaman says:

    Correction: “… #4 has the same implications as #2 as far as where to search….” should read “… #4 has the same implications as #3 as far as where to search….”.

  405. Victor Iannello says:

    @Andrew: You wrote a paper entitled “MH370 – The Left Engine Restart/APU Auto-start Problem”. In light of the current discussion, I wanted to re-read it, but the link no longer works. Can you please again make it available?

  406. Nederland says:

    I appreciate there is a certain reason why a pilot can turn off the IFE from the cockpit, but I still don’t get why the IFE should necessarily log on, although turned off, when the SATCOM link alone works for data/messages, voice from the cockpit, and ACARS, and the in-flight entertainment system is of use just for the cabin/PAX.

  407. Andrew says:

    @Victor

    Try: MH370 – The Left Engine Restart/APU Auto-start Problem

    Note the caveats in Section 6 regarding Mike’s simulator trials:

    •The simulator modelled a B777-200ER with two-spool PW4090 engines. The PW engines are likely to have different relight and acceleration characteristics to the three-spool RR Trent engines on 9M-MRO.

    • The RR Trent engine’s auto-relight function is disabled if the engine does not restart before the N3 runs down below 35%. However, the PW engine’s auto-relight function is not disabled unless the engine is shut down, which allows a much longer start time.

    • The accuracy of the simulator systems modelling is unknown, especially in relation to the availability of residual fuel after the initial flameout.

    That last caveat ties in with Don’s earlier comment:

    “The primary purpose of the FFS is to provide a testing environment in which all the training for published tasks and procedures can be tested in a realistic, yet synthetic, environment. A situation where the aircraft has no fuel would be a consequence of ignoring so many non-normal events, and reactive procedures, that the nuance of a few litres of fuel in a line is quite irrelevant.”.

    Simulators are training devices that are flight tested and certified to replicate the aircraft’s behaviour in certain scenarios required by the airlines and their regulators for pilot training and assessment purposes. The simulator standards required by the FAA’s AC 120-4B (Airplane Simulator Qualification) stipulate that (my emphasis):

    “Simulator systems must simulate the applicable airplane system operation, both on the ground and in flight. Systems must be operative to the extent that normal, abnormal, and emergency operating procedures appropriate to the simulator application can be accomplished.”

    Given the system design, we know that an APU autostart and/or engine relight might occur in the event of a flameout caused by fuel exhaustion. That scenario was indeed modelled by the simulator used in Mike’s trials, but it is not something that is typically required for pilot training or assessment. Consequently, there is no guarantee the behaviour was modelled with a high degree of fidelity.

  408. Nederland says:

    @Don

    #9 and 15 have in common with the other debris parts that their core is intact.

    In case of #22 the outer skin (on the right) is basically intact, but the inner skin (towards the left) is badly fractured, which to me indicates that the force of impact came from left to right, pushing the inner towards the outer skin, and “squashing” the inner skin.

  409. Richard says:

    @airlandseaman

    Many thanks for the reminder of the possible reasons that the IFE system transmission was not received.

    The ATSB reports that the APU uses 2 lbs of fuel every 55 seconds under load. As you say, with 30 lbs of fuel remaining for the APU from the left tank, when the left engine flames out at around 00:17:30 UTC, this gives a nominal amount of fuel to operate the APU until 00:31:15 UTC. This gives a margin of over 10 minutes of APU operation beyond the time of the expected IFE system transmission.

    The reason the IFE signal was not received when expected at 00:21:06 UTC can only be that MH370 crashed or the aircraft attitude was so unusual as to break the line of sight from the aircraft’s satellite antenna to the satellite.

    I agree with your conclusion that either way, MH370 will be found relatively close to the 7th Arc at around 34.4°S.

  410. George G says:

    If, during high speed descent, the wing broke off outboard of the engine thus releasing the flaperon already damaged by flutter and causing break-away of part of the flap also already damaged by flutter, would the subsequent descent profile of the main part of the aircraft then change from the descent prior to break-up ?
    If so. would it be possible that the change in descent profile might lead to the impact being further away from the Arc than the nominal 25 NM ?

  411. David says:

    @Victor. You wrote, “I remain suspicious of the fidelity of ALL the Level D simulators in their ability to accurately model behaviors after fuel exhaustion with no pilot in the loop.” I join you in that.

    As to the Boeing simulations these were blessed by the ATSB, SSWG, even with their limitations apparent. They must have had reason to do so if more suitable simulators had been available, so it seems likely that there weren’t.

    In the First Principles Review there is the statement, “Results from recent simulations showed high rates of descent broadly consistent with the BFO analysis.” There is nowhere an explanation as to why the simulations conducted should be accepted, given their limitations, or why there was apparently no effort to overcome these.

    The word “broadly” suggests subjectivity, bearing in mind that there are 3 requirements to meet the BFOs; speeds of descent, the average descent acceleration between them and timing. Though some descents match the BFO rates and acceleration, on the timing, none is consistent so far as I know. In which case ‘broadly consistent’ is misleading.

    The most recent bunch of 10 simulations was conducted presumably because those earlier did not support the BFO-related descent. So other circumstances were simulated, though apparently still missing the jackpot.

    This is not to say that with a larger sample including more of abnormal configuration and, as you say, with APU start and relights included the criteria could not be met broadly but no argument along those lines, complementary to the simulation results, was advanced by the ATSB.

    There is the possibility that APU starts and relights might result in crashes further from the 7th arc than was evident in the simulations. As I understand it your judgement is that that is most unlikely. For my part I would note that even so, the number of simulations conducted is small and the outcome of a large sample, with APU start and relight, might see a few go far.

    If so, the aircraft might have crashed beyond the extreme allowed for, based on simulations.

    Potentially this raises the likelihood of the wreckage having been missed, even assuming there was no manned glide.

    But then you go on to say that, “..if the aircraft reaches an increasingly steep descent (downward acceleration of 0.67g) as suggested by the final BFOs, and there are no pilot inputs, then the aircraft is banked and will impact the sea not far (<10 NM) from that point.”

    In effect you are saying that if the BFO interpretations are sound, there need be no notice taken of the simulations. The trouble is that the simulations are there, extant, and as they stand, contrary to the BFO interpretation IMO.

    Further, again on your observation that if APU start and relights were included that could well bring the simulations into line I think that is based on the United Airline outcomes reported by ALSM. We then return to the start point as to how representative they are of what the Boeing simulator would have disclosed had it been programmed satisfactorily.

    I think it might be prudent, since the wreckage has not been found, to allow the possibility that search width should be wider than has been estimated.

  412. David says:

    @ALSM. About why the Boeing engineering simulator was used despite its shortcomings you think that was, “Probably because of the large number of sims they conducted, and the convenience of the engineering simulator to their work.” That would be a Boeing-centric approach though surely if there were better alternatives the ATSB or SSWG would have intervened?

    I would indeed approach Boeing about it except that they did not respond to an earlier request of mine for information and anyway, as Honeywell explained to me, there are inhibitions to releasing information to foreigners.

    You might have more success.

    As to why Boeing used their engineering simulator you write that, “Your pure speculation that they did it because the engineering simulator was superior is baseless nonsense.” I would withdraw such a comment to that effect had I made it, though I didn’t.

  413. Richard says:

    @Andrew

    Many thanks for your excellent paper entitled “MH370 – The Left Engine Restart / APU Auto-Start Problem”.

    You describe 2 cases where a Left Engine Restart may have occurred using the remaining 30 lbs or 13.6 kg of fuel, (a) a successful restart to CLB thrust would require 14.7 kg of fuel, which is more than the available fuel and SDU would not reboot (b) a successful restart to an intermediate thrust would require 12.0 kg of fuel but the engine would deplete the remaining fuel before the SDU could complete its reboot sequence, which takes around 60 seconds.

    In the MH370 flight path scenario described in the article above, the aircraft is assumed to be at FL390, which is way above the altitude supported by one engine operation, after the right engine flames out. The left engine will be increased to CLB thrust following the right engine flameout to try to maintain speed and altitude. If a left engine restart would have been successful, which is in line with your first case (a) above, the remaining fuel would have been depleted before the SDU reboot sequence could have started. Clearly, the left engine did not relight (as you mention, this is not surprising given the aircraft’s assumed altitude is well outside the demonstrated in-flight start envelope), the APU did come online and the SDU did reboot to provide the satellite logon request at 00:19:29UTC and logon acknowledge at 00:19:37 UTC.

  414. David says:

    @Don Thompson. Thank you for that informed response.
    You say, “The fidelity of a Part 60 FFS is demonstrated by replicating a real aircraft flight test program……” My understanding then is that unless the inputs are taken from flight testing there will be no replication in a Level D FFS. I doubt that flight much beyond flight boundaries would be done other than by analysis (the 737 at high Mach required this and flutter calcs for the Silk Air investigation).

    Flight testing of left fuel tank fuel exhaustion, right engine stopped might be too expensive and dangerous for the marginal gain foreseen and likewise spiralling high speed descents, high bank etc also I note your, “A situation where the aircraft has no fuel would be a consequence of ignoring so many non-normal events, and reactive procedures, that the nuance of a few litres of fuel in a line is quite irrelevant.”

    So in the absence of such data, flight and conditions would not be simulated consistent with Part 60 requirements. Specifically you say, “I think it very unlikely that an airline training simulator will deal with the situation where fuel dregs are scavenged from the left main tank and APU fuel line.”

    So what would account for ALSM’s experiences if inconsistent with Part 60?

    You go on to imply that Boeing could have adapted the engineering simulator to suit had it wished or that had been requested and I assume that this would be outside Part 60 also; but at least the source data would be known, authoritative and the output qualified accordingly.

    You say, “I have a recollection that the Trent 892 and PW4000 engines have differing auto-relight requirements.” That is mine too, the RR igniters firing when the engine speed drops below flight idle, the PW on flame out. Also the 3 spool response might be different to the 2.

  415. DennisW says:

    @Richard

    I agree with your conclusion that either way, MH370 will be found relatively close to the 7th Arc at around 34.4°S.

    I don’t agree with your conclusion above for reasons I have already made clear. The 34.4S latitude is far from compelling, and the area close to the 7th arc has already been searched.

  416. Richard says:

    @DennisW

    1. The MH370 end point at around 34.4S is valid even if, like you, one decides to ignore the BFO.

    2. In light of the San Juan experience, you must admit it is possible to miss debris, in an underwater search.

  417. dennisW says:

    @Richard

    Please provide a link to the The MH370 end point at around 34.4S is valid even if, like you, one decides to ignore the BFO..

    I must have missed it. I do agree that it is valid, but the possibility that the debris field was not found is low (~10%). Our disconnect has nothing to do with validity. It has everything to do with probability.

  418. DennisW says:

    @Richard

    BTW, I would not characterize my position as ignoring the BFO. It (BFO) is useful for knowing the plane went South and came down rapidly at fuel exhaustion.

    The interpetation of the BFO data having a stationary variance is just plain wrong. You know, or should know, that the variance of a random walk increases as the square root of time (a proxy for the number of samples). Weighting the samples early in the flight the same as samples later in the flight under-estimates the variance later in the flight. One could (arm waving here) call your BFO variance an “average” variance. The variance at the end of the flight is at least 1.414 times your “average”.

    In my view, the search should be continued North from 25S, to maximize the probability of finding the aircraft.

  419. Victor Iannello says:

    @Andrew: Thank you for access again to that excellent paper. There does not seem to be enough residual fuel for both an engine restart (producing thrust asymmetry) AND running the APU.

    In addition to not knowing the position of the switches controlling the IDGs, BUGs, and bus ties, I also raise the following points:

    1) If the fuel crossfeed valves were opened, it would shorten the time between the flameout of the right and left engines, and also make it difficult to determine which engine flamed out first.

    2) If the left engine fails first and the left bus is automatically tied to the right, can we be certain that the break time does not cause a reboot of the SDU? The SATCOM is designed with enough internal bus capacitance to ride through a power interruption lasting 200 ms, but do we know what the actual break time would be for an in-flight transfer?

  420. TBill says:

    @all
    Re: San Juan analogy-
    MH370 was not like a submarine crawling along the ocean bottom getting stuck in a low spot. I have always hoped, that if we at least got in the ballpark of the MH370 debris field, we mnight find a piece or two that had some buoyancy and drifted a little. The 34.5 South area has been searched so many times, I already said Victor had 3 chances to be correct on that end point, and I think there might actually have been 4 times when OI went back to recover the lost AUV and rescanned again some areas. I finally got tired of telling Victor how many shots ghe had at it…of course at that time I was also very interested in 180 South, so I was not opposed to it.

  421. Victor Iannello says:

    @All: Don Thompson alerted me that the Ocean Infinity has found the French submarine Minerve, missing since January 1968. It was found in the Mediterranean Sea, about 30 miles from Toulon, at a dept of about 7,800 ft. That’s another feather in Ocean Infinity’s cap.

    https://www.bbc.com/news/world-europe-49068823

  422. Don Thompson says:

    @David,

    You asked, “what would account for ALSM’s experiences if inconsistent with Part 60?

    As I wrote, the performance of an airline FFS must be demonstrated to comply with a baseline derived from a flight test program.

    However, it’s typical that the simulator manufacturer must gain customer acceptance of the new FFS prior to regulator approval. During this phase the customer’s senior pilots will assess the simulator to ensure they are satisfied with its performance, customer requirements may exceed Part 60. Often, ‘snags’ outside the requirements of Part 60 will be highlighted and must be resolved to the satisfaction of the customer.

    A Level D FFS is not the ideal tool for testing an MH370 end-of-flight scenario unless the underlying ‘model’ and implementation is fully understood. The manufacturer’s Engineering Simulator should provide a closer to ideal tool. From our vantage point, many unknowns stand for the detail of the evaluations/tests performed.

    Now… I have some further thoughts for an uncontrolled end of flight scenario. The timings that have been set out for APU start, AES reboot from power interruption, are all ‘best case’. However, the AES Log On Request may not have been the first LOR burst transmitted by the AES, the GES can only record what it successfully receives and decodes. Had the AES transmitted a LOR prior 00:19:29, a finite delay is implied before its wait, for the Log On Confirm response, times out and a new LOR is transmitted. I am not suggesting that the 00:19:29 event is in doubt, rather that the interval from prior events may be longer, which potentially shifts the 00:19:29 event further into the sequence of an uncontrolled descent.

  423. Victor Iannello says:

    @TBill: Much of the seabed around the 34.3S arc crossing was scanned once, not three or four times, and some of the seabed that WAS scanned did not produce usable data due to terrain masking and steep slopes. You are much more certain of the quality of the data than I am. Is this based on solid analysis, or another hunch?

    I make my statements about the possibility of having missed the debris field based on examining the bathymetry data and the SAS data from GO Phoenix, as well as conversations I’ve had with those that were responsible for analyzing the data during the search. In addition, Don Thompson is doing a much more detailed, quantitative analysis of all the data, and has shared some preliminary results.

  424. Don Thompson says:

    @TBill wrote “the 34.5 South area has been searched so many times

    Please stick to facts: the area surveyed by the GO Phoenix/Phoenix International/SL Hydrospheric team was covered once with the ProSAS-60 towfish. Surprisingly few gaps/holidays were declared and, as a consequence, the area was given very little re-inspection by Fugro’s “Echo Surveyor” Hugin AUV.

    I say ‘surprisingly few’ because the bathymetry isn’t significantly different to areas where Fugro worked, areas that were “riddled” with gaps/holidays that required re-survey by the “Echo Surveyor”.

    Ocean Infinity’s work overlapped the edges of the ProSAS-60 area and made good on deficiencies in one particular reqion (S34.54, E93.85) but, the ProSAS-60 surveyed area has not been searched more than once.

  425. Don Thompson says:

    Further to Victor’s comment, the Geoscience Australia archive of the Phase 2 search data includes source data as recorded during the seafloor surveys and post-processed data.

    The post-processed data does include representation of the towfish survey lines in GIS compatible shapefiles, unfortunately, the shapefiles don’t include any Z, elevation, data.

    I’m particularly interested in the towfish elevation as it progressed along its survey line. The source data, in a proprietary format, does include elevation: the towfish depth and its altitude off the seafloor. I have deduced the proprietary format and am presently sifting through and “cleansing” the millions of position records.

    Towfish position was recorded at one second intervals during active survey operations throughout the 7 month campaign. My starting point was 95GBytes of source data. At this time, my progress is steady, and positive.

  426. TBill says:

    @Victor
    @Don
    To be clear, I am not suggesting the same area was re-searched 4 times. I am saying at first we had the middle searched on Arc7 by Go Pheonix, then we had OI return later and each outer side searched out further, and if I recall maybe a litle rescan of some missed lanes. If it was a hard crash with many pieces, I would hope we could lucky and find something even if we were not on top of the debris field.

  427. airlandseaman says:

    Victor: Re: “There does not seem to be enough residual fuel for both an engine restart (producing thrust asymmetry) AND running the APU.”

    That assumes the APU required fuel “in the tank”. But that is not the case. There was ample fuel in the rather long fuel line running from the left tank to the rear of the plane to supply the APU for several minutes once the fuel in the tank was exhausted.

  428. Victor Iannello says:

    @airlandseaman: Maybe. Once the suction to the DC pump goes dry, the DC pump can no longer pressurize the APU fuel line. I don’t know how the APU’s boost pump would operate with low pressure in its suction line, and whether the entire residual fuel in the APU fuel line would be available. Perhaps others can comment.

  429. airlandseaman says:

    Victor: I don’t know for sure about the 777, but in all the low wing aircraft I have flown that use wing tanks, the fuel pump is in the engine compartment above the fuel tank.

  430. airlandseaman says:

    …I am only referring to the APU pump locations, not the ME pumps.

  431. Hank says:

    @airlandseaman

    I don’t understand the issue about starting the APU after fuel depletion shuts down the engines? There could be residual fuel that the APU could access that the main pumps cannot. But the RAT automatically deploys when the engines shut down and provides electric and hydraulic power for some of the systems.

    Is there some issue about the RAT operating all the way to impact?

  432. airlandseaman says:

    Hank: The discussion is about the reason why there was no IFE logon at 00:21 as would be expected following the AES logon at 00:19:37. This has nothing to do with the RAT, which is believed to have been deplyed circa 00:18. But the RAT does not power the AES.

  433. Hank says:

    @airlandseaman

    Thanks. Hank

  434. Andrew says:

    @Richard
    @Victor

    Thank you for your kind words. Given the residual fuel quantity at fuel exhaustion, I agree that it’s unlikely that both the engine and APU relit/started. In that case, the engine would have consumed some fuel during the relight attempt, but the bulk would have been available for the APU. Let’s say the engine wound down to 35% N3 in 30 seconds without a relight. At the end of that period, the EEC would have shut off the fuel and ignition and disabled the auto relight function. Using the light-off fuel flow figure from the report, the fuel consumed during the relight attempt would be 140 kg/hr x 30/3600 hr = 1.2 kg. That leaves 12.4 kg of fuel available for the APU auto-start, which provides plenty of running time for the SATCOM transmissions.

    RE @Victor’s comment: “I don’t know how the APU’s boost pump would operate with low pressure in its suction line, and whether the entire residual fuel in the APU fuel line would be available.“

    I doubt the APU would continue to run after the wing tank ran dry. I note ALSM’s comment regarding the placement of the engine pump and wing tanks in the light aircraft he has flown, but in the MH370 case there is a long fuel line and a considerable height difference between the tank and the APU. The auto start is also assumed to have occurred at very high altitude. @David previously suggested that vapour lock would prevent the APU suction feeding under those conditions and I agree. If the main tank boost pumps fail on one side, the associated engine won’t reliably run by suction feed above FL380. I therefore think it’s unlikely the APU would run at high altitude without boost pump pressure, especially given that it is mounted higher than the tank.

    RE @Victor’s comment: ”If the left engine fails first and the left bus is automatically tied to the right, can we be certain that the break time does not cause a reboot of the SDU? The SATCOM is designed with enough internal bus capacitance to ride through a power interruption lasting 200 ms, but do we know what the actual break time would be for an in-flight transfer?”

    I remember it being discussed some time ago and I seem to recall there was an opinion that the power interruption would not be sufficient to cause the SDU to reboot. That said, I don’t know the duration of the break time for the power transfer.

  435. David says:

    @Don Thompson. Thank you for your answer on the suitability of a level D FSS for testing an MH370 end-of-flight scenario.

    On the possibility of a shift of BFO timing further into an uncontrolled descent sequence can you estimate/guess what the time out for a log-on confirm response would be?

    Conceivably there might be more than one LOR that fails, particularly if the link has been broken via aircraft attitude.

  436. TBill says:

    @Andrew
    Regarding the last point, FSX PMDG shows an apparent quick reboot upon left engine fail, which is of course questionable fidelity to real world, but sometimes gets dragged into the Twitter MH370 debates.

  437. Viking says:

    @DennisW

    I agree with your objections concerning recommending an area near 34.4S. In particular your point concerning the performance of the BFO data is important. It is not acceptable to assume Gaussian statistics (even if it is only implicitly assumed) for the error distribution, since after a violent thermal and pressure cycling of the SDU without power on the thermal stabilization, the oscillator is likely to behave erratic for quite some time. Such an erratic behavior produces a tail-heavy data distribution.

    For those unfamiliar with this kind af data distribution it means that ‘outliers’ far beyond the normal standard deviation are orders of magnitude more likely. If one tries to squeeze this distribution into a normal distribution with a 4Hz standard deviation it leads to an artificial shift towards the optimum for the BFO data (around 27S) compared to the Ashton paper result, and in addition other possible solutions may be overlooked.

    I have just returned from vacation and quickly read through the last couple of hundred contributions, so further comments will come gradually over the coming days.

  438. mash says:

    re: IFE (log) missing [& (re-)search priority, if any?!]

    From a (general) system point of view, IFE is a subsystem, and an add-on subsystem too [actually don’t know how it works at all]. But from this point of view, is it really impossible to [‘physically’] disconnect IFE [on the fly] from the (main) system? Of course, this would only likely happen in a completely different scenario …

    By the way, does it make more sense (cents) to search the wider part first [if both parts (34.x) selected]?

    More remotely, wonder whether the dive-glide-dive (if happened) could also be due to ‘system constraints’ too (i.e. a different kind of “struggle”)?

  439. DennisW says:

    @Viking

    Even without any unusual perturbations, the math associated with using Gaussian stats on a non-ergodic and non-stationary process is deeply flawed. The advanced degree people in the DSTG were obviously sleeping in stats classes.

    My computer science professor daughter knows how to compute mean and variance, but has no idea if that is meaningful. Don’t ask me how I know. Very dissappointing.

  440. Nederland says:

    In the above I asked if anyone has definitive confirmation that the ATSB statement that you can switch off the IFE and therefore prevent it from logging on is wrong. I take no answer for no.

  441. Niels says:

    @Don Thompson
    I would like to understand the towfish scanning a bit more in detail.
    – Could you give an indication for typical towfish altitude off the seafloor?
    – Would you know how towfish position was determined?

  442. David says:

    @Victor, Andrew. Victor asked, “If the left engine fails first and the left bus is automatically tied to the right, can we be certain that the break time does not cause a reboot of the SDU?………………”

    Def of Underwater Search Areas, p8, 3rd Dec 2015. About bus tie breakers joining left and right AC buses, “This power switching is brief and the SDU was designed to remain operational during such power interruptions. To experience a power interruption long enough to generate a log-on request, a loss of both AC buses, or a disabling of the automatic switching would be required.”

  443. Don Thompson says:

    @Niels,

    The ‘ideal’ altitude for the ProSAS-60 was 150m. In practice, the altitude varied considerably as the crew navigated it over the seafloor features. I’ll pull together some plots.

    I cannot emphatically describe how the towfish position is determined but these inputs are used: USBL techniques with the tow ship (referenced to a DGPS derived position for the ship), a DVL and INS on the towfish, towfish depth from water pressure, and a downward looking sonar on the towfish recording the towfish altitude off the seafloor.

    Position data was recorded, at 1sec intervals, by a software application on the ship: 1 record with towfish position and 1 with towfish attitude.

  444. Richard says:

    The MH370 Flaperon arrived at the Direction Générale de l’Arment Techniques Aéronautiques (DGA TA), a French Ministry of Defence facility for aviation expertise, in Toulouse on 1st August 2015. A tribunal of the French Ministry of Justice requested DGA TA to prepare a report on the Flaperon. This was completed on 26th July 2016 and consisted of 238 pages, 61 photos or figures and 13 appendices. The report contains the correspondence with Boeing via the French police and Boeing’s lawyers in Annexe 12 (the questions asked on 11th February 2016) and Annexe 13 (the responses received on 18th March 2016). Boeing responded to all questions, but refused to reveal the specific makeup of the composite portions of the Flaperon as it was proprietary and commercially confidential information. Boeing gave a general overview of the composite portions of the skins, spars and ribs in the Flaperon and the composition, characteristics and manufacturing method of the standard modulus epoxy impregnated carbon fibre fabric used in the Flaperon.

    The report concludes “In the absence of data from Boeing, and despite the deterioration of some fracture surfaces, a hypothesis was nevertheless formulated: taking into account the results of the examinations, it appears that the flaperon impacted the water while still attached to the aeroplane and that at the time of the impact it was deflected.” and “The little data supplied by Boeing did not enable the examination to be progressed by making calculations that would have made it possible to confirm or reject the proposed hypothesis.”

    The Flaperon was manufactured by Construcciones Aeronauticas S.A. (CASA), an Airbus subsidiary based in Spain at San Pablo, Seville and delivered to Boeing on 18th February 2002. On 24th August 2015, François Grangier, an expert appointed by the French Supreme Court and Pierre-Charles Rolland an Airbus Engineer appointed to assist the court, contacted Airbus in Seville to confirm the part numbers and serial numbers of the Flaperon components. Airbus Seville responded in full on 2nd September 2015, confirming the Flaperon components were from the assembly of aircraft unit number 404 for the delivery to Malaysia Airlines as part of the WB175 build, which was subsequently registered as 9M-MRO.

    If the various French experts, appointed by the court, were in direct contact with Airbus, the manufacturer of the Flaperon, why did they want more information from Boeing? What would Boeing know about the Flaperon, that the manufacturer Airbus did not know? Why did DGA TA complain that Boeing supplied little data, when they were in direct contact with the Flaperon manufacturer? Why did DGA TA not use their direct contacts with Airbus to get the data they needed to complete their calculations and arrive at a firm conclusion?


  445. Niels says:

    @Don Thompson
    Thank you, Don. I think it is important to see the (aimed) 150m in the light of typical swath width (“opening angle”) and then imagine a non-flat seafloor.
    It would be good to have an estimate on accuracies regarding the position determination.

  446. Richard says:

    @Nederland

    You stated “In the above I asked if anyone has definitive confirmation that the ATSB statement that you can switch off the IFE and therefore prevent it from logging on is wrong. I take no answer for no.”

    Why do you believe that an Airworthiness Directive issued by Boeing is not a definitive confirmation that the ATSB statement was incorrect?

    Please see the comment from Don Thompson on 21st July 2019:

    https://mh370.radiantphysics.com/2019/06/30/a-comprehensive-survey-of-possible-mh370-paths/#comment-24265

  447. Richard says:

    @Nederland

    Here is the FAA notice describing the reasons for Boeing issuing the Airworthiness Directive to include a cockpit switch on the overhead panel 5 in order to be able to cut the power to cabin components of the IFE, cabin lighting and passenger seats in case of fire or smoke, as explained by Don Thompson.

    https://www.dropbox.com/s/ewzb5tb32iz2aas/EASA_AD_US-2014-16-11_1.pdf?dl=0

  448. TBill says:

    @Nederland
    Unfortunately re: confirmation, as you know there are many closed lips re: MH370 (Boeing, Malayasia, etc etc) thus as the public we do not have the advantage of confirmation of many facts. DrB has some fuel consumption info on secrecy agreement, so that is the nature of the problem.

    The implication is however any future search effort by OI or whoever should try to get some secrecy agreements so they can rank search area proposals/assumptions.

  449. Victor Iannello says:

    @David: Thank you for the reference. Considering the SATCOM is designed to ride through 200 ms of power interruption, and the contactors make and break in about 20 ms, it’s not a surprising result.

  450. DennisW says:

    @Richard

    I give up.

    The most amusing of your links is the “unexplained” 20Hz BFO error during a routine flight chronicled in the DSTG book. Apparently you and DrB are comfortable ignoring it.

  451. Richard says:

    @TBill

    I agree we are not party to the full truth, but at least we can accept the truth, where we have convincing evidence.

    We have been very fortunate to have received data from a number of whistle blowers in Malaysia.

    @DrB has also been very fortunate to have received data under a confidentiality agreement.

    Please do not try to knock it as “unfortunate”.

    Please do not try to imply any future search is only possible, when dependent on confidentiality agreements.

    Such an implication, is false.

  452. Richard says:

    @DennisW

    You regurgitate your same nonsense.

    I give up too!

    Do not be surprised, if I regurgitate, what you consider, “our same nonsense” back to you.

  453. Shadynuk says:

    @Richard Concerning the questions you pose at the end of your comment above (#24322), don’t you think the French investigators are simply establishing a ‘caveat’ in case their ‘conclusions’ are shown to be inaccurate?

  454. Richard says:

    @shadynuk

    I had the same thought, but then I thought they are professional scientists and engineers.

    I then thought their paymasters are the French Goverment and sadly concluded, their statements were politically motivated more by the enduring battle between Boeing and Airbus.

  455. DennisW says:

    @Richard

    What I would prefer is a link to the analytics substantiating your claim of an optimal route based on BTO alone.

  456. Andrew says:

    @Richard

    RE: ”If the various French experts, appointed by the court, were in direct contact with Airbus, the manufacturer of the Flaperon, why did they want more information from Boeing? What would Boeing know about the Flaperon, that the manufacturer Airbus did not know? Why did DGA TA complain that Boeing supplied little data, when they were in direct contact with the Flaperon manufacturer? Why did DGA TA not use their direct contacts with Airbus to get the data they needed to complete their calculations and arrive at a firm conclusion?
“

    If the components were manufactured by Airbus under licence to Boeing, I expect there would be agreements in place to protect Boeing’s intellectual property. In that case, Airbus couldn’t legally release the information without Boeing’s permission.

  457. Victor Iannello says:

    @Shadynuk: It comes down to which is more credible: the hypothesis of the French, or the later failure analysis of the flap by the ATSB. Ideally, the French would incorporate the observations and conclusions from the later investigation of the flap into their results and determine if the witness marks on the flap match their hypothesis. (The ATSB believes they don’t.)

  458. Sid Bennett says:

    re:Allan variance

    In the present context the Allan variance seems to be applied to the characteristics of the on-board oscillator, presuming that the oscillator in the satellite is quite stable (not) as is any oscillator at the earth station.

    The Allan variance of a frequency of an oscillator, in the time range less that 100 sec or so is dominated by angle random walk. This would be true for any time-discrete measurement of BFO. When used in inertial navigation systems, the sensor ARW is said to result in a navigation error that increases as sq-rt hr. (That is, starting at a particular location, the location at 1 hr differs from the computed location by a value proportional the Allan variance. It can be anywhere within +/- the value. How this applies to the present case is unclear.)

    Interpreting the Allan variance at long times is only possible semi-quantitatively as the behavior is highly dependent of the specific error mechanisms (such as temperature variations, lack of complete cancellation of bias offset, long term aging etc.)

    It seems to me that the typical behavior of a crystal oscillator, including the effects (at least qualitatively) of turn-on transients as been thoroughly discussed over the past years. That a measurement on an another flight apparently exhibited a large unexplained error (without access to the completed data and algorithm used) should not divert our attention from the data set we have.

    The original Inmarsat paper has some nuggets of information on the BFO errors due to the “eclipse effect” during the period 19-22UT, where the most likely BFO error originated (and can be estimated). I would dearly love to see plots such as Fig. 11 for a period not dominated by the eclipse effect.

  459. Richard says:

    @Andrew

    You stated “If the components were manufactured by Airbus under licence to Boeing, I expect there would be agreements in place to protect Boeing’s intellectual property. In that case, Airbus couldn’t legally release the information without Boeing’s permission.”

    In that case, it begs the question as to why Boeing put their commercial advantage through protecting their proprietary intellectual property, before solving what is arguably the greatest aviation mystery involving a Boeing aircraft.

  460. Richard says:

    @Andrew

    You stated “If the components were manufactured by Airbus under licence to Boeing, I expect there would be agreements in place to protect Boeing’s intellectual property. In that case, Airbus couldn’t legally release the information without Boeing’s permission.”

    If this was the case, then it begs the question as to why Boeing put their commercial advantage through intellectual property before solving what is arguably the greatest aviation mystery involving a Boeing aircraft.

  461. Richard says:

    Oops! The web site swallowed the first comment. NOT!!

  462. Nederland says:

    @Richard

    Thank you for that AD file!

    First, the ATSB statement was, on the contrary, that it is possible to switch off the IFE from the cockpit and in so doing prevent a logon. Without strong evidence to the contrary, I’d say this is correct.

    In the AD, I can’t find confirmation that the IFE will still logon to the SATCOM connection. On the contrary, I read that you can turn off electrical power to the IFE system. In the FI, I read that there is an interface between the IFE and the SDU (p. 49). I would therefore think that the IFE won’t logon if it is no longer powered.

    @TBill

    Of course if there is confidential information to the contrary, that could make a difference, but in this case I haven’t heard of any (not even that such information could exist). And I’d also say this is not some industrial or investigative secret, if it has even been addressed in a published ATSB report.

    Victor wrote: It comes down to which is more credible: the hypothesis of the French, or the later failure analysis of the flap by the ATSB. – There is also the middle way that the flaperon was not deflected but still came off on impact rather than in flight. In fact, there is nothing in any ATSB report suggesting otherwise.

  463. Richard says:

    @DennisW

    You stated “What I would prefer is a link to the analytics substantiating your claim of an optimal route based on BTO alone.”

    Links already supplied, more than once, but you are not listening, so I rest my case.

  464. DennisW says:

    @Richard

    Your name appears 133 times in this section of Victor’s blog. I looked at all of them (again). All I could find was your claim of me not listening and Nederland being obtuse.

  465. Richard says:

    @Nederland

    You stated “I read that you can turn off electrical power to the IFE system.” That is not true. That is not stated in the document that I provided.

    Electrical power to the complete IFE system is not turned off. Only the components in the passenger cabin are turned off, in case of fire or smoke.

    The AD states that the new switch removes power from “the passenger cabin in-flight entertainment (IFE) system, cabin lighting, and passenger seats”.

    I previously pointed out the comment from Don, where he has gone through the work instructions of the AD in detail and concludes that the comms functionality in the MEC is not affected.

    Don further reports that Andrew and he had gone through the work instructions to implement the AD. The PIIC and the EPESC provide integrations between the IFE and PSS, PAS, CSS. There is no evidence based on their analysis that the PIIC and EPESC are shutdown.

    Do you have evidence to the contrary?

  466. Richard says:

    @DennisW

    My apologies! I was referring to the 4 links provided earlier today.

  467. Nederland says:

    @Richard

    There is no evidence based on their analysis that the PIIC and EPESC are shutdown.

    I’m not sure that this answers the question.

    SATCOM, voice, data and so on do come online again. That is true. But how can the IFE log on if it is unpowered? And how can we be sure that this will definitely happen?

  468. Andrew says:

    @Nederland

    I’m away from home at the moment and don’t have access to the documents, but IIRC, the electrical relays associated with the IFE/PASS switches do NOT affect power to the PIIC (passenger inflight information computer) or the EPESC (enhanced passenger entertainment system controller), the two components responsible for the IFE connection.

  469. Nederland says:

    @Andrew

    How about if the passenger inflight information computer was not operational for the flight?

  470. Richard says:

    @Nederland

    I will let @Don Thompson and @Andrew speak for themselves, but in my view they are both competent and knowledgeable individuals with respect to the B777 systems and work instructions for a Boeing 777 AD.

    If they tell me, that they have been through the work instructions of the AD and it only effects the cabin passenger part of the IFE and not the MEC comms IFE functionality, that is good enough for me.

    Of course, if your judgement of their expertise is different, you are entitled to your opinion.

  471. Nederland says:

    Just saying, there were perhaps some issues with the IFE. A software downgrade a few days before the flight, and also the following defect: In-Flight Entertainment (IFE) Airshow does not show arrival time/time
    to destination logged time & problem still persists.
    (FI, p. 28)

    And can we also be sure that the AD was implemented? In other words, can we be 100% sure that nothing has been missed and that the ATSB was definitely wrong?

  472. Don Thompson says:

    @Ned(erland)

    I will repeat: the purpose of the Service Bulletin that describes the implementation of the IFE/PASS switch on panel P5 in the flight compartment is to “provide new switches on the flight deck than can remove power from the cabin systems if smoke or flames occur. Smoke and flames in the passenger cabin can cause injury to the passenger and damage the safety critical cabin equipments.

    The SB makes no mention of the LRUs located in the MEC. The LRUs located in the MEC provide functions that are integrated with other systems: the Cabin Services System including Passenger Address, Passenger Service, and Cabin Interphone functions. The FCOM description for the IFE/PASS switch makes no mention that CSS/PSS/PAS/CIS functions are affected by operation of the switch.

    Further, and more importantly, there is nothing in the conditions observed during the flight that the IFE/PASS switch was toggled to OFF. At 1825UTC, the IFE system opened its connections over the SATCOM datalink. There is nothing to suggest that the IFE/PASS switch state changed after 1825UTC.

    When considering the hypothesis that, after Main Engine Fuel Exhaustion, the APU auto-started and ran for a finite time the expected sequence of events is similar to the 1250UTC SATCOM initiation on the ground at KLIA: the aircraft was powered by the APU, SATCOM boots up and after an interval the IFE initiates its connections over the SATCOM Datalink.

    You wrote, above, “[I can’t find confirmation that] the IFE will still logon to the SATCOM connection.

    logon to the SATCOM connection. is word spaghetti – those words don’t mean anything. If you have no clue about the IFE’s use of the SATCOM datalink, why even enter the discussion? ARINC 628, give us a run through of that?

    I am not finding the engagement constructive.

  473. Richard says:

    @Nederland

    You asked “How about if the passenger inflight information computer was not operational for the flight?”

    I can assure the IFE was operational for the flight.

    I can even tell you what music was playing! Pet Shop Boys!

    https://www.dropbox.com/s/0blbth9nsapmosa/Pet%20Shop%20Boys.png?dl=0

  474. Nederland says:

    @Don

    I’m sorry you don’t find the engagement constructive.

    I still have to say though (Further, and more importantly) that the perp could have forgotten to turn off the IFE after the diversion and before the 18:25 logon. He could have changed that immediately thereafter (probably one of the messages that show up on the screen?). I’d suspect that he wanted to make sure that no message from the cabin gets through.

    The ATSB report also seems to raise that possibility (the IFE system being selected off from the cockpit overhead panel at some point after the 18:25 logon).

  475. Nederland says:

    @Richard

    Yes, the IFE was operational, not sure about inflight information though. Time to destination seems to be an important part of that, and that didn’t work.

  476. Richard says:

    @Nederland

    You asked “And can we also be sure that the AD was implemented?”

    That is irrelevant.

    If the AD was not implemented then there is no switch in the cockpit to remove power from the cabin passenger IFE system.

    Don already stated “Prior to implementation of that AD, the cabin crew/purser had an ability remove power from the seat position equipment using a switch at the purser station. Implementation of the AD gave control of power to ALL cabin located equipment from the flight compartment.”

  477. Richard says:

    @Nederland

    You stated “IFE was operational, not sure about inflight information though. Time to destination seems to be an important part of that, and that didn’t work.”

    The last ACARS messages show the destination as ZBAA.

    There is no ACARS update showing the destination changing to the South Pole.

    How do you know the time to destination didn’t work?

  478. Nederland says:

    @Richard

    Again, you want to look up p. 28 of the FI (p. 47 of the pdf). The heading 1.6.3.7 (“Deferred Defects”)

    In-Flight Entertainment (IFE) Airshow does not show arrival time/time
    to destination logged time & problem still persists.
    dating to the end of 2013.

    ACARS worked correctly yes,that is well known, but it seems the IFE did work for Pet Shop Boys, but not for passenger inflight information.

  479. Nederland says:

    The LRUs located in the MEC provide functions that are integrated with other systems: the Cabin Services System including Passenger Address, Passenger Service, and Cabin Interphone functions.

    In that case, too, the Cabin Interphone was rubbish as it was NOT connected to the ground (no need to logon therefore)

  480. Richard says:

    @Nederland

    You stated “I still have to say though (Further, and more importantly) that the perp could have forgotten to turn off the IFE after the diversion and before the 18:25 logon. He could have changed that immediately thereafter (probably one of the messages that show up on the screen?). I’d suspect that he wanted to make sure that no message from the cabin gets through.”

    I suspect that the real reason was, that the perp did not like Pet Shop Boys.

  481. Nederland says:

    There was still the possibility to send a text message (SMS or email) via IFE from the cabin.

  482. Richard says:

    @Nederland

    You stated “There was still the possibility to send a text message (SMS or email) via IFE from the cabin.”

    In my view, the pilot powered down the SDU to prevent any SMS or e-mail communication between someone in the cabin and the ground. MH370 only had seat to seat calling from each handset and no seat to ground communication via SATCOM. If the aircraft was depressurised at high altitude, well before 18:25 UTC, then the perp would no longer be concerned by possible SMS or e-mail communication between the cabin and the ground. The passengers and cabin crew would all be dead.

  483. DrB says:

    @David,

    You asked: “Should it be that it does fall within tolerances, the crash site might be expressed as a band along the arc rather than a point?”

    A band always been the recommendation, as the path from Arc 6 to Arc 7 is highly uncertain.

    You also asked about “the ‘alternate’ electrical generator loading configuration, whereby the SDU/SATCOM re-boot initiation would occur at right engine failure , , , ,”

    Here are my fuel and endurance model predictions for a LNAV 180 route:

    1. NORMAL Electrical Configuration:

    REFE: 00:10:00, LRC, FL390, FFF 1.07%, M0.8356, 483.5 KTAS,
    260 KCAS, 3414 CUM AIR NM

    LEFE: 00:17:30, INOP, FL 363, FFF 1.07%, M0.665, 385 KTAS,
    208 KCAS, 3467 CUM AIR NM

    2. ALTERNATE Electrical Configuration

    REFE: 00:17:30, M0.830, FL390, FFF 1.07%, 480.5 KTAS,
    258 KCAS, 3439 CUM AIR NM

    LEFE: 00:25:00, INOP, FL 348, FFF 1.07%, M0.623, 362 KTAS, 208 KCAS, 3494 CUM AIR NM

    The alternate electrical configuration can have about 55 NM of powered flight after 00:17:30, with a MEFE at 00:25.

    The main argument I have against this possibility is the lack of IFE communications at 00:21. If the APU started at REFE at 00:17:30, it would continue to provide power until after LEFE at 00:25. It’s not clear to me why or how the SDU or the MEC IFE equipment could lose power between 00:19:37 and 00:21, so that the IFE communications would not occur, even though in this case the aircraft is still powered by the left engine and the APU is still running at 00:21.

  484. Nederland says:

    @Richard

    Two concerns:
    1) If someone attempted to send a message before 18:25 (which is likely) while SATCOM was down, would the system not try to resend that message some time after the IFE came online again? (Would it not therefore make sense to depower the system just in case?)
    2) If there was a decompression (commonly assumed but not proven), it can hardly have lasted for as long as an hour. At least this is what Peter Foley said who rightly pointed to the risk of decompression sickness for any person in the cockpit (for example, Z. was still a smoker, given his age a candidate for COPD). There was probably a chance that someone in the cabin could recover from unconciousness, especially the cabin crew who had used the portable oxygen bottles. (So why risk that?)

    I agree that SATCOM was down still before the diversion. To make sure no messages are sent. If the IFE was depowered also, that would surely have raised early suspicions in the cabin. On the other hand, passenger inflight information was dead, and no one could have noticed any change in the flight plan. The perp could have decided to turn on SATCOM again to check for incoming calls/messages (or for any other reason), but would then have had to depower the IFE if only as a precautionary measure. That would explain why the IFE logged on regularly on ground and at ~18:25 but not during the final logon.

  485. Richard says:

    @Nederland

    I assume that if the cabin was depressurised, that it lasted a maximum of around 30 minutes after diversion.

  486. Nederland says:

    Even at a very high setting (10 l oxygen per minute), the portable oxygen bottles would last longer than 30 mins. It is easy to fall unconscious, but for example passengers of Helios 522 were still alive after ~3 hours of depressurisation.

    Still another possibility is that the perp deliberately left on the IFE until 18:28 to not make it look like deliberate action (depending on the degree of preplanning).

    Would there be no automatic transmission (not necessarily Pet Shop Boys) if the IFE was online for several hours?

  487. Victor Iannello says:

    @Nederland asked: Would there be no automatic transmission (not necessarily Pet Shop Boys) if the IFE was online for several hours?

    That’s a very good question. (I believe the “Pet Shop Boys” was a password for logging on to the IFE server using very rudimentary security.)

    In the case of the Inmarsat network, after inactivity of around an hour, the GES pings the AES to determine if the connection is still alive, and the response from the AES is what produced the BTO and BFO data. However, in the satellite logs, we see no attempt by the IFE server to ping the IFE client using the Inmarsat network. Perhaps after a certain time period, the IFE server simply marks the connection as inactive without verification.

    So, I think the quick answer to your question is there are no automatic transmissions for the IFE connection.

  488. David says:

    @Dr B. Thank you for looking into that.

    In the ALTERNATE configuration, for the same last leg duration but with that same right engine fuel extended by 7½ min to match NORMAL LEFE (and hence BFO) timing, means a lowered the right’s fuel consumption/hr. That requires less speed.

    To cover greater distance (3439 less 3414 = 25 NM), with that fuel means a shade less fuel consumption/mile. So fuel/hr and fuel/mile have both improved by slowing, meaning neither was quite at the most economical?

    Irrespective, whereas my premise points (which must be askew, though where I know not) had the ALTERNATE crash site further north this places it a little further south, the course being say ½˚ west of south. Neither the placement nor course I would assume would affect BTO’s and BFO’s appreciably.

    That is reassuring thanks: the effect of ALTERNATE on the modelling apparently is within bounds (as you thought earlier).

    I missed the mention of the BFO position band along the arc. Is its extent roughly clear as yet?

  489. Tom O'Flaherty says:

    @Nederland

    I assume the perp was not concerned with anyone attempting to send a message as only he would be in a position to do so.

  490. Don Thompson says:

    @Ned

    Earlier you wrote, “the perp could have forgotten to turn off the IFE after the diversion and before the 18:25 logon.

    The simple assumption for the loss of SATCOM service during the hour prior to 1825UTC is that the L AC Bus was isolated, not that a specific circuit breaker located in the MEC was pulled.

    Isolating the L AC Bus removes half of the power for passenger amenities in the cabin, should the R Main Bus also be isolated (and there is no specific evidence for that) then all pax cabin amenity services are unavailable.

    No opportunity to use the seat position IFE to create an message that you consider may have been queued.

    It’s helpful to assess the observations about the SATCOM datalink as secondary to other actions, rather than absolutely central to the observed events. Contriving actions that are ascribed solely to align with SATCOM service is a common mistake.

    Before I go on to address comments following the one I have quoted, please give me some idea of your understanding of ISO-8208 SSN communications so that I have an idea where to begin.

  491. TBill says:

    @Nederland
    If MH370 experienced intentional depressurization as severe as it seems possible (FL380+), the portable O2 bottles and masks are essentially useless.
    The strong vacuum sucks the pressure out of your lungs. If you knew depressurization was coming and were trained scuba-type person (as maybe one of the Russian PAX was) you might be trained how to handle that. But most likely there was chaos in the cabin with the aircraft unexpectedly making extreme turns and climbs.

    Helios had bleed air on and the outflow valves were partially closed, so it only got to about FL250 in the cabin, which means the PAX were alive but unconscious when it crashed. In that case the O2 bottle helped the flight attendant.

    If MH370 wanted to do it quickly, bleed air would be shut off, and outflow valves set to full open. Altitude would be FL350+. I can think of a few more less likely things to further speed up depresurization, such as open cockpit windows when the cabin pressure allows that, which I would not rule out however.

  492. Don Thompson says:

    @David,

    In reply to your question above concerning a retransmission delay in the AES.

    First, I’ll reiterate that the AES can tolerate an interruption in its reception of the P-channel signal from the ground, up to ten seconds.

    I’m suggesting that it’s possible the GES did not successfully receive and decode the first LOR transmission from the AES. Perhaps an R-channel ‘collison’ occurred, or momentary aircraft attitude caused loss of LOS. It’s worth noting that a single burst on a 600bps R-channel has a duration of 960ms.

    The AES waits for a period of between 12 and 19 seconds for a response to its first Log On Request. After that period expires, it retransmits. A variation in the period is determined by a limited range random number (the random element increases as failures repeat).

  493. Nederland says:

    @Don

    Thanks for your reply, and I’m afraid I don’t know much about ISO-8208 SSN communication, but happy to listen.

    I do know, though, about the left AC bus and the different theories proposed in that regard. My own theory still is that shutting down SATCOM was part of the plan.

    Even if the IFE was not operational while the left AC bus was isolated, once power was restored at ~18:25 and the IFE came online again, in theory someone could have been able to send a message, so this could explain why the IFE might have been switched off just after. That would make sense even as a precautionary measure.

    @TBill

    Even if it is possible to depressurise the aircraft very quickly, than the perp in the cockpit would have experienced similar effects (despite the advantages of 100% oxygen and pressure breathing). In the extreme scenario of open cockpit windows, that would presumably kill anyone in the cockpit, too.

    I don’t think anything like this has ever happened and it’s all speculation at the moment.

  494. DennisW says:

    @Nederland

    I’ve never been a supporter of the depressurization scenario for three reasons.

    1> It is difficult to reconcile with a negotiation scenario.

    2> The compatibility of the 18:25 BFO with the likely aircraft heading and speed at that time.

    3> There is no hard evidence to support it. Pure speculation.

  495. DrB says:

    @David,

    The air speed I used for the NORMAL electrical configuration was Long Range Cruise because that is used in the overall most probable route. LRC is actually defined by a 1% fuel mileage degradation from the maximum, which itself defines Maximum Range Cruise. The air speed for the ALTERNATE electrical configuration is less than LRC and closer to MRC, so it’s fuel mileage, endurance, and range are slightly higher. This increased endurance allows REFE at 00:17:30 instead of circa 00:10.

    There are routes which can be fitted at the ALTERNATE air speed, but only at a significantly reduced probability compared to the NORMAL air speed. This is a second strike against the ALTERNATE configuration, in addition to the lack of an explanation for the missing IFE messages at 00:21.

    In general terms, the BTOs are much more discriminating among trial routes than are the BFOs. Our next paper will present some figures with probabilities considering BTOs + BFOs and BTOs only.

  496. TBill says:

    @Nederland
    Pilots have pressurized masks which admittedly not risk free but whole different ballgame re: survival. I am not the least bit certain about open cockpit window scenario but I am thinking that is last resort method for cockpit smoke removal during fire, and thus could be employed to make it look like that action was necessary, or also possibly to eject some evidence. So I am keeping the option open.

    @DennisW
    Yes you have been consistent on that. Appears to me negotiation if any was probably over by IGARI. Over the years I listened to the other experts to see if they agree with you that the BFO behavior proves lack of depressurization, and there is not agreement on that point. Intentional depressurization to me is not speculation, but what a pilot would have needed to do to cut off outside communications and maintain a certain mission plan without interruption. Basically the next logical step after Flight 93.

  497. Greg says:

    @TBill

    Negotiations?
    I missed the moment when this option appeared. Could you please indicate any source where I could learn more about it.

  498. DennisW says:

    @Greg

    Sorry. It is also simply speculation. No evidence exists for a negotiation having taken place.

  499. TBill says:

    @Greg
    Since about 2014 I believe, a participant on the blogs has claimed knowledge that there was a plan to divert MH370. Here is my prior essay on it, in the reference section I document what Victor has summarized about it.

    https://twitter.com/HDTBill/status/1080510470841348096?s=20

  500. Greg says:

    @TBill, @DennisW, thank you for your answers

  501. David says:

    @Dr B. Thanks again and for the LRC economy explanation. I do recall LRC is an optimisation between fuel per mile and trip duration.
    While noting that the 1% MRC range increase would take the NORMAL configuration aircraft 35 miles farther, my understanding is that that would reduce consistency with BTOs/BFOs.

    “There are routes which can be fitted at the ALTERNATE air speed, but only at a significantly reduced probability compared to the NORMAL air speed.” That joins the IFE issue (though see below) in making the NORMAL the more likely from those aspects though NORMAL looks the less probable from a final BFO simulation perspective. Even so simulation weaknesses have been the subject of recent discussion.

    More about that, in the ALTERNATE electrical configuration, APU auto-start (and autopilot disconnect) would have been prompted by depowering of the transfer buses. That also would cut power to all fuel boost pumps. The APU DC fuel pump would have started for the APU, though that supplies just 1430 kg/hr at its normal operating pressure and the left engine drawing part of its consumption through it gravity feed would mean that manifold pressure was very low. If very low, APU DC pump output might increase under the decreased back-pressure but with that the supply to the APU would be prejudiced by vapour lock and the APU could fail. Conceivable the left engine could fail also from lack of supply or, possibly, air embolism from the main tank at altitude that @Andrew mentioned. That illustrates I think how careful and comprehensive simulator modelling must be to be authoritative.

    On the lack of IFE log-on in ALTERNATE, possible explanations are an early crash, achievable if the 2nd BFO’s ROD continued, or the above, or extreme aircraft attitude (spiral or bank).
    On the last, it is unclear to me what the AES and GES reaction would be to the AES not setting up a ground connection for the IFE SMS/email application, the first IFE connection step. It looks like the GES would not interrogate. Presumably the AES would try again after some time-out. If so, the abnormal attitude would have to persist for there to be no more transmissions. So maybe attitude is a doubtful explanation.

  502. David says:

    4th line of the fattest paragraph, ‘..it gravity..’ should read ‘..its gravity..’

  503. Viking says:

    @DennisW

    I agree that the statistics of the oscillator drift is non-Gaussian even without the rough cycling in temperatrue and/or pressure. However, the statistics gets significantly ‘worse’ with cycling.
    In addition, there are other problems with the BFO values. Some of these I described in my paper. This only strengthens your objections.
    However, at the moment I would like to move on to some other problems.

    @All

    The general issue which makes me most uncomfortable with recommendation of a new search near 34.4S is that this solution is based on several implicit assumptions. We have already discussed the BFO issues. This is a conceptually important problem since it is related to basic science and technology. However, there are many other hidden assumptions.

    One of these is that the reason for the diversion must have been suicide. In case of any other underlying reason than suicide, the pilot would never have chosen to aim the airplane for the south pole, since it would leave absolutely no possibility to survive.
    The detailed arguments for this conclusion are longer than a typical Blog contribution, but I think this short version is a good start.

    Jumping to an intermediate conclusion, we do not have solid arguments to exclude other possibilities than suicide. I do agree that the reason must be a deliberate diversion, but we cannot be sure it was done for committing suicide:

    (1) One alternative is the suggestion for negotiations
    (2) Another one is some kind of scam (several possible ones)

    My problem here is the implicit nature of the assumption. A discussion is needed before we can justify such a radical assumption.

  504. David says:

    @Dr B. A correction to my 8:24 PM post please. I see that the isolation valve that rleases APU DC pump fuel delivery to the left engine fuel manifold opens when there is no manifold pressure but also when the engine is not operating, so there will be no supply from it to the left engine despite it having started for the APU.
    In that case the supply to the APU will be unaffected unless the left fails due to vapour lock in its gravity supply or air entrainment.

  505. David says:

    @Don Thompson. Thank you once more. Transmission does require appreciable time and time out is lengthy.
    Iterating, just like IFE non-completion has been attributed to the possibility of aircraft extreme attitude, multiple unsuccessful LOR transmission attempts could delay the final BFO timings appreciably, bringing those closer to simulation timings.
    However given that the LOR’s BFO indicated a descent in the 2900 – 14,800 fpm range (Holland) more likely it would be in the upper part of that range for attitude (bank or a spiral) to be a likely cause of a transmission failure. It is hard to see how such a break in link would last long enough to cover multiple unsuccessful re-transmissions, that is unless aircraft motion caused antenna tracking to go haywire.

  506. DennisW says:

    @Viking

    My preference for a negotiation scenario is based largely on the early response to the diversion by Malaysia. It was handled at high levels in the government, and not by the usual resources who would deal with an emergency situation. It is likely that people controlling the response knew full well that it was not an emergency, and expected the aircraft to land somewhere.

  507. TimR says:

    Viking says:
    @All “… (1) One alternative is the suggestion for negotiations”

    “(Adapted from Victor Iannello)
    • Since 2014, Freddie has been writing to Victor Iannello (and online discussion groups) with a rumour out of Malaysia. It has not been possible to substantiate the rumour to date.
    • In the days after MH370’s disappearance, a politically connected individual allegedly related to Freddie what was planned for MH370, as follows:
    • The objective of the MH370 diversion was to embarrass Prime Minister Razak into agreeing to call new elections”

    I still maintain the following took place.
    MH370 flown by Captain Zaharie proceeded out into the Andaman Sea came back and around to BEDAX and on to a heading due South and slowed to 370 knots.
    Using waypoints at ISBIX, Cocos Islands and Christmas Island crossed arcs 2, 3, 4, 5, maintaining a steady 370 knots.
    After Christmas Island slowed slightly and crossed the 6th arc on a heading to Yogyakarta airfield.
    The last signal, the 7th arc, placed MH370 some 160nm from the airfield and only 80nm from land.

    Nothing that I have been told or seen since 2014 is convincing enough to say that the Captain deviated from the plan. However for some unknown reason he didn’t land and ran out of fuel and ditched before reaching Yogyakarta.

  508. Nederland says:

    @DennisW
    @TBill

    Have you ever asked yourself the question of why MH370 was ordered to turn off (ACARS data) VHF coverage? Surely switching it off during flight would have left a record in the data logs.

  509. Greg says:

    @TimR

    “a steady 370 knots”
    By the way: it is very close to the performance of Global Hawk.
    But I’m not suggesting anything.

  510. Richard says:

    @Don Thompson

    I have a question about the IFE system message.

    We know that the IFE message expected at 00:21:16 UTC from the aircraft was not received. We also know there are a number of possible reasons why this IFE message was not received, including the possibility that the APU had run out of fuel, the aircraft attitude prevented a line of sight between the aircraft’s satellite antenna and the orbiting satellite relaying the signal or the aircraft had crashed.

    A controlled glide scenario would take between 14 and 22 minutes and the aircraft attitude would normally allow satellite communication.

    We are also told that there was 13 minutes 45 seconds of fuel estimated to be available to the APU.

    Andrew’s analysis shows it is unlikely that there was an engine restart, because if it happened, there would not be any fuel left to power the SDU long enough for its reboot cycle to complete and we would not have received the 00:19:29 UTC Logon Request and 00:19:37 UTC Logon Acknowledge messages.

    It is conceivable, but unlikely, that the aircraft first entered a steep descent of around 15,000 feet per minute, in line with the final satellite transmission received at 00:19:37 UTC and coincidentally, at the time of the IFE message 89 seconds later, was recovering in an attitude that prevented the signal reaching the satellite.

    At 18:28:10.260 UTC, on Channel IOR-T1200-0-36D7, Superframe Number 12980, SU Numbers 0 to 5, we find the IFE message containing the text “Pet Shop Boys”.

    This is followed by a “Acknowledge User Data” message.

    My question is: If MH370 was in a glide at 00:21:06 UTC, but the aircraft attitude was such that the IFE message did not get through and no acknowledge message was received, would the IFE system not automatically repeat the IFE message for a preset number of times until an acknowledge was received?

  511. David says:

    @Dr B. Re my last para in my 8:24 PM above I see @Richard has posed one very similar to @Don Thompson. Don’s answer therefore should address that speculation I raised with you.

  512. Richard says:

    @David

    Apologies! I did not read your comment at 8:24 PM carefully enough and missed the last sentence in particular, where you state “If so, the abnormal attitude would have to persist for there to be no more transmissions. So maybe attitude is a doubtful explanation.”

    So if:

    1. Aircraft attitude is a doubtful explanation.
    2. APU running out of fuel is a doubtful explanation.
    3. Engine restart is a doubtful explanation.
    4. Alternate Electrical Configuration is a doubtful explanation.
    5. The RAT deployment does not power the AES.
    6. The cockpit switch (if implemented) does not cut the power to MEC IFE components.

    May we conclude the aircraft crashed before 00:21:16 UTC?

  513. Andrew says:

    @Richard

    RE: ”6. The cockpit switch (if implemented) does not cut the power to MEC IFE components.”

    The MAS FCOM shows that the IFE/PASS SEATS switch was installed on 9M-MRO.

  514. Richard says:

    @Andrew

    Many thanks for the correction.

    So if:

    1. Aircraft attitude is a doubtful explanation.
    2. APU running out of fuel is a doubtful explanation.
    3. Engine restart is a doubtful explanation.
    4. Alternate Electrical Configuration is a doubtful explanation.
    5. The RAT deployment does not power the AES.
    6. The cockpit switch does not cut the power to MEC IFE components.

    May we conclude the aircraft crashed before 00:21:16 UTC?

  515. David says:

    @Richard. No prob at all.
    “May we conclude the aircraft crashed before 00:21:16 UTC?”, depending on Don’s answer and based on those ifs, I would have thought so.

  516. Brian Anderson says:

    @TimR

    “a steady 370 knots”

    Not long after the first BTO data became available there were many guesses at the aircraft speed.
    I was able to demonstrate, quite conclusively I think, that the mid-flight average speed, at least between 19:41 and 20:41, was around 490 knots. The methodology was rather unique, and quite different from subsequent track models, and it is interesting that the best and most elegant models, almost 5 years later, are also proposing a very similar speed. It is just not possible for the speed to be as low as 370 knots.

  517. Don Thompson says:

    @Richard asked “would the IFE system not automatically repeat the IFE message for a preset number of times until an acknowledge was received?

    Not quite.

    The IFE’s peer is a remote service on the ground, somewhere off in ‘ATN land’. Only when the ISO-8208 connection is established can the IFE ‘talk’ with its peer. The satcom datalink provides the ISO-8208 SSN (satellite sub-network), and operates as a Reliable Link Service. Any retries in establishing or operating the ISO-8208 SSN connection would be evident in the link level exchanges between the AES & GES.

    In the period following 00:19:38, the expected datalink activity is an ISO-8208 SSN Connection Request (in the form of two R-ch bursts). That Conn Rqst sets up the first SVC for IFE. It’s the second Connection Request that establishes the second SVC which carries the infamous ‘Pet shop boys…’ string.

  518. Andrew says:

    @Richard

    If engine restart is doubtful, then perhaps we could also conclude the high rate of descent was the result of deliberate action?

  519. Richard says:

    @Don Thompson

    Many thanks for the response.

    You have pointed out previously that following the cold start 12:50 GES Log On IFE related link activity began 156 seconds after the AES Logon Acknowledge was recorded. You also pointed out above in your comment to David that “the AES waits for a period of between 12 and 19 seconds for a response to its first Log On Request”. Elsewhere you stated, had the first burst in the IFE related exchange been lost, then circa 8-15 seconds additional delay would result, before a retry occurred.

    The maximum delay waiting for the missed IFE message, including waiting for a retry is therefore 156 + 15 = 171 seconds, say 3 minutes. You also remind us above that the AES cannot tolerate an interruption in its reception of the P-channel signal from the ground, for longer than ten seconds.

    So is it reasonable to assume, that the crash occurred within 3 minutes of 00:19:37 UTC, because we saw neither an IFE message, nor an IFE message retry, nor a new Logon request following loss of sync on the P-channel?

    If this assumption is valid, then MH370 is within 30 NM of the hot spot at 34.4°S on the 7th Arc.

  520. Richard says:

    @Andrew

    I agree with your conclusion from the analysis in your paper, that without engine power, the high rate of descent calculated from the final BFO data value at 00:19:37 UTC at around 15,000 fpm and in particular the acceleration of around 0.67g from the previous BFO data value at 00:19:29 UTC, is unlikely without a pilot’s deliberate action.

  521. Victor Iannello says:

    @Richard, @Andrew: If the steep descent suggested by the final two BFO values was caused by a nose-down pilot input, it doesn’t change the fact that the missing IFE log-on request can best be explained by an impact. If this transpired, it remains unexplained why the pilot chose to wait about 2 minutes after fuel exhaustion to enter the dive, but it’s certainly possible.

  522. Andrew says:

    @Victor

    RE: ”If this transpired, it remains unexplained why the pilot chose to wait about 2 minutes after fuel exhaustion to enter the dive, but it’s certainly possible.”

    Perhaps he was simply summoning up the courage for one last act to end it all.

  523. Richard says:

    @Victor

    2 minutes is maybe the time for a final prayer.

  524. TBill says:

    @Richard
    “May we conclude the aircraft crashed before 00:21:16 UTC?”

    I remind readers I said a few days ago I am taking “passive” approach, not voicing active pilot opinions here. When @DonT gave his final list of reasons for no IFE, I was silent, versus my normal rebut after 2-minute delay.

    @Nederland
    “Have you ever asked yourself the question of why MH370 was ordered to turn off (ACARS data) VHF coverage? Surely switching it off during flight would have left a record in the data logs.”

    Please say more about what you find interesting. I was under the impression that China did not use radio for ACARS so I was not aware of any point of interest. Here again it would be nice if we could benchmark against other MH370 flight practices.

  525. TBill says:

    @TimR
    Thanks for checking in. To me the passage of time has not been helped the Xmas Island scenario. It looks a whole lot like the PAX saw depressurization at IGARI, which is not a new idea, it is just reconfirmed that that seems like the logical deduction. Perhaps someone in Malaysia knows more, because the inference of intentional depressurization came from some source way back in 2014…I am not sure where that idea came from but, it seems to be correct.

    Thus if the rumor was correct, the actual threat T (perhaps unknown to the ground team) was apparently loss of aircraft, and the negotiations were over by IGARI. So the question becomes what was ZS’s ultimatum plan? whereas peaceful surrender to Jakarta authorties seems unlikely.

    Further above I list reasons why northern locales above 25 North are frowned upon by many. One reason is drift experts DrGr and DrP are saying not likely.
    Another reason is two big MH370 groups (IG and the 38 South Crowd) don’t agree with north. I would also note that the MH370-Captio/Xmas crowd seems to be awful quiet, suggesting to me they perhaps see limited merit of that idea. Of course, in truth all ideas have limited merit, so that is the base case.

  526. Viking says:

    @DennisW

    It was indeed remarkable that the case was handled at so high level from the start. It indicates that the political leaders knew something deep about the case and feared that an embarrassing secret could have leaked out. Negotiations is one possible explanation, but perhaps there could also be other possibilities?

  527. TBill says:

    @TimR
    Correction above:
    should say- “I list reasons why northern locales above 25 *South* are frowned upon by many…”

  528. Viking says:

    @TimR

    Your information is interesting. However, I find it strange that a flight following such a route did not leave any contrail when two other flights left relatively clear contrails in the area at exactly the same time.

    I also find it strange if Shah told so many details about a plot to ‘hijack’ the plane to anyone. It would have been extremely risky and completely unnecessary.

    Finally, I think it looks unlikely that such an experienced pilot would run out of fuel by some error in the calculation.

    Is it not more likely that he wanted people to believe that this was what happened and instead decided to bail out and disappear?

  529. Richard says:

    @TBill

    You stated “I am taking “passive” approach, not voicing active pilot opinions here.”

    I am voicing active pilot opinions here.

    Please see my comment above @Andrew:

    “I agree with your conclusion from the analysis in your paper, that without engine power, the high rate of descent calculated from the final BFO data value at 00:19:37 UTC at around 15,000 fpm and in particular the acceleration of around 0.67g from the previous BFO data value at 00:19:29 UTC, is unlikely without a pilot’s deliberate action.”

  530. Viking says:

    @TBill

    Indeed I have experienced that most people do not like routes north of 25S. The only strong argument I can see against these is the beaching pattern (including timing). However, I strongly disagree with this argument. First of all there is one (or perhaps two?) hotspot area(s) for the origin of debris somewhere between 25S and 19S.

    More importantly (in my opinion) if the debris originates further north than 19S it may manage to get caught by the tropical hurricane passing two weeks later and get dragged to around 20S where the hurricane looses strength and heads west. It is delicate to hit this narrow window, but not impossible. A really interesting detail is that the end of the contrail I found seems to fit perfectly with this scenario as far as I can see with simple extrapolation methods.

    @Richard

    Are you willing to make one more debris-tracking simulation using these coordinates (instead of the end position I calculated from the Inmarsat data) and include the influence of the hurricane?

  531. TBill says:

    @Richard
    Thank you for considering active pilot, yes I understand.

  532. airlandseaman says:

    Richard, Andrew, others…

    Re “I agree with your conclusion from the analysis in your paper, that without engine power, the high rate of descent calculated from the final BFO data value at 00:19:37 UTC at around 15,000 fpm and in particular the acceleration of around 0.67g from the previous BFO data value at 00:19:29 UTC, is unlikely without a pilot’s deliberate action.”

    I do not agree. Most of the simulations Paul Matson and I conducted on Nov 2, 2014, and many of the Boeing simulations demonstrate that very high rates of descent can be expected for uncontrolled post MEFE case. For example, Boeing’s Case #6 resulted in a ROD >45,000 ft/min. See details here: http://bit.ly/2ZbP31R

    In addition, there are multiple examples of actual flights where the ROD exceeded 30,000 ft/min. So, it is entirely possible…in fact likely…the ROD was 15,000 ft/min at 00:19:37 (whether controlled or not) as indicated by the BFO.

  533. Victor Iannello says:

    @airlandseaman: Even though a high descent rate is achievable without pilot input, what is difficult to match is the 0.67g downward acceleration, which did not occur in the airline simulations you witnessed except for the case where the engine relight caused a rapidly increasing bank angle. There either has to be nose-down pilot input, or a large amount of aerodynamic asymmetry (e.g., excessive rudder out-of-trim, thrust asymmetry) to reach the downward acceleration suggested by the final BFO values within 2 minutes of flameout.

  534. airlandseaman says:

    Victor:

    In the Boeing Simulation I cited above, the acceleration varied between 39 and – 20 ft/sec^2. Also, in the case of Payne Stewart’s flight, that Lear Jet descended 48,900 feet in 150 seconds, exceeding 30,000 ft/min after MEFE. It does happen.

  535. airlandseaman says:

    Here is another example of the high rates of descent that can occur. This one was a crash of a Grob G102 sailplane in Boulder Colorado…one I am very familiar with. The pilot became hypoxic, lost consciousness at 23,000 ft and the plane crashed about 10 minutes later. It underwent dozens of phugoids before finally descending 20,000 feet in the final 2 minutes. The plane hit a high power transmission line shearing off one wing before coming to rest in a field a few miles from KBDU. http://bit.ly/2OkTczz

  536. Don Thompson says:

    @ALSM,

    The crash of Learjet Model 35, N47BA, is not a good comparison to make with MH370. N47BA reached maximum altitude, while still attempting to climb, under autopilot control.

    One engine suffered flame-out, the thrust loss caused a stall, an uncontrolled descent ensued while the remaining engine continued to provide thrust.

  537. Victor Iannello says:

    @airlandseaman: What a number of us have tried to explain was the high rate of descent, AND the high level of downward acceleration, AND the timing within 2 minutes of flameout. With an engine relight a minute after flameout, all of these can be explained. In the Boeing simulations without the relight, the high level of downward acceleration occurs either minutes later, or not at all. On the other hand, a nose-down pilot input would explain what is suggested by the timing and levels of the final BFO values.

    So either the Boeing simulations did not capture the physical mechanisms that might lead to a 0.67g downward acceleration 2 minutes after flameout, or there was a nose-down pilot input.

  538. Andrew says:

    @airlandseaman

    As @Victor mentioned, the issue is not the rate of descent per se; it’s the downward acceleration without control input, and the timing.

    A couple of observations re the Payne Stewart accident:

    1. According to the NTSB, ”The type and degree of damage observed on the left engine was indicative of engine operation at the time of impact.”

    2. Earlier models of the Learjet had nasty handling issues in the event of a high-speed upset, including Mach tuck and aileron ‘snatch’. Several accidents involving high-speed upset led the FAA to undertake a special certification review of the Learjet in the early 1980s.

  539. airlandseaman says:

    Wow! I’m more than a little surprised to hear all this new theory coming up after 5 years. This is the first I have heard anyone suggest the BFO data is only consistent with a high speed piloted descent. Payne Stewart’s flight may have been a poor comparison if one engine was still operational when it stalled, but there are other examples. And gliders don’t have an engine at all, but still have accelerated very rapidly as in the case I cited above.

    I remain convinced that the data is consistent with either a piloted or piloted descent. We simply cannot know for sure until we find the plane, and then we may still not know. But either way, the data indicates the plane is very likely close to the 7th arc, not 100 nm south.

  540. airlandseaman says:

    Correction: “…either a piloted or piloted …” should read “…either a piloted or unpiloted… “

  541. Sid Bennett says:

    @DennisW
    re: your post June 2, 2019 at 11:54 am on Allan variance.

    Thank you for posting the Fig 5.5 excerpt from the DSTG report, saving me the trouble of finding the original :-).

    I refer you to Figs. 11-13 and related text of the Ashton Inmarsat paper where temperature effects of the sun eclipse period are discussed.
    Once the underlying Doppler effect of the satellite motion is compensated, there remains a transient effect due to the strong temperature variation during the eclipse time. Since the comparison flight is on 2 March, the time of peak deviation will be different, but appears consistent with an eclipse temperature variation. The presented data is not anomalous. Rather it shows the uncompensated eclipse effect.

    Thus, the DSTG data should not be used to denigrate the value of the BFO data.

  542. DennisW says:

    @Sid

    I am not denigrating the DSTG data. I am denigrating the subsequent mathmatical interpretation.

    BTW, sk999 and I have beat the eclipse effect and the evolution of the March 8 data to death.

  543. Andrew says:

    @airlandseaman

    RE: ”Wow! I’m more than a little surprised to hear all this new theory coming up after 5 years. This is the first I have heard anyone suggest the BFO data is only consistent with a high speed piloted descent.”

    This issue has been discussed previously. I can clearly recall comments that suggested the timing of the descent and the vertical acceleration were unlikely to occur without some kind of yaw input (eg. engine relight), or a deliberate control input.

    The paper I wrote regarding the relight issue was first posted some 12 months ago, however, I don’t recall the implications for a piloted vs unpiloted descent being discussed at that time; it only resurfaced in a recent discussion. If seems to me that if an engine relight was unlikely, the most likely explanation for the vertical acceleration and timing of the descent is a deliberate control input. Nevertheless, I totally agree that the aircraft is likely to have impacted the water close to the 7th arc. I am not a fan of the dive/recovery scenario that some have suggested.

  544. Victor Iannello says:

    @ALSM: What new theory is coming up after 5 years? This topic has been discussed extensively under the following two posts:

    MH370 End of Flight with Banked Descent and No Pilot
    End of Flight Simulations of MH370

    Those posts and discussions are consistent with my previous comment: So either the Boeing simulations did not capture the physical mechanisms that might lead to a 0.67g downward acceleration 2 minutes after flameout, or there was a nose-down pilot input. That was not a “new theory”.

    The timing of the increasingly steep descent has never been explained for the case of no pilot inputs, other than by invoking physical mechanisms not included in the Boeing simulations, which I don’t rule out.

  545. Sid Bennett says:

    @DennisW
    Can you point me to any posts or papers that I can review?

  546. Barry Carlson says:

    @all,

    I think we have sufficient evidence to conclude that a benign stall at circa FL350 will rapidly lead to a descent of 15,000 ft/min with the wings being held level and the attitude maintained through the ‘barn door’ effect of control surfaces. Should a wing drop, then the nose will also drop and the drag decrease accordingly, resulting in acceleration and a higher RoD. If still under engine thrust at this time, the acceleration will continue to increase, otherwise the RoD will be limited by the drag to reach the terminal velocity.

    None of the above can conclusively prove a piloted or non piloted descent.

  547. Nederland says:

    @TBill

    I am currently on travel and responding slowly. Nevertheless, I think it is a bit strange that there was an individual briefing to turn off ACARS/VHF if this was standard practice on all flights to China (because of the different network, ARINC rather than SITA). This was an information that was released only after the FI and SIR reports came out.

    An early statement by a MAS pilot was:
    It is Mas procedure to switch ACARS, VHF, and High Frequency selection off but this is only for flights to China as the service provider for Mas does not cover China. Some if not all pilots switch them all off for a while and then later switch SATCOMM back on to force the system into SATCOMM mode.
    Not sure if that means the normal procedure was to switch off ACARS VHF mode in flight before entering Chinese airspace.

    I think because ACARS was in SATCOM mode only, there were no suspicious log records when ACARS was disabled before the diversion.

    I also think the test message at 18:39 was sent via VHF, which is a bit strange too since the VHF mode was disabled.

  548. Victor Iannello says:

    @Nederland: At 15:54:31, there was an ACARS message via VHF sent from MAS OPS to 9M-MRO to switch VHF3 (the radio normally used for VHF data) from DATA (131.550 MHz) to VOICE if the SATCOM was serviceable. Shortly after (15:54:53), the VHF ACARS link was lost and subsequent ACARS messages were sent via SATCOM. So it appears that the VHF link was disabled before the flight.

  549. TBill says:

    @Nederland
    Last sentence, I believe you are referring to the ACARS TEXT message that MAS sent to MH370.

    I am still not understanding totally, but are you suggesting that there was a superfluous (unneeded) message to deselect VHF from ACARS?

    Wonder what MH371 did with ACARS, on the day before? Speculation, but also could be coded message to say “no Go” on negotiations, but with ZS on his cell phone on the runway, we already know “sterile cockpit policy” was not in force, and clandestine conversations are possible.

  550. Ventus45 says:

    @TBill

    Perhaps, (if there was a need to confidentially conduct or reject negotiations) it may have been an attempt to get MH370 to clear VHF3 from data to voice, so that the frequency (that very few people would be listening to as “voice”) could be used by any other MAS aircraft (hopefully one within 400nm range) to attempt contact by voice, in Malaysian, instead of using any open frequency.

  551. DennisW says:

    @Sid Bennett

    @DennisW
    Can you point me to any posts or papers that I can review?

    I am not sure what information yu are seeking.

  552. Sid Bennett says:

    @DennisW
    I have read your paper of June 2 and wanted to know if there was any other available information on the analysis by you and sk999 of of the BFO data.

  553. DennisW says:

    @Sid Bennett

    The short answer is no. Our conclusion was that the eclipse effect is captured by the current Inmarsat BFO data.

  554. Viking says:

    @All

    I am surprised to see no reaction to my claim that an end position near 34.4S means that the motive for the diversion must have been suicide. In particular since a lot of details speak against suicide:

    (1) In all other cases of pilot suicide leading to loss of large passenger jets the suicide took place within a few minutes after the PIC brought himself into exclusive control (typically by locking the cockpit door). This makes MH370 a completely unique case.

    (2) Shah was not a typical suicide candidate, since he did not suffer from any (even secretly diagnosed) mental illness, did not have severe economical problems, had no links to terror, and perhaps most importantly seemed to have an outward reacting personality (just look at his videos), who rarely commits suicide.

    (3) A lot of effort was used to make the airplane disappear, so if the purpose was suicide it must have been extremely important to hide it. However according to TimR, Shah told one or more friends of some plot against the flight, and according to leaked information he cancelled all future appointments in his calendar. In addition, he made a tentative simulation on his private flight simulator and deleted the results in a careless way. These details do not point to any intentions of a hidden suicide. I find this contradictive evidence disturbing!

  555. Victor Iannello says:

    @Viking: We can’t know the motivation for the diversion, but that doesn’t mean we can’t scientifically deduce the who, how, and where. Suicide remains a possibility, as well as a diversion that went wrong.

    Regarding your three points:

    1) Yes, this is a completely unique case in MANY ways. You can’t use the uniqueness of any scenario to argue against it. (I would say even your preferred scenario is quite unique.)

    2) Our knowledge of the captain is from unofficial interviews of people around him, his online activities and the RMP report. Those around him have described behavior that is detached, unhappy, and reckless. I don’t put a lot of weight on the RMP report, which I believe was deliberately written to put him in the most favorable light. The RMP has a history of writing reports under political pressure to whitewash crimes.

    3) Rumors like TimR presents are not proven (despite my many attempts to get more facts). Those rumors could be disinformation. As for deletions on his simulator, I don’t know how many people understand how Volume Shadows work, or even if the deletions were meant to be found.

  556. ArthurC says:

    Is it at all possible that someone (passed out or otherwise incapacitated) in the pilot’s seat fell forward and tipped the column forcing a downward descent (possibly explaining the 0.64g) AFTER the plane flamed out and maybe took a slight dip down?

  557. Viking says:

    @Victor

    Pilot suicide in itself is fortunately very rare. However, all the known cases share one important detail: They were carried out quickly. I guess this is not at all strange, since if one uses too much time there is a significant ‘risk’ someone may manage to get through the reinforced door with the fire axe. The captain on German Wings was working hard with the axe and almost made it. Have you ever seen the axe in use during a flight? I have, and it is BIG.

    Indeed some of the conflicting information is unconfirmed by reliable sources. However, it seems to come from different sources, so the probability is relatively small it is all biased in the same direction.

    Concerning the rarity of a scam, I am not so sure you are right. For instance the trigger for my detailed investigation of MH370 was when a Danish whistleblower confirmed that the Scandinavian Star accident was not an accident, but a scam (back in the beginning of 2016). The number of casualties were comparable to MH370. You may argue it was a ferry, not an airplane, but I do not think it makes much of a difference concerning the motive.

  558. Victor Iannello says:

    @Viking: You are either misinformed or cherry picking information if you think all of the rumors are “biased in the same direction”. Far from it.

    In the recent work undertaken by Bobby, Richard, and me, there is no presumption of the motivation of the disappearance. The only assumption is that of automated flight after 19:41, which is presented as a null hypothesis that is then proven.

  559. Victor Iannello says:

    @Viking: Let me add that I don’t totally dismiss the possibility that the captain jumped. If so, it might have happened over one of the sparsely populated Nicobar Islands. However, the only way I can imagine this occurred is for a jump out the bulk cargo door, and the only way that door was accessible was if the cargo was not loaded in the aft bay in accordance to the load sheet. That would mean collaboration on the ground at KLIA as well as collaboration near the jump site. The scenario rapidly gets very complicated.

  560. Tim says:

    @Victor,
    May I ask, what is the evidence you have for an autopilot ‘on’ flight from 1941z ?

    This is important, if it was off, MRO would have entered a right-hand roll and steep spiral descent as soon as the right engine failed. And assuming an alternate power scenario with the right IDG and right Backup having been the power source, the APU/RAT would come online even though the left engine was still functioning.

  561. Victor Iannello says:

    @Tim asked what is the evidence you have for an autopilot ‘on’ flight from 1941z?

    If there is only one automated path the meets unbiased criteria for an acceptable solution, it is very unlikely that this path was not followed. As I’ve said before, the challenge is to show that the criteria is proper and unbiased, and the path uniquely satisfies the criteria. In the next paper, we’ll learn more about this.

    This is important, if it was off, MRO would have entered a right-hand roll and steep spiral descent as soon as the right engine failed.

    If the left engine was still operating and the main left bus or the left backup generator was supplying power to the left transfer bus, then the flight control mode would remain as NORMAL and the thrust asymmetry would be mostly compensated after the right engine failed.

    And assuming an alternate power scenario with the right IDG and right Backup having been the power source, the APU/RAT would come online even though the left engine was still functioning.

    True, after their respective delays. In the alternate configuration, as the right engine spooled down, the thrust asymmetry compensation (TAC) would be progressively applied until the flight control transitioned to SECONDARY with the loss of power to the right transfer bus. Then later, after the left engine failed, the TAC would remain frozen, and the aircraft would yaw and bank to the left.

  562. Peter Norton says:

    @all:
    Does anyone here have contacts at Malaysia Airlines ?

    I was thinking that it would be very interesting if they compared MH370’s data and ATC coms to previous flights.

    Some examples that come to mind:

    – Did the pilots always read back the frequency at handover ?
    – Did the pilots ever report arriving at/maintaining a specific altitude? Twice?
    – Were any stats out of the ordinary?
    – Anything (occurring prior to IGARI) that was different than at all other flights?

    There might be clues buried in a comprehensive comparison of flight characteristics/statistics.

  563. Mick Gilbert says:

    @Victor Iannello

    Victor, somewhat off piste to the current discussion, can I please ask what is your understanding of the FBI’s involvement in the investigation.

  564. DennisW says:

    @Victor

    If there is only one automated path the meets unbiased criteria for an acceptable solution, it is very unlikely that this path was not followed. As I’ve said before, the challenge is to show that the criteria is proper and unbiased, and the path uniquely satisfies the criteria. In the next paper, we’ll learn more about this.

    I am waiting, but running out of patience.

  565. Nederland says:

    @TBill

    Victor wrote: Shortly after (15:54:53), the VHF ACARS link was lost and subsequent ACARS messages were sent via SATCOM. Yes, in order to disable ACARS you need to first turn VHF mode from data to voice, and that leaves a record in the SITA protocol. Now if that had happened after the aircraft was “taken over”, that means shortly before the diversion, then it would be clear even to the die-hard accident proponents that this was an intentional, nefarious act. Also, Z. arrived early in the cockpit. That means he would be the natural suspect if he turned off ACARS VHF, while on his own, without instruction.

    Any possible further conclusion would depend on what the standard policy is. Again, I would have thought that if it is general company policy to turn off ACARS VHF mode for any flight to China, that there would be no special message. Pilots would simply know it. If it is general policy to turn off ACARS VHF mode AND send a message requiring the pilots to do that, than this specific flight to China was ideal for one of the pilots to disappear without leaving a trace. If, on the other hand, MAS OPS does not normally send such a message, and the pilots turn off ACARS VHF mode at some point later during the flight, then yes perhaps this could be seen as a coded message.

  566. Richard says:

    @DennisW

    Please accept our apologies that you are having to wait for the next paper and are running out of patience.

    It is a hell of a lot of work!

    We have tons of stuff. Fuller explanation of the method, more analysis, more precise results, test cases, sensitivity studies, …

    In short, the paper will not be short.

  567. David says:

    @Andrew. Given that high altitude reduces the prospect of an engine relight the question arises as to what is its effect on the APU auto-start probability? Had that failed but another succeeded at lower altitude that could bring the log-on BFO timing more into line with the Boeing simulations. Something needs to.

    As a corollary, with no engine relight at high altitude, as you know the RR would have shut down when its N3 dropped to 35%, with no prospect of a relight thereafter. Thus an aborted APU start would remove the possibility of a relight affecting the descents simulated by Boeing. That would leave just the drag from the APU’s open inlet during the first start attempt, plus 30-40 secs either side, while that opens and closes, unsimulated. (I leave aside the simulation verification issue).

    The APU has a centrifugal compressor, which should have a less sensitive airflow than the main engines’ axial. At the same time the total air pressure recovery at its compressor inlet will be less given its bent intake, its intake being in the fuselage turbulent boundary layer, airflow then being split between it and its pneumatic load compressor, and the intake air providing a cooling bleed; less that is when compared to the recovery to be expected in the engine intakes. Also, in the air, the APU’s lubricating oil heater is powered by the right main ac bus, so de-powering that could increase starting drag had there been several-hours cold soak at altitude.

    Given the delay from an aborted start and its likely increase on descent rate, the 00:19:29 LOR might not have been transmitted when the aircraft had a low ROD. However, you are aware that Holland (DSTG) allows up to 14,800fpm for that, the associated 00:19:37 LOA descent rate being 25,300fpm, with the approx. 0.68 ‘g’ descent acceleration between.

    As to whether this hypothetical transmission delay might have increased the chances of the link to the satellite failing due to aircraft attitude, should this LOR have failed the Safety Investigation Report (p384) implies that re-transmission would employ the low gain antenna. Since in the event the LOR was recorded at the high gain antenna frequency it would seem that attitude did not intervene.
    (The 8 secs between LOR and LOA indicates that there was no need for LOA re-transmission.)

    Also, a log-on delay like this after an the increased descent would lead to those transmissions having been from a lower altitude, possibly offering a smaller prospect of recovery, manned or not.

    So those transmissions being the result of a second APU auto-start would go some way to reconciling Boeing simulations with an assumption that there were no pilot inputs from the time of the final course being set.

    Thus a question is whether flight testing of APU auto-start at high altitude has demonstrated that it could be relied on to auto-start there. If not, or it faltered, another question is whether a second auto-start would be attempted, and in time, once the first failed. The Training Manual (TM) says, “You can try 3 APU starts in a 60 minutes period with the electric starter motor.” It does not say what the minimum gap between first and second should be or whether that is just guidance.

    The TM acknowledges that there can be ignition failure during a start.
    To me the nominal conditions for an auto-start are repeated at shut down from an aborted start, those being no transfer bus power, aircraft in the air. While a pneumatic start would not be available for either start, the APU battery could well provide for a second try, given the three starts mentioned above, supposing that limit is dictated by battery re-charging. If for APU cooling, delays between the starts would not apply anyway since the APU would be cold; even if the starter motor isn’t.

    So questions are to what altitude, installed in the aircraft, has the APU auto-start been verified (the TM just lists its sea-level operating limits) and whether its auto-start initiator, the ELMS, and controller, the APUC, would set it starting a second time, where the higher air pressure lower down might help, and in time.

    The TM says, “To shutdown the APU after an autostart you must first move the APU selector to ON and then back to OFF”. However it does not describe what shuts it down if there is just an auto-start attempt. However it does say elsewhere, “During an APU start, if the APU stops its acceleration for more than seven seconds, the APUC shuts down the APU.” It does not say though what shut down procedure would be followed or how long that would take – the normal shut down above 14,000ft (Aircraft Maintenance Manual (the older TM 22,000) takes about 4 mins, much of it for general cooling; below that, 3.

    The TM does say also, “APU engine RPM must be less than 12 percent RPM and the air inlet door must be fully open before the start.” However the circumstances envisaged there are unclear since normally the inlet door starts to close on run down to 15% and it needs to be must be fully open a restart will commence.
    Perhaps you have access to the APU manual or another source of information on this and other questions? I have found nothing in the FCOM and test bed reports that I have are not pertinent.

    One other thing. While able to deliver 120kW AC at all altitudes, APU fuel consumption during SDU reboot would rise a little at a lower altitude and would step-increase beneath 22,000ft, when it would start assuming pneumatic load. The pneumatic delivery would power the air driven hydraulic pumps (supplementing the RAT hydraulic supply) and the environmental control system. However there should be enough fuel for the reboot providing aircraft attitude did not reduce that accessible. That has not been modelled or I believe flight tested. About that, in the ALSM simulation where the engines suffered fuel exhaustion together, there was what appeared to me to be a second APU start, when the aircraft nose was high. It stopped again when the nose dropped. However I might have misinterpreted the indications of that.

  568. Tom O'Flaherty says:

    I’ve never been a big fan of the negotiation theory.

    At that time of night with most of the chain of command leading to Najib Razak in bed, the perp would likely have run out of fuel before any meaningful negotiations could be concluded.

    The stakes would be raising by the minute and ‘the house’ (i.e. the government) is going to win. Better to negotiate on the ground. Also with the plane going dark how could reliable communications be established?

    If it comes under the category of a rumour that started to circulate once no wreckage was found in the initial surface search then the cynic in me would dismiss it as a manufactured theory – just like so many of the others.

    Returning to the topic of discussion I’d just like to say thanks once again to the authors- this discussion has underlined the possibility that someone was in control right to the end.

    I remain confident that reasoning and refinement of the analysis of what we do know will be what allows this plane to be found.

  569. Don Thompson says:

    @Ned wrote “Yes, in order to disable ACARS you need to first turn VHF mode from data to voice, and that leaves a record in the SITA protocol.

    Switching VHF3 from data to voice leaves no record.

    It seems that every attempt @Ned makes to assert a fact, when making a comment here, involves incorrect information.

    Suffice to say, the remainder of that >comment is nonsense.

  570. Victor Iannello says:

    @Don Thompson said: Switching VHF3 from data to voice leaves no record.

    At 15:54:31, the ACARS link was VHF, and after MAS OPS advised 9M-MRO to switch VHF3 to voice, the SATCOM link was established and the VHF link was reported (over the SATCOM link) as lost. So it would seem that if switching VHF3 from data to voice changes the active link, there would be a record.

    On the other hand, to address @Nederland’s comment, those conditions were not met in flight, as the active link was already SATCOM.

  571. TBill says:

    @Nederland
    If you and Victor are correct about the record, that would be another reason for picking the Beijing flight, aside from 2 crew members and other factors. DennisW always wondered the reasons. Also the Chinese NoK PAX leader believes (incorrectly) that mechanical failute is indicated due to lack of signal …he is convinced there would be an automated signal from the aircraft if comms were cut by the pilot. @Haxi was going to follow-up with him for me.

    @David
    If engine re-light is difficult at higher altitude, then you gave me another point for my list of 10-possible reasons why the pilot did an intentional dive at the final BFO.

    @Tom O’Flaherty
    If we accept William Langewiesche and ALSM’s re-confirmation that intentional depressuization seems to be the logical deduction at IGARI, I agree it seems that negotiation (if any) was over by the time aircraft reached IGARI.

    I have just added a new page (July-2019 Addendum) to my Freddie/@TimR rumor essay reflecting that the actual ultimatum threat might have been crash of aircraft, whereas surrenduring to authorities in Jakrata was a bluff. I also mention it is not just TimR’s account, there were similar rumor accounts in social media, which again are difficult to verify.

    https://twitter.com/HDTBill/status/1080510470841348096?s=20

  572. DennisW says:

    @Richard

    My comment was not intended to be negative (although it came out that way). Please ignore it. I do appreciate the difficulty.

  573. Don Thompson says:

    @Ned,

    The statement “Yes, in order to disable ACARS you need to first turn VHF mode from data to voice, and that leaves a record in the SITA protocol.” is incorrect.

    If the statement refers to circumstances at 15:54:53, where ACARS over VHF data service was disabled, then it’s incorrect to say “you need to first turn VHF mode from data to voice. My observation, reading the ACARS Traffic Log is that intervention was first made at the ACARS Manager, before the operating mode of the VHF-C radio was reported.

    As ever, there is technical nuance in the operation and control of the datacomms services.

    VHF data comms, as existed on 9M-MRO, would be termed 'AoA', ACARS over Analogue (radio). That implementation is very basic and does not involve an underlying link layer function. The implication of this statement is determination for a failure is only possible when an expected message is not received, e.g. a periodic broadcast from the ground network, or an air-to-ground transmitted message is not acknowledged.

    At 15:54:31 the ACARS Log shows that MAS OCC sent the msg instructing the crew to change the state of VHF3;
    At 15:54:41 the ACARS Log shows an SA msg indicating that SATCOM is the active medium, SATCOM AND VHF remains available;
    At 15:54:53 the ACARS Log shows an SA msg indicating that VHF service has been lost.

    The ACARS Log shows that VHF was regarded as available for data service when the SATCOM became the active link, then some 12 seconds later the VHF service was lost.

    The determinant action to make SATCOM active was to uncheck VHF in the ACARS Manager screen.

    The remainder of the comment above is irrelevant. The instruction to manually control the active data service, VHF vs SATCOM, was driven by the compararive cost of using ARINC/AEROTHAI’s communications services vs SATCOM within the service scope MAS contracted with SITA. Relying on the default priority/selection choice within DCS for VHF would mean roaming onto the ARINC/AEROTHAI VHF network. ARINC/AEROTHAI provide services throughout much of SE Asia.

    [*] DCS – i.e. all elements related to Data Communications, including the AIMS Data Communications Management Function.

  574. Victor Iannello says:

    @All: For improved security, I have implemented https on this site, which only allows encrypted data exchange. The site should automatically redirect http requests to the https protocol. Please let me know if you are experiencing any problems caused by this change.

  575. Don Thompson says:

    Apologies, correction required above (so many acronyms):

    would be termed ‘AoA’, ACARS over Analogue (radio).

    should read

    “would be termed ‘POA’, Plain Old ACARS.”

  576. Andrew says:

    @David

    RE: ”Thus a question is whether flight testing of APU auto-start at high altitude has demonstrated that it could be relied on to auto-start there. If not, or it faltered, another question is whether a second auto-start would be attempted, and in time, once the first failed.”

    The APU in-flight start is very reliable, even when the APU is cold-soaked. The EDTO (formerly ETOPS) guidance material states that ”It should be shown that there is a very high probability that, after the failure of any one or two generated sources of power, the APU can be started without delay at any altitude up to and including the aeroplanes certificated altitude.” The airline where I work regularly conducts APU in-flight starts as part of its APU health monitoring program. The APU is cold-soaked for several hours during a scheduled flight and is then started at any level above FL300. The APU is required to start within three attempts, but in my experience it has started on the first attempt every time. The start is normally conducted towards the end of a flight, after the aircraft has reached its highest cruising level, typically around FL390.

    I think the APU autostart would be similarly reliable, except in the case of an inoperative APU oil heater, as you mentioned. The TM states that the heater ”increases starting reliability when the APU has been cold-soaked”. I know the APU autostart function is checked during some airworthiness test flights, but I do not know the altitude at which the check is conducted.

    The autostart is initiated by a signal from ELMS to the APUC upon loss of power to both transfer buses. The manuals do not describe what happens if the autostart fails, but I suspect there would only be one attempt. If that failed, the crew would be alerted and would then need to attempt a manual start. The APU start limitations are intended to prevent the starter motor overheating. The time limit between start attempts is one minute.

  577. ST says:

    @Victor – Thank you for the update to the secured site. There are no error messages and the blog loads smoothly for people like me who read this thread several times each day.

  578. Victor Iannello says:

    @ST: Thank you for the feedback!

  579. Nederland says:

    The way ACARS is disabled is described in the SIR, p. 375f.:
    On the event flight, voice was selected for the Center VHF on the ground which resulted in the ACARS using SATCOM for the data transmissions, as shown in the SATCOM Ground Station Logs … If both boxes are deselected, ACARS loses the capability to send downlink messages, but can receive and display uplink messages.
    So, turning ACARS VHF from data to voice does leave a record.
    It is still not clear if MAS sends such a message, advising any flight to China to force the system into SATCOM mode, regularly as part of the ground briefing. As far as I know, the VHF link works with no roaming charges for Malaysia and Vietnam.

  580. David says:

    @Andrew. Much obliged. That possibility gets bowled out therefore except for a special circumstance and one that would not allow the ALTERNATE configuration.
    I was about to raise an erratum to what I had said and thank you for not being distracted by an error or two.

    So under the assumption that a left engine relight is unlikely, in my opinion the final descent criteria remain inconsistent with Boeing simulations, the last possibility to correct that now being the inclusion of the effect of a first-try APU auto-start.
    Supposing engine windmilling hydraulics and RAT deployment are simulated already, the principal effect from an APU start most likely would be reduction to left roll rate from any flaperon asymmetry induced by RAT deployment. This asymmetry would arise only if the L1 and right ACEs had both been unpowered by PMG and, in the L1’s case, the small battery powering the left PSA had been exhausted.
    That effect or none, APU power would tend to stabilise the aircraft, so reducing further the chance the aircraft would meet those descent criteria, while the APU was running at least.

    In my view that reinforces the descent being man-made and reduces the simulations’ standing both as support for the current interpretation of the BFOs and as the basis of unmanned search width. However the interpretation of the BFOs can stand alone anyway and the simulations have neither the sample size nor fidelity to set search width. To me that depends instead on the likelihood or not of a person inducing a rapid descent then recovering. I think that unlikely.
    A conclusion that most likely the descent would continue would be the basis for replacing simulator predictions.

    @Victor. No problems down where I am.

  581. Victor Iannello says:

    @David: I think you are landing the same place as me. The base case is a crash close to the 7th arc, whether the dive was controlled or not controlled, and this should be used to define the primary area. However, the possibility of a recovery glide cannot be dismissed, and should be used to define the secondary area. If there was a dive-glide-dive, I think it is likely that the plane continued on the due south path (or close to it), since we believe that was the course for the preceding 4.5 hours.

  582. Victor Iannello says:

    @All: There is an excellent article in The Telegraph about the MH370 investigation, including references to the French investigation. There are many sensible quotes from Mike Exner.

  583. Richard says:

    @Victor

    You stated “However, the possibility of a recovery glide cannot be dismissed, and should be used to define the secondary area.”

    I disagree. A glide can be dismissed.

    Andrew has shown that an APU start is almost guaranteed even at altitude.

    He also has shown an engine restart is unlikely at altitude and can be dismissed, otherwise we would not have seen the SDU reboot at 00:19:29 UTC.

    Don has shown that the IFE system message would be repeated within a short timeframe, if the aircraft attitude was an issue, and repeated until an acknowledge was received.

    In a glide the aircraft attitude is not an issue for satellite comms.

    Don has also shown that if the satellite connection was lost, a new logon request would be made after a short timeframe.

    Don and Andrew have shown that the IFE switch on the overhead panel does not effect the IFE comms, by inspecting the work instructions of the AD.

    Andrew has shown that a downward acceleration of 0.67g can only be achieved with pilot input.

    Bobby has shown that the alternate electrical configuration does not fit the data we have.

    The RAT does not power the SDU.

    I do not see any possibility of a glide scenario.

  584. Victor Iannello says:

    @Richard: We don’t know for sure that the APU was running after the last AES transmission. Extreme attitudes could have led to lowering the tank level below the pipe inlet.

    From a practical point of view, if OI decides to search a 30-NM radius around the hotspot at -34.3,93.78 and comes up empty, I would recommend continuing the search to the south based on the possibility of a glide. If there were no possibility of a glide, it would be pointless to continue the search.

  585. DennisW says:

    @Victor

    From a practical point of view, if OI decides to search a 30-NM radius around the hotspot at -34.3,93.78 and comes up empty, I would recommend continuing the search to the south based on the possibility of a glide. If there were no possibility of a glide, it would be pointless to continue the search.

    I find your comment a bit unsettling since the “hotspot” to which you (and others) refer has not been subjected to peer review.

  586. Victor Iannello says:

    @DennisW: It was a hypothetical, not a recommendation. On the other hand, if the analysis shows that the BEDAX-SouthPole path is objectively the “best”, I think a recommendation similar to that is in order. We’ll see.

  587. Ventus45 says:

    @Richard

    As I see it, use of the “missing IFE” to justify the “no glide” assumption, itself depends on the assumption that it should have been received, because the primary assumption is that the APU was still running at the time the IFE logon was expected.
    But, the question is, was it “still running” ?

    APU run time depends firstly on what amount of fuel was actually able to be supplied to the APU, and secondly, whether or not it could all be pumped to, and or sucked through the line, to actually keep the APU running.
    My reading of all the APU fuel discussions does not convince me that the APU “must have still been running” for the IFE logon, given the length of line, attitude, fuel slosh, g-loads, pump inlet placement etc etc.

    In short, I think it is a brave leap to assume the APU kept running throughout the descent as most seem to have assumed and accepted.
    If we can not determine “with certainty”, that the APU was not starved of usable fuel at some point prior to the expected IFE log on time, then logically, we can not determine, “with certainty”, that it was still actually running, and powering the SDU.
    Without that certainty, the “no IFE” logic to justify the “no glide” determination, falters somewhat, in my opinion.

    I now see that Victor has made a similar comment re APU run time.

  588. sk999 says:

    Question r.e. frequency calibration in the SATCOM system.

    It has been suggested (I believe it was ALSM) that the SATCOM on MH370 carried out a recalibration of the onboard OCXO, causing a jump in the BFO of 16 hz. Is there any other recalibration done along the signal chain from the aircraft to the output of the receivers at Perth? In particular, would the reference frequency generator onboard the satellite also be recalibrated periodically? Or possibly that of the ground station (seems less likely)?

  589. Victor Iannello says:

    @sk999: First, here is what we know from the ATSB:

    There is an automatic SDU calibration process to correct log-term OCXO drift. Analysis of the GES logs from 9M-MRO from 04/03/14 to 08/03/14 shows three such probable calibration events at: 04/03/14 11:43:32 (-15Hz), 06/03/14 00:35:25 (+16Hz) and 07/03/14 16:00:28 (-17Hz). The actual adjustments will have been +/-16Hz, and these three calibrations appear to show a system adjusting around nominal as seen in SDU logs from other aircraft. This calibration is performed after logging-on and, if necessary an adjustment is made. For an adjustment to be made, one hour must have elapsed since power on, and 26 hours of on-time must have elapsed since the last adjustment. It is not possible to know from the GES logs, when and for how long, the SDU on board 9M-MRO had been powered on prior to flight MH370.

    We also know that the GES is synchronized to a very accurate, stable oscillator, and the MITEQ pilot receiver compensates both for long term frequency drift of the satellite and diurnal frequency variations due to Doppler shift. It would seem that there is no need for re-calibrations other than the AES.

  590. Don Thompson says:

    @Victor

    Concerning the statement from the ATSB, “It is not possible to know from the GES logs, when and for how long, the SDU on board 9M-MRO had been powered on prior to flight MH370.

    With sight of the full Stratos GES Log, showing the complete set of metadata for each record, it is possible to know when and for how long the AES had been powered on prior to flight MH370.

    9M-MRO made an initial Log On with POR at 12:50:19UTC, it remained powered before it executed two Log On renewals via POR at 15:55:57 and 15:57:49, then commanded a Log Off from POR, immediately followed by an initial Log On with IOR at 15:59:55.

    The APU shutdown report, generated at push back time, shows that the APU run time was 4 hours (rounded). The APU report corroborates my reading of the GES Log that the AES was powered on throughout the period from 12:50:19UTC.

  591. Victor Iannello says:

    @Don Thompson: I don’t think this was in question after Ian Holland released his paper. He uses the GES logs to determine the range of time that the SATCOM was shutdown prior the log-on, and reported the results in Table II. That table shows no power-ups between 12:50 and 18:25. The statement from the ATSB preceded Holland’s work.

  592. Don Thompson says:

    @Victor,

    I agree with your comment.

    My intent was to point out, that with the information now to hand, we can be confident about circumstances for the period 12:50 through to the calibtration event at 16:00:28

  593. Niels says:

    @VictorI, others

    Is it known what the calibration “target” / criterion is for applying this +-16 Hz adjustments?

  594. Richard says:

    @ Ventus45

    ATSB confirms there was at least 30 lbs of fuel for the APU in the fuel line, which is sufficient for 13 mins 45 secs under load.

    Andrew confirms that the APU is very reliable to start and to continue once started.

    I have no reason to believe that the APU died after 89 secs.

  595. Victor Iannello says:

    @Niels: The frequency error is generated by the difference between the received and expected frequency of the P-channel signal. Because that signal is not pre-compensated for Doppler shift (I think), the calibration will not be “exact”. As the BFO is around 150 Hz and not zero, that means there are other offsets in the path, including the Doppler shift.

  596. Victor Iannello says:

    @Richard: Here’s what the ATSB says:

    The APU fuel inlet is located in the left main tank. The APU is estimated to consume (when electrically-loaded) approximately 2 lb of fuel in 55 seconds. In a standard flight attitude (1° pitch), the difference in location between the left engine fuel inlet and the APU fuel inlet would result in approximately 30 lb of fuel being available to the APU after a left engine fuel exhaustion. From this information, the APU had a maximum operating time of approximately 13 minutes and 45 seconds. The pitch attitude would have an effect on the usable fuel for the APU; an aircraft not under control may exhibit dynamic changes in pitch attitude (i.e. phugoid motion) which could have limited the APU’s ability to receive fuel. In-flight acceleration forces could also affect the distribution of fuel in the tanks.

    First, once the fuel level fell below the APU fuel inlet, it is doubtful that any fuel in the fuel line would be available to the APU. The DC pump would cease to pressurize the line, and likely the line would cavitate, especially at high altitude.

    Second, we don’t know if there was an extreme attitude that caused the level in the tank to fall below the APU inlet. If this occurred for any significant length of time, the APU would likely run dry.

    It’s possible that the APU continued until impact, but I don’t think we know this for sure.

  597. Sid Bennett says:

    I am glad to see that the BFO issues are being revisited. I am in the process of consolidating some of the published evidence and trying to locate some of the studies I circulated to the IG in 2015 or so.

    I see the issue a little differently.

    The ATSB published a graph showing the effects of the solar eclipse of the satellite and indicating that this would produce an uncompensated error of about 8Hz. (I believe that the error is slightly greater. The initial analysis used linear interpolation between the ping times.)

    I digitized the Fig. 12 and 13 of the Inmarsat paper and was able to show a high correlation between the temperature variation of the translation oscillator and the frequency variation. Thus, although only measured at the pings, we have a continuous trace of the actual frequency error due to temperature variations in the satellite. There are smaller frequency corrections that can be made (due to the eclipse effect) at the 1941 and 2141.

    I determined a coefficient relating temperature and frequency error, but omitted to look into the resultant statistical error in using the coefficient. If we posit that most of the variation in BFO is related to the temperature variation of the translation frequency, then the difference between the calculated (using temperature) and the measured oscillator frequency would yield a residual which could be taken as an estimate of the actual statistical error in this measurement.

    I was prompted to revisit the subject by the recent citing of Fig. 5.4 of the DSTG report where BFO residual data for a flight of 2 March exhibits a large excursion in the 22h region. Both the magnitude and the shape of the deviation are consistent with Fig.12 of the Inmarsat paper and is likely to be a thermal effect. As such the large estimated error used by DSTG seems to me to be in error and means that the model is not as sensitive to the actual measured values as it should be.

    I will attempt to organize some of this data this week.

  598. Richard says:

    @Victor

    You cannot have a long glide scenario and at the same have an extreme aircraft attitude.

  599. Victor Iannello says:

    @Richard: “Extreme” was probably not the right word. We’re really talking about a change from the 1° of pitch that was used to derive the estimate of 30 lb of available fuel for the APU. In the dive-recovery scenario, you could have both nose-down and nose-up pitches much different (>10°) from 1° before the plane enters the long glide. Even during the glide at an efficient speed close to the MMS, the AoA might be around 6° and the FPA around -2.4°, with a pitch of around 3.8°.

    Based on what we know, I think it is premature to assume a glide did not occur.

  600. Viking says:

    @Victor

    I am happy to hear that you do not exclude the jump scenario. In this context it is interesting that you point to the Nicobar islands as the most likely place(s). Initially I thought the same. However, I had problems making the fine details fit, and at the end there was only one of these islands which I could not exclude totally.

    For this particular island there was a funny coincidence that I noticed. At some time in 2017 one of the national German television channels (ZDF as far as I remember) sent one of their best reporters to that particular island to make a comprehensive report. His report was well made but quite uninteresting. At the time I had the suspicion that someone had the same idea as I and that was the true reason for sending the excellent journalist to an otherwise uninteresting and extremely remote place.

    It may of course also have been a completely random coincidence.

  601. Victor Iannello says:

    @Viking: Please be aware that when I say I don’t dismiss or exclude a scenario, it doesn’t mean I think it is likely. The coordination with collaborators that is required both at KLIA and at the jump site are obvious weaknesses of this scenario.

  602. Richard says:

    @Victor

    My point is that an abnormal aircraft attitude would not last long and with an IFE system message being repeated every 8 to 15 seconds until acknowledged, the argument based on a continuing aircraft attitude is not valid.

  603. Viking says:

    @Victor

    I would also like to add that I do not exclude the suicide scenario. However, my worry is that setting course directly for the South Pole excludes any other scenario. I would be happy to see more discussion of that particular (partly overlooked) detail.

  604. Victor Iannello says:

    @Viking: It is a common misconception that the use of the BEDAX and SouthPole waypoints is an assumption of the analysis. Rather, a path following those waypoints is derived from an analysis that attempts to use an unbiased objective function to discriminate among all paths.

  605. Viking says:

    @Victor

    The coordination is indeed tricky for the jump scenario. However, I am sure Shah had connections in many places after so many years as a pilot. The need for coordination is also a good explanation for leaked information (eg. via @TimR). For the suicide scenario I have severe problems understanding these leaks.

  606. Viking says:

    @Victor

    I agree that you have deduced the South Pole direction so the suicide is not by any means a direct assumption. The problem is that it becomes an indirect one if other assumptions lead to this and only this solution (S.P. course).

  607. sk999 says:

    Victor,

    OK, thanks. The spec for the MITEQ receiver says it has a “frequency range” of +/- 50 khz, so I guess that is enough to compensate for satellite oscillator drift over the life of the satellite.

  608. Niels says:

    @VictorI
    About the frequency calibration you wrote:
    “The frequency error is generated by the difference between the received and expected frequency of the P-channel signal.”
    Ok, thank you. Is it correct that the calibration is only carried out with the aircraft stationary on the ground? In which case the relative satellite motion may still impact the calibration.

  609. Victor Iannello says:

    @sk999: I’d be interested to see where you saw the “frequency range”.

    The MITEQ receiver has an “acquisition range” that is settable between +/- 1 kHz and +/- 55 kHz, which is the bandwidth for acquiring the pilot signal. The LTE can be set to within +/- 999.999 kHz, which I assume is also the range of LTE that can be automatically tracked and compensated.

  610. Victor Iannello says:

    @Niels. I believe so because it would make sense, but I’m not sure. In any event, there was no recalibration during the flight because 26 hours of on-time hadn’t elapsed since the last recalibration.

  611. sk999 says:

    Victor,

    Here’s the document I came up with after a brief search:

    https://nardamiteq.com/docs/D-169.PDF

    No idea if it is the actual model installed at Perth. “Frequency Range” is in the section “Acquisition/tracking characteristics”.

  612. David says:

    @Richard. Supporting what @Ventus and @Victor are saying (and I implied), a momentary uncovering of the APU DC pump inlet could flame out the APU once the ensuing air or vapour bubble reached it.
    There is of course the question of it being restarted, even manually, but that would have led to another LOR.

    @Andrew, ALSM. If a future search supposed that the priority was to search within the likely radius of an intent to continue rapid (as per the LOA) descent, the question is what would that be?

    One issue is the possibility of pitch-up when unbanked.

    In investigating the Silk Air 737 crash, Boeing concluded from simulations (ie presumably within the 737’s data base) that the aircraft would have pitched up at very high speed if full manual nose down trim had not been applied or alternatively the pilot applied a 50 lb nose down force to the control column. That was with both engines running so full flight control hydraulics.

    You will see where I am going with this. With APU running and an aircraft of a similar wing sweep and taper, would there be a similar nose up pitching moment is one question. A second is whether loss of the APU as above make a difference?

    I note that Boeing/ATSB made no mention of this as an outcome of 777 simulations, simply mentioning that in some the aircraft exceeded the data base. Implicitly, nothing similar could be expected within the 777 data base.

    Can you throw any light on this from your experience?

    Incidentally Andrew, having racked my memory I can add a better rationale for the APU having a more reliable start at altitude than the main engines. Combustion chamber/atomiser/injector etc design, two generations ago anyway, could be optimised for flame stability at altitude or economy. You will recall the smoke trails left by the early 707’s; and military aircraft, US in particular, where the trade off was more flame stability versus visibility.

    I suppose that the main engine trade off these days allocates a higher priority to pollution though in the APU’s case, noting its relatively low fuel consumption, that is not such an issue; so carbon content in its exhaust is more acceptable. Even so I have a report about Zurich airport and the monitoring of ground APU-running there from the pollution perspective.

  613. Victor Iannello says:

    @sk999: That’s what I thought. I don’t believe the acquisition range is the same as the LTE range, as I said above.

  614. Andrew says:

    @David
    @Richard

    I agree with the previous comments regarding the likelihood of an APU flameout. A small nose down change in the pitch attitude could easily uncover the pump inlet and disrupt the flow of fuel to the APU, given the small quantity of residual fuel in the tank. My earlier work/comments were not intended to imply the APU would keep running until all the residual fuel was exhausted.

    @David

    RE: ”With APU running and an aircraft of a similar wing sweep and taper, would there be a similar nose up pitching moment is one question. A second is whether loss of the APU as above make a difference?”

    1. I would think so. The FBW control laws would take care of any change in the aircraft’s longitudinal stability at high speed (eg Mach tuck, if any) and the aircraft would tend to return to its previously trimmed speed.

    2. I doubt it. The left elevator would remain powered by the C hydraulic system (via the RAT) and controlled by the C ACE. The FBW control laws should continue to maintain the aircraft’s stability, which would tend to pitch the aircraft nose-up if it were flying faster than its trimmed speed.

    RE: ”You will recall the smoke trails left by the early 707’s; and military aircraft, US in particular…”

    The older designs certainly produced a lot of smoke (and noise):
    https://i.pinimg.com/originals/36/e4/31/36e4314837fa2aafa4253a5a1ed4067f.jpg
    Fortunately, those days have long gone!

    RE: ”I suppose that the main engine trade off these days allocates a higher priority to pollution though in the APU’s case, noting its relatively low fuel consumption, that is not such an issue; so carbon content in its exhaust is more acceptable. Even so I have a report about Zurich airport and the monitoring of ground APU-running there from the pollution perspective.”

    Even so, airports around the world are becoming increasingly sensitive to pollution from APUs. At many airports the APU must be shut down after arrival and only re-started shortly before departure.

  615. David says:

    @Andrew. Thank you. In either case no pitch up in this aircraft with its FBW.
    Re your photo, at least that might show those behind where wake turbulence is!

  616. David says:

    @Andrew. I wracked my memory again for how to spell racked.

  617. Victor Iannello says:

    @David: In your scenario of large nose-down elevator input followed by release of the column would provide a large perturbation for an extended phugoid. I don’t believe there would be a sudden pitch up when the column is released.

  618. David says:

    @Victor. What I was wondering was whether there would be sufficient 777 pitch up moment to result in a pitch up. @Andrew has indicated that the moment would be countered by FBW so that would not arise, as I understand it.

  619. Victor Iannello says:

    @David: As far as I can tell, whether in NORMAL or SECONDARY control mode, with hands off the column and level wings, the vertical acceleration is limited to around +/- 0.3g and the pitch changes slowly unless in NORMAL mode and overspeed protection kicks in. If you have a specific case in mind, e.g., flight control mode, engine thrust, speed, etc., I can run it on the sim to demonstrate this.

  620. David says:

    @Victor. I may not be understanding you. The question I raised was based on whether the 737’s tendency to pitch up at transonic speeds without a substantial manual input to correct this would lead to a like pitch up tendency in the 777 such that MH370 would fly farther than in a straight line accelerating descent. I had it in mind that the person at the controls might not prevent that but am now reassured that the 777 fly-by-wire pitch stability would obviate it anyway.

    Hence there is no specific case I have in mind: it is general. To summarise, there is no risk from that hypothetical so, as I can see it, the issue is resolved.

  621. Victor Iannello says:

    @David: I thought your question referred to a nose-up pitch response. Mach tuck causes a downward pitch, doesn’t it?

    In any event, if your hypothetical doesn’t occur, the issue resolved, as you say.

  622. flatpack says:

    @Richard

    “Don and Andrew have shown that the IFE switch on the overhead panel does not effect the IFE comms, by inspecting the work instructions of the AD.”

    The switch does not affect power to the back end however if the cabin clients are not powered up when the server boots up, then boot time may be extended to allow more polling time. CPU cycles may be consumed generating an error dump etc.

    Has this possibility been considered?

  623. Richard says:

    @flatpack

    Don analysed the various instances where we have the data and timing for the IFE system message following a SDU reboot.

    There is quite a range of timing from 62 secs to 156 secs.

    There are obviously reasons for this wide range of timing.

    I took 3 minutes as a safe assumption.

    I can only hope my assumption covers the majority of possibilities.

    Since a long glide would take more than 10 minutes, there is a certain margin.

  624. TBill says:

    @Andrew
    APU questions-
    a) I assume APU can be manually turned on when engines are on
    b) Does the APU have a visible (flame) at night?
    c) What does flame out of APU look like at night?
    d) What would be eco-issues with APU? I know there can be around 2000 ppm Sulfur in there, unlike gaso and diesel the sulfur I think is still allowed in there to keep jet fuel cost as low as possible

    I never could find a picture of main engine flame out at night.

  625. airlandseaman says:

    Re BFO calibration and accuracy…The pilot receiver acquisition range has nothing to do with the accuracy of the BFO values. The accuracy is a function of the reference oscillator accuracy only. The GESs have a “station standard” (typically a rubidium reference) that is distributed to all CUs (demodulators). Thus, the error contributed by the pilot receiver and demod’s is <<1Hz. The dominant error in BFO observations is the small SDU OCXO drift. Everything else has been calibrated out very well, including the eclipse induced anomoly.

  626. Victor Iannello says:

    @airlandseaman said: The pilot receiver acquisition range has nothing to do with the accuracy of the BFO values.

    Nobody said this. @sk999 asked whether other components in the link could also recalibrate like the AES. We know the GES does not because it has an accurate reference that is used by all the GES components. My claim was that the satellite also does not undergo a recalibration because the long-term drift is compensated in the pilot receiver, which can accommodate a lot of long-term drift of the oscillator in the satellite.

    I suspect the fact that the BFO is not zero after a recalibration on the ground is partly due to uncompensated Doppler in the pilot receiver.

  627. airlandseaman says:

    Victor: I agree with all that. All the transponders are sync’ed to a common (free running) reference oscillator on the s/c, like the GES, but it is not a high accuracy reference like the GES uses. It does not need to be as accurate because, as you note, the pilot system compensates for the s/c reference oscillator error and drift over time.

  628. Viking says:

    @All

    More details on suicide and South Pole directed route:

    In case we accept suicide as the cause, it is still worth digging one level deeper into the details. From everything we know about MH370 it is clear that a lot of effort was put into hiding the route and end destination for the airplane. This must mean that we are dealing with a concealed suicide. In other words the perpetrator wanted us to believe that it was not suicide, but rather some kind of accident. While this is possible, there are a number of inconsistencies in the details. We have already discussed some of the leaked information. This is all quite strange, but unfortunately we have no way of knowing if large parts of it may be false or misleading.

    However, we do know one thing for certain. Immediately after the airplane left radar coverage the SDU was turned on again. This is extremely strange in case the perpetrator wanted to conceal a suicide. Actually it seems completely contradictory to all the rest of the events, since it proves that the airplane is still under intelligent control, and now able to communicate again.

    I think this is as close as we come to a proof that it was not an attempt to hide a suicide.

  629. Richard says:

    @Andrew

    You stated “A small nose down change in the pitch attitude could easily uncover the pump inlet and disrupt the flow of fuel to the APU, given the small quantity of residual fuel in the tank. My earlier work/comments were not intended to imply the APU would keep running until all the residual fuel was exhausted.”

    We are both agreed with the ATSB statement “the difference in location between the left engine fuel inlet and the APU fuel inlet would result in approximately 30 lb of fuel being available to the APU after a left engine fuel exhaustion”.

    In your paper, you state “If the left main tank is empty, the APU DC fuel pump cannot supply fuel to keep the APU fuel supply line pressurised. Continued operation of the APU would therefore rely on the boost pump in the APU fuel cluster to continue drawing the remaining fuel from the APU fuel supply line.”

    This statement in your paper appears to imply that the boost pump in the APU fuel cluster would ensure supply of the remaining fuel at least from the APU fuel line.

    Are you now saying that the boost pump in the APU fuel cluster would not ensure supply of the remaining fuel from the APU fuel line, if there was a nose down attitude?

    I understand “a small nose down change in the pitch attitude could easily uncover the pump inlet and disrupt the flow of fuel to the APU, given the small quantity of residual fuel in the tank” but what about the fuel already in the fuel line?

  630. Andrew says:

    @David
    @Victor

    David, I think you may have misunderstood my previous comments. The NTSB and Boeing conducted a number of simulations in Boeing’s M-cab simulator, in a bid to reproduce the descent profile of the Silkair B737. The NTSB’s findings in Appendix N of the Silkair accident report state:

    “The engineering simulations just discussed indicated that manual manipulation of the primary flight controls in multiple axes would result in a descent time history that was similar to the last recorded radar points. Without the use of horizontal stabilizer trim, this would require control column forces greater than 50 pounds and large control column inputs; if those forces were relaxed, the airplane would have initiated a return to a nose-up attitude due to its inherent stability. However, the simulations indicated that a combination of either control column inputs and/or changing the stabilizer trim from about 4.5 to 2.5 units nose-down trim (which would have “unloaded” the high control forces) in conjunction with aileron inputs, would result in a descent time history similar to that of the last recorded radar points.”

    I believe the B777 would behave similarly, with or without the APU operating.

  631. Andrew says:

    @TBill

    RE: “APU questions-
    a) I assume APU can be manually turned on when engines are on
    b) Does the APU have a visible (flame) at night?
    c) What does flame out of APU look like at night?
    d) What would be eco-issues with APU? I know there can be around 2000 ppm Sulfur in there, unlike gaso and diesel the sulfur I think is still allowed in there to keep jet fuel cost as low as possible”

    a. Yes.
    b. No.
    c. Flame-out simply means the flame inside the combustion chamber has extinguished, which results in an APU shutdown. It doesn’t mean there’s a visible flame out of the APU.
    d. Noise and exhaust emissions. The following study might be of interest:
    Aircraft APU Emissions at Zurich Airport

  632. Andrew says:

    @Richard

    RE: “Are you now saying that the boost pump in the APU fuel cluster would not ensure supply of the remaining fuel from the APU fuel line, if there was a nose down attitude?”

    The comment you quoted was not intended to imply the APU would continue running under suction feed. It was meant to highlight that the APU would need to use suction feed if it were to continue running long enough to allow the final SATCOM messages. However, I don’t believe that would have occurred, as was discussed in a previous post:

    “I doubt the APU would continue to run after the wing tank ran dry. I note ALSM’s comment regarding the placement of the engine pump and wing tanks in the light aircraft he has flown, but in the MH370 case there is a long fuel line and a considerable height difference between the tank and the APU. The auto start is also assumed to have occurred at very high altitude. @David previously suggested that vapour lock would prevent the APU suction feeding under those conditions and I agree. If the main tank boost pumps fail on one side, the associated engine won’t reliably run by suction feed above FL380. I therefore think it’s unlikely the APU would run at high altitude without boost pump pressure, especially given that it is mounted higher than the tank.”

  633. Victor Iannello says:

    @Andrew: When it is stated that if those forces were relaxed, the airplane would have initiated a return to a nose-up attitude due to its inherent stability, I interpret that to mean a phugoid would begin, with the nose slowly rising. A high pitch rate up would mean high angle of attack which means high wing loading which means high upward acceleration. I don’t see how any of this is possible with no pressure on the column. And I said in my comment to David, Mach tuck would LOWER the nose, not raise it, as the center of lift moves backwards.

  634. Victor Iannello says:

    @Andrew: The discussion about the fuel fed to the APU has encompassed many comments and I think the significance may be lost. Here’s what I believe we are saying:

    It’s possible and perhaps likely that the nose-down attitude suggested by the final BFO values would uncover the inlet to the APU fuel line, which would starve the APU as the pressure in the pipe is lost. That means the missing IFE log-on request could have been a consequence of the nose-down pitch, and not indicative of an impact that occurred before the IFE log-on sequence began. That means that a long glide is possible.

  635. Richard says:

    @Andrew

    Many thanks for the clarification!

    I had seen the post, but was quoting your paper.

    I had failed to understand that your post was a correction to your paper.

  636. Richard says:

    @Victor

    Many thanks for your summary of a long and complex discussion on a possible long glide scenario.

    I stand corrected!

    You left out a statement requiring an active pilot to enter a steep descent in order to match the BFO data, followed by the requirement of an active pilot to recover from a steep descent and control a glide at the right speed and descent rate.

    Can we assume an active pilot at the end of flight?

  637. Victor Iannello says:

    @Richard said: Can we assume an active pilot at the end of flight?

    I think it is a possibility. I’m not sure it is our base case. Let’s continue to discuss and think about it.

  638. airlandseaman says:

    Re: discussion about “…uncover the inlet to the APU fuel line [due to the downward pitch angle]…”

    …we do not know the attitude, velocity and acceleration profiles for the entire descent, but we have data suggesting that the average acceleration between 00:19:29 and 00:19:37 was about -0.68 Gs. Given that, is it not possible that the residual fuel was actually sloshing back toward the trailing edge, not collecting at the leading edge? Questions we may not ever answer without the FDR.

  639. Victor Iannello says:

    @airlandseaman: In your glider, if you push the nose down and enter into an accelerating steep descent, are you pushed forwards or backwards in your seat?

  640. airlandseaman says:

    Victor: I rarely do that, but your body would feel “weightlessness” when you move up in the straps and back in the seat. To be clear, I did not intend to infer precisely what was happening between 00:19:29 and 00:19:37. I was simply pointing out that the acceleration during that time suggests some dynamics going on over the descent, and fuel probably sloshing around in the tank. I doubt that it was collecting at the leading edge and staying there.

  641. Andrew says:

    @Victor

    RE: “When it is stated that if those forces were relaxed, the airplane would have initiated a return to a nose-up attitude due to its inherent stability, I interpret that to mean a phugoid would begin, with the nose slowly rising. A high pitch rate up would mean high angle of attack which means high wing loading which means high upward acceleration. I don’t see how any of this is possible with no pressure on the column.”

    I agree. I interpreted David’s comment “the aircraft would have pitched up at very high speed” to mean the aircraft would have pitched nose-up if the control forces were relaxed while the aircraft was flying at very high speed. I don’t think he meant the pitch rate would be high. @David??

    RE: “And I said in my comment to David, Mach tuck would LOWER the nose, not raise it, as the center of lift moves backwards.”

    Yes. I said to David “The FBW control laws would take care of any change in the aircraft’s longitudinal stability at high speed (eg Mach tuck, if any) and the aircraft would tend to return to its previously trimmed speed.” In other words, the FBW would prevent Mach tuck and the aircraft would therefore behave in the conventional sense.

  642. Victor Iannello says:

    @airlandseaman: I’m not criticizing you. I’m also thinking about these examples before I draw the gravitational and acceleration vectors and show mathematically what occurs.

    Yes, you would be lighter in your seat, but also the component of gravity along the axis of the plane would push you forward.

    Here’s another thought experiment. Suppose we place a ball in the aisle of a B777 and pitch forward into an accelerating descent. Does the ball go forward or backward?

  643. airlandseaman says:

    It depends on the acceleration magnitude…which is why I made the point that I was not referring to the .68G time frame, except to suggest that the accelerations could have been quite variable and over -1 G at times.

  644. Victor Iannello says:

    There’s another interesting aspect of allowing pilot inputs. When the fuel starts running low, the crossfeed fuel valves can be opened, which means both engines will flameout at nearly the same time, and the fuel range will be maximized. That means the double flameout would have occurred around 00:17.

  645. David says:

    @Andrew, Victor. My interpretation of the NTSBs passage about Boeing Silk Air 737 simulations that @Andrew quotes remains that the 737 would pitch up as it reached high transonic speeds were there no pilot intervention, ie either full manual down trim (as apparently was applied in the Silk Air case) or, had that not been done, a 50 lbf nose down force by him on the control column.

    Mach tuck in that report is described as mild, occurs at high normal cruise speeds, is controlled automatically normally and, were it not, it is easy for the pilot to control anyway.

    High transonic pitch up results I believe from loss of lift behind “normal”(ie perpendicular to the airflow) shocks where the airflow is sped up to sonic locally by lift there, ie mainly on the wing upper surface. That shifts the wing centre of lift forward and hence the pitch up moment, or tendency. As I recall this was a problem in some USAF 100 series fighters, uncontrolled pitch up approaching the sound barrier resulting in some cases in loss of the aircraft.

    In that quoted description of the Silk Air 737 descent, had full manual nose down trim not been applied it was implied (to me, but see below) the pilot efforts at controlling the consequent pitch up moment might well have been unsuccessful due to the continuing high forces required. The context of that passage was to indicate that unless full manual trim nose down had been applied the aircraft would have pulled out despite pilot efforts, possibly breaking up in mid-air due to wing loading (as distinct from speed related flutter, that did happen).
    Since it did not break up in mid air and the impact site was consistent with the descent plotted by radar, the NTSB used this in its assertion that the crash was deliberate. The Indonesians denied that and have since.
    What the NTSB was saying was that the full down trim found post crash was not the result of forwards inertia at impact but would have been applied deliberately and disclosed pilot intent.

    Now while your answer to my question 1, “..would there be a similar nose up pitching moment” in the 777 you said, “I would think so” but that FBW stability would intervene. I took that to mean that FBW stability differentiated the 777 from the 737. That must be where I misunderstood you.

    Your most recent, “I believe the B777 would behave similarly, with or without the APU operating” renews my interest in whether the intent I draw from the LOA descent rate and acceleration, of holding the nose down to impact, would not be realised if the 777 reached high transonic, it “pulling out” despite pilot efforts to counter that (assuming he did not trim nose down) and indeed the aircraft could have broken up mid-air in the process.

    Possibly I have misinterpreted what the NTSB meant by, “..the airplane would have initiated a return to a nose-up attitude due to its inherent stability.” as you both imply and that is why I spell out at such length above what I have understood to be its context.

  646. Ventus45 says:

    @Andrew

    A question.
    If we assume APU flameout during descent, due to fuel slosh / momentary starvation (perhaps for some seconds, even tens of seconds), but with fuel still theoretically available, will the APU continue to attempt to restart automatically, or not ?
    In other words, is it plausible, that in a phugoid descent, with the pump inlets alternating between being “dry” and being “wet”, might the APU attempt to restart itself again, or is the autostart function a “one shot deal” ?

  647. Andrew says:

    @Richard

    RE: “I had failed to understand that your post was a correction to your paper.”

    For what it’s worth, it wasn’t so much a correction as an ‘addendum’! I made the original comment in relation to the Case 2 scenario discussed in the paper, where the engine relit and accelerated slowly to an intermediate thrust setting before flaming out again due to fuel exhaustion. The analysis showed there would be sufficient fuel to also allow the APU to autostart, but that all the fuel in the tank would be exhausted before the SDU rebooted and made the final transmissions. The original comment was meant to highlight that the Case 2 scenario would require the APU to rely on suction feed, using fuel in the APU fuel line, if it were to continue running long enough for the final SATCOM transmissions to take place. The paper failed to reiterate that suction feed was unlikely to occur.

    Note: The analysis was based solely on the estimated residual fuel quantity and did not consider the effects of sloshing fuel in the tank!

  648. Andrew says:

    @Victor

    RE: “It’s possible and perhaps likely that the nose-down attitude suggested by the final BFO values would uncover the inlet to the APU fuel line, which would starve the APU as the pressure in the pipe is lost. That means the missing IFE log-on request could have been a consequence of the nose-down pitch, and not indicative of an impact that occurred before the IFE log-on sequence began. That means that a long glide is possible.”

    I agree.

  649. Andrew says:

    @David

    RE: “High transonic pitch up results I believe from loss of lift behind “normal”(ie perpendicular to the airflow) shocks where the airflow is sped up to sonic locally by lift there, ie mainly on the wing upper surface. That shifts the wing centre of lift forward and hence the pitch up moment, or tendency.”

    Transonic pitch-up occurs primarily due to the tip-stalling tendency of swept-wing aircraft, even at relatively low angles of attack. When the wing tips stall, the centre of pressure (CP) moves forward very quickly, which can cause a violent pitch-up. Mach tuck, on the other hand, is caused by the rearward movement of the CP as the normal shock wave moves rearward at higher transonic speeds.

    RE: “The context of that passage was to indicate that unless full manual trim nose down had been applied the aircraft would have pulled out despite pilot efforts, possibly breaking up in mid-air due to wing loading…”

    The report only says “if those forces were relaxed, the airplane would have initiated a return to a nose- up attitude due to its inherent stability.” It does not say there would be a violent pitch-up that would cause the aircraft to break up.

    RE: “Your most recent, “I believe the B777 would behave similarly, with or without the APU operating” renews my interest in whether the intent I draw from the LOA descent rate and acceleration, of holding the nose down to impact, would not be realised if the 777 reached high transonic, it “pulling out” despite pilot efforts to counter that (assuming he did not trim nose down) and indeed the aircraft could have broken up mid-air in the process.”

    I’d be very surprised if the pilot didn’t trim if the intention was to cause a high speed impact. If trim wasn’t applied and the pilot relaxed the control forces (or couldn’t overcome the natural pitch-up tendency), I doubt any subsequent pitch-up would cause the aircraft to break up.

  650. Andrew says:

    @ventus 45

    RE: ‘In other words, is it plausible, that in a phugoid descent, with the pump inlets alternating between being “dry” and being “wet”, might the APU attempt to restart itself again, or is the autostart function a “one shot deal” ?’

    As I said in an earlier reply to @David:

    The [APU] autostart is initiated by a signal from ELMS to the APUC upon loss of power to both transfer buses. The manuals do not describe what happens if the autostart fails, but I suspect there would only be one attempt. If that failed, the crew would be alerted and would then need to attempt a manual start.

    As far as I’m aware, it’s a ‘one shot deal’.

  651. David says:

    @Andrew. My reading of the report is that as speed increases so does the nose up pitch moment. Once it has increased to 50 lbf down force (more for the 777?)being required, were that released slowly I think pitch up rate would accelerate. If suddenly (in the 777 the pilot was overwhelmed say) then I think it could be violent.

    My assumption was that the pilot may not know the extent of the forces building and if he did his mind might be distracted anyway, focussed elsewhere.

    However since apparently the 737 pilot was well aware of that in advance while quite possibly in a similar state of mind, that supports the likelihood that the 777 man would apply trim like wise.

    That now, is a good reason to put this pitch up possibility aside again, and noting that Boeing/ATSB made no mention of it also suggests it has no likelihood also.

    Thank you for your further remarks.

  652. Ventus45 says:

    @ Andrew

    Thanks, I just wanted to explore whether an APU will auto relight just like some main engines will, i.e. to consider what would (could / might) happen in the APU’s own controller logic, if it flamed out without an “external command” to shut down, i.e. being switched off.

  653. David says:

    @Victor. You wrote, “There’s another interesting aspect of allowing pilot inputs. When the fuel starts running low, the crossfeed fuel valves can be opened, which means both engines will flameout at nearly the same time, and the fuel range will be maximized.”

    The crossfeed valves being open, piloted or not, is interesting in another respect I have pondered about. The APU DC pump could deliver relight fuel to both engines.

  654. TBill says:

    @Victor
    Interesting if not exciting.
    Are you saying the tank equalization can only be done at low fuel level? So therefore, if true that could be active pilot?

    I think you are saying 00:17 because that is 2-min before 00:19 Arc7. So you are saying APU came on long enough to reboot SATCOM but short enough to not logon IFE.

    That implies if we say straight flight over Arc7 at constant speed, I get about 435 ground speed for 180 South path. This could be refined, but I wonder if that is a little slow? I wonder if that implies moving closer to 30 South is more consistent. Admittedly I am showing my bias but I am trying to say, what can we deduce? Of course once we say active then that leads to possible end-flight maneuvers.

  655. TBill says:

    @Victor
    PS- on above calcs, I am just looking at Arc6 to Arc7 distance.

  656. Victor Iannello says:

    @TBill: I said open the crossfeed valves when the fuel level is low so that each engine would draw from its respective tank and therefore there would be little weight unbalance. The valves in fact could be opened at any time, but the draw from the left and right tanks might be very uneven.

    I am suggesting the possibility that both engines flamed out at 00:17, the APU come online a minute later, and the SATCOM logged in a minute later during an accelerating steep descent. (Of course, it might have taken longer for the SATCOM to log-on, which means the engines flamed out earlier.) The dynamics of that descent might have uncovered the APU fuel line inlet and caused the APU to fail with a little fuel still in the tank, which cut the power to the SATCOM. It’s possible there was a recovery and glide after that.

    Here’s an interesting possibility: Let’s suppose the pilot intended to perform a powered, flaps-down ditching. Unfortunately, he started his descent late, ran out of fuel while still at relatively high altitude, and he couldn’t lower the flaps, which require an altitude less than 20,000 ft. When the APU started, he eventually realized that this might provide the opportunity to lower the flaps, but he also knew that he had limited time before the APU ran dry. He descended quickly to get below 20,000 ft while the APU was still running, but the descent uncovered the APU fuel inlet, and the APU failed before the flaps were lowered.

  657. TBill says:

    @Victor
    Another way to put that I think is that we may start to have water-impact consistent flaperon explanation. I found myself re-reading Langewiesche to see how he handled the flaperon issue, and I think he ducked (was silent). In general when he came a fork in the road, he took it, which has story value it seems.

  658. Viking says:

    @All

    Concerning: More arguments against South Pole route.

    Until today I have discussed two arguments against routes towards the South Pole:
    (1) The BFO data is not sufficiently strong from a data statistical point of view to determine the best solution among southern routes.
    (2) A route towards the South Pole is only likely if the reason was suicide. All known pilot suicides in large passenger jets were not particularly well hidden, and they happened within a few minutes. MH370 does not fit this picture. In a broader perspective (not limited to airplanes, but including other types of murder/suicide cases) there are a few cases of well hidden suicides. Also this type does not fit to MH370, in particular because of the rebooting of the SDU.

    In other words I think we have to come up with another type of suicide if the South Pole route is to be our best proposal.

  659. Richard says:

    @Viking

    May I ask if you have read the paper available above and followed the discussion with meanwhile 660 comments?

  660. Viking says:

    @All

    Concerning: Biofouling on debris speaks against most southern routes.

    If the airplane crashed somewhere near 34.4S in late summer, it would initially get mostly sub-tropical (mixed with some temperate) climate biofouling. As the currents predominantly move debris from this area towards northeast (later north), this movement combined with onset of autumn (later winter) will secure that temperatures remain almost constant for a long time. After 6-7 months with a modest fall they will start rising to values typical around Reunion.

    I am sure this will allow some of the original subtropical biofouling to settle well and remain on the debris for a long time. Replacing this with purely tropical biofouling is not likely to have completed when the flaperon beached.

    However, it was found with only one species of tropical biofouling. Finding only one species is unusual in its own, and in addition a detailed chemical analysis pointed to a rather unusual temperature history, starting at extreme tropical values (well above those near Reunion), then going down to sub-tropical, and finally approaching values near Reunion.

    This is simply totally inconsistent with expectations!

  661. Viking says:

    @Richard

    Unfortunately only DennisW seems to agree with the BFO data issue. Concerning the suicide issue there are somewhat mixed opinions.

    However, I have not seen any arguments which can seriously sweep my two initial points off the table.

    Please note that I am not saying the mathematics and logic method behind your analysis is wrong. On these points I agree with practically all your arguments, and I really think you did a good job. What I claim is that there are some questionable hidden assumptions behind. These need very careful consideration and discussion before we can draw any final conclusions.

  662. Viking says:

    @Richard

    Sorry, I can see our contributions crossed. I hope it does not cause confusion that I started with the next issue.

  663. Richard says:

    @Viking

    Was that a yes or a no?

  664. Victor Iannello says:

    @Viking: Some comments about your points:

    1) The BFO is not being used in isolation to discriminate among paths to the SIO. In fact, Bobby thinks the path BEDAX-SouthPole might be the “best” path using criteria that does not include BFO statistics. Before rejecting the path on this basis, I suggest you wait until the analysis is presented.

    2) Suicidal pilots have in the past have tried to misdirect investigators, including turning off the FDR and CVR to make it difficult to determine the cause. What’s never occurred before was a misdirection in the path as well as misdirection in the cause. That certainly is not a reason to dismiss the scenarios that include pilot suicide. We all acknowledge that events occurred in this incident that have never been seen before.

    3) The marine biology results are at best inconclusive. If the plane crashed along the 7th arc near 34S, the debris traveled north and eventually to the west. Young, tropical marine organisms were found on the debris. That doesn’t seem very surprising to me. That said, I’m not a marine biologist. In fact, I wish a true marine biologist, i.e., somebody that does this for a living, would weigh-in to definitively accept or dismiss the possibility of a crash in southern latitudes.

  665. Viking says:

    @Richard

    That was mostly a yes.

    I skipped a few subsections where I could see that I would agree at least 99%, or where I knew I had no competence (sorry for that, but I wanted so save time), but I presume that is not a problem?

  666. Viking says:

    @Victor

    I would also like to discuss the biofouling with an expert.

    My knowledge on the subject is partly of general nature (other biosystems) and partly based on results from archeology. A nice example of the resilience of marine life with good adhesion is that certain clamps never managed to cross the Atlantic Ocean for millions of years, but they made it under viking ships 1000 years ago. This is one of many proofs that vikings made it to USA from Europe (and back) long before others.

    In that case the temperature differences are actually greater (15 degrees or more).

  667. Richard says:

    @Viking

    Dear Professor Martin Kristensen, many thanks for admitting you are short on time and competence.

    I have also run out of time for someone, who quotes such experts as Kate Tee, Mike Chillit and Jeff Wise and cherry picks the satellite data with such a convoluted mathematical approach, camouflaging a parachuting from a Boeing 777.

  668. Sid Bennett says:

    RE:BFO

    I have consistently used both BFO and BTO in path studies and a joint minimum error can be found for a plausible path(s). The BFO data including that at 00:11 is consistent with the hypothesis that there was a 6nmi north offset from 18:22+ to 18:40 and the value for the first phone call is either a turn to BEDAX or an incomplete turn to 186T (ISBIX?).

    For the time period from 18:22 + a minute or so, until 00:11, the Satcom equipment appears to have been powered on continuously. Even if there is a fixed jump in average frequency at turn on, the stability of the crystal oscillator in the aircraft when at constant temperature is known to be excellent. So, at best, a simple change in the BFO fixed offset would suffice to correct all of the data. I believe that it is not quite so simple, but it may be good enough for the period of time in question, but do note use exactly 150Hz.

  669. Niels says:

    @Sid Bennett

    A constant (changed) bias frequency for the interval 19:41 – 00:19 is what I employed in my recent paper, please see (and link therein):

    https://mh370.radiantphysics.com/2019/05/31/a-new-methodology-to-determine-mh370s-path/#comment-23535

    It gives some interesting results, which to me indeed seem to suggest that frequency stability was good.

    The paper will be updated soon, with a “Case C”. All key results are indicative for a crash latitude in the S33 -S35 range, and consistent with constant true track, LRC near FL340 – 350.

  670. Viking says:

    @Sid Bennet, @Niels

    Your approach looks interesting, and qualitatively comes much closer to the behavior one might expect from the oscillator after a rough cycling in temperature and pressure. It tends to get stuck at some offset, which gradually decays over a long, and not particularly well defined time (typically following something like a stretched exponential curve). My guess is that for the period in question it will work fine. As a side result it will also be able to include effects from an open door (but with a somewhat larger and different offset).

    This is exactly what what I have been arguing on concerning the figures in the Ashton paper for a long time.

    If one plots measurements performed at fixed time intervals on such a system over a very long period of time (in this case probably several days or even weeks) the data statistics will show a tail heavy distribution. One will get the same qualitative distribution (but with a smaller spread) without the cycling as DennisW correctly points out.

    A real offset (such as flying permanently with an open door or an incorrect temperature on the stabilized furnace) will behave qualitatively different for long times (no decay), but for shorter times it will look very similar.

    @Richard

    Using this kind of knowledge in the analysis is not cherry picking of the data. However, with so few data points as we have available, one must be very careful. Probably I was not sufficiently careful on this point in my initial analysis, but in order to limit any damage I gave the BFO data very low weight.

  671. Viking says:

    @Niels

    I am looking forward so your Case C (and perhaps later D).

    Have you tried to look for solutions with a U-turn (eg. soon after 19.41)? It may not automatically be found with standard polynominal fitting, since it may be necessary to insert a sign change somewhere in the procedure.

  672. Viking says:

    @Richard

    I had very good reasons to use Topology Optimization as my mathematical procedure. One reason is that I had written and documented the validity of the necessary software and procedures years ago. Another reason is that I knew my particular version is able to handle rough terrain (including things like a sign change).

    This is exactly the opposite of hiding something. Rather it is able to find the nasty needle in the haystack!

  673. David says:

    737 Max saga followers. From the below, “Air Canada is the only North American carrier with a MAX full-flight simulator. This could influence the return-to-service timeline if regulators mandate additional simulator training as a condition of operating the MAX again. ”

    I though the Max needed no special training? If so why would there be one and why just Canada?

  674. David says:

    ‘thought’ vs ‘though’

  675. Don Thompson says:

    @David,

    It was intended that the 737 MAX would not require specific procedures training, it was to be regarded as sufficiently common with the NG series that there was no specific requirement for a FFS replicating a MAX flight deck. While the cockpit displays systems are considerably different, the information that is displayed was deemed to be common. As Boeing had underplayed the impact of MCAS, it was not deemed necessary to be part of the training curriculum: why would it? MCAS was not even described in the FCOM.

    As airlines transitioned their fleets to MAX it would be normal for the airline’s simulator fleet to reflect the mix of variants. My experience was that a rule of thumb was 25-30 aircraft to 1 simulator for an aircraft such as the 737. A range of factors will drive the ratio.

    It remains to be seen what conditions will permit the MAX to fly again, modification may be required to the simulators, just as the aircraft will undoubtedly require mods. It has been reported that Boeing made assurances that the MAX would not require specific training in FFS.

    Air Canada, and other training providers who had made early acquisitions of MAX simulators, will be well placed to move quickly to meet whatever requirements are set out for return to service. Flight Global publishes an annual census of simulators, you may find it a handy reference.

  676. David says:

    @Don Thompson. Thanks. So as airlines 737 numbers increase, understandably they need more simulators and added Maxes are best served by adding Max simulators, albeit other 737 simulators will do if you need no more in gross.

    Looking at a Flight Global summary I see that CAE of Canada has over half the world market so I suppose that would be a good reason why Air Canada apparently was first (in North America that is: I see that there are others).
    The Chinese had one (‘Tru’ manufacture (US based)) qualified as Level D in October 2018 and Gatwick had two of Tru’s qualified to Level D last year also, the earlier in August.

    I noticed on a quick web scan that there is interest in developing simulators for accident investigation!
    Who knows how they could be adapted to suit different types/functions but maybe their role would be limited. Perhaps these would be available for all national aircraft accident investigative agencies to hire, since I daresay they would be expensive. Perhaps ICAO is involved somewhere.
    They might be envisaged for when the aircraft manufacturer’s do not suit, or there is none for the aircraft type, or none suited to the research specifics in question.

  677. Andrew says:

    @David

    RE: “I noticed on a quick web scan that there is interest in developing simulators for accident investigation!”

    The Use of Full Flight Simulators for Accident Investigation

  678. Don Thompson says:

    @David

    CAE has dominated the civil, and possibly military, simulator business for many years. I worked with CAE’s teams many times, I was always impressed by their capability. L3-Link remains a competitor having evolved out of the legacy of TTS/Rediffusion/Link-Miles, companies that once, collectively, had a much greater market share. I’ve seen TRU mentions popping up more frequently.

    It’s my understanding that a flight test data pack, supplied by the a/c manufacturer, is now a significant part of the simulator cost. It’s not a volume business, long product delivery cycles, not so attractive to startups.

  679. TBill says:

    @Victor
    I know your scenario (failed-flaps down landing) was a brainstorm, but why do a flaps down landing, and if so, why mess it up so bad? Why delay steep descent until SATCOM reboot, and then we also have the ELT signal expected. How about cabin pressurization rationale for descent, assuming possible need to move about the aircraft or MEC Bay?

    In any case, I feel like fuel equalization potentially makes Arc6-to-Arc7 more constant speed and that makes Arc7 more important data point.

  680. Victor Iannello says:

    @TBill: I’ll answer your questions again recognizing that I am providing food for thought. I don’t necessarily prescribe to this scenario.

    1) “Why do a flaps down landing?” Flaps-down reduces the stall speed, which means less damage when the aircraft contacts the water. However, with no thrust available, the increased drag also means the descent rate increases, so the descent rate has to be properly managed in the flare. Complicating this is the lack of a good visual reference, but the radio altimeter (RA) would help.

    2) “Why mess it up so bad?” I think you are asking how the plane was allowed to run out of fuel before reaching an altitude where flaps could be deployed. I don’t know. Perhaps there were conflicting goals of maximizing range (i.e., stay at high altitude as long as possible) and ditching safely (i.e., deploy flaps at an altitude below 20,000 ft).

    3) “Why delay steep descent until SATCOM reboot?” Perhaps the timing of the steep descent was related to the APU coming up, and the realization that a flaps-down configuration was possible if below 20,000 ft. The SATCOM reboot was a consequence of the APU start-up, and may or may not have been noticed.

    4) “How about cabin pressurization rationale for descent”? When bleed air from the engines is lost, the outlet valves would automatically close, and the depressurization would slowly occur. Perhaps the rising cabin pressure altitude did prompt a rapid descent.

  681. Richard says:

    @Viking

    (1) What I mean by cherry picking is:

    (a) Discarding the satellite data at 18:25 UTC, quoting the Inmarsat Paper. Whereas the Inmarsat Paper states that corrected values my be derived.

    (b) Excluding the satellite data at 19:41 UTC, because you claim MH370 was heading North West, just prior to performing a U-Turn at that time, and not heading South East.

    (c) Ignoring the change in the satellite height, because it didn’t fit your model.

    (d) Guessing at the AES compensation algorithm, but ignoring that the AES compensation algorithm is known.

    2. What I mean by a convoluted mathematical approach is, initially assuming a flat earth, then updating to a spherical earth, then updating to an oblate sphere.

    3. I have run your flight path to Christmas Island and beyond though my model. I have included the U-Turn at 19:45:20 UTC on an initial bearing toward your point at 20:41:05 UTC (roughly toward Perth YPPH). I have included the change in track at 20:41:05 UTC or “bend” of 6.3° on an initial bearing toward your point at 21:41:27 UTC (roughly toward Christmas Island YPXM) and change of speed to fit your point at 00:11:00 UTC. I assumed an altitude of 11 km, LNAV navigation mode and a variable Mach speed.

    The BTORs are fine only when the latitude and longitude are forced to your points, which implies your LNAV navigation mode is out and/or there are speed changes in each leg. Your mean BTOR is -2.3 µs and the standard deviation is 4.6 µs, which is too good to be true. Without forcing the latitude and longitude at each handshake, the mean BTOR is -6.4 µs and the standard deviation is 137.8 µs.

    The BFORs are way off, the standard deviation is 39.64 Hz.

    The PDA is -0.16%, compared with a nominal -1.5%, which implies there was 218 kg fuel remaining at a MEFE at 00:17:30 UTC. The bleed air system would have to have been switched off for 74 minutes, to save this much fuel.

    In addition, your end point does not fit the drift analysis, as shown previously.

    https://www.dropbox.com/s/gm4sc2elblt74eq/MH370%20Flight%20Path%20Model%20V19.8%20RG%20LNAV%20VM%20FL367.6%20MK%20Full%20Report.png?dl=0

  682. George Tilton says:

    @Victor

    Curious….What is the One-Engine Inoperative Cruise Speed (ETOPS) for a B777 and at what Flight Level will the crew fly the plane?

    Thx in advance.

  683. David says:

    @Andrew. The cricket intervened.

    Thank you for that interesting reference. His interest is principally in the FSS/pilot interface, warning as to their use in replicating pilot reactions. To me he has unrealistic expectations as to the FSS improvement potential when given the variety of aircraft configurations (hardware and software) and limitations of data outside the flight envelope. For example he notes that they do not replicate flutter, while implying that pilots should not encounter situations they have not been exposed to in an FSS. That seems incompatible with flight trial limitations.

    What work is now envisaged to develop simulation specifically for accident investigation, as distinct from pilot training, remains unclear. Perhaps in engineering simulation the requirement for flight trial verification could be waived, Boeings assessment of 737 elevator flutter (for the Silk Air investigation) being just an example of such assessments that could be added into the data base, albeit with lowered proof of fidelity.

    On a lighter note I wonder as to weaknesses in the fidelity of the simulation now used in cricket to confirm or deny whether the bowled ball would have hit the wicket (the batsman then being ‘out’) had it not struck the batsman’s pads first. Its fidelity and authority seem unquestioned.

    But could the Poms nobble it?

  684. David says:

    @Don Thompson. “It’s my understanding that a flight test data pack, supplied by the a/c manufacturer, is now a significant part of the simulator cost.”

    Then there are the updates. Outside my data base here but while @Andrew’s reference touches on formal ‘qualification’ requirements it must be difficult and expensive to see that hardware and software changes are incorporated, configuration then being monitored formally, so maintaining the FSS integrity even with adjustments in configuration to suit the aircraft model simulated.

    All this with FSS having evolved from an assistant in training to an examination tool.

  685. David says:

    @Victor. Complementing your answer to TBill’s 3) “Why delay steep descent until SATCOM reboot?” Surprise at the roll induced by RAT deployment, that continuing to APU generator loading, may have contributed?

  686. Viking says:

    @Richard

    You are bringing up many issues. I have answers for them all, but it would take up too much space to answer them all in one Blog contribution. Therefore I will answer them in smaller groups. Here comes the first bunch:

    (a) In the optimization I ignored all points before 19:41. For anything before (and including) 18:25 (and the first subpoint at 18:28), I was simply worried that the uncertainty was too great since the oscillator was most likely still drifting significantly. It is correct that a major error contribution is coming from a simple digital effect which can be handled as described by Ashton. I did in fact use this information during the reconstruction of the flight between MEKAR and the position at 19:41. My quoting of Ashton is therefore referring to this general issue, and it is wrong that I discard the data. However, I would still insist that it should not be included in the optimization, since there is still some drift, and almost certainly still ongoing course changes.

    (b) Concerning the point at 19:41, I was particularly cautious in the analysis. As you can see from my paper, I applied several different methods to analyze the situation. From this I concluded that the most probable solution is that there was a U-turn immediately after 19:41. However, I have never excluded the other ‘branch’ which results in a solution very close to the one found by Ashton (and not significantly different from the solution that I presume you prefer). In case of a U-turn one must exclude the 19:41 point from the optimization. However, it is still included in the reconstruction of the rest of the flight after MEKAR.

    (c) It is clear from the numbers in table 9 that the Ashton paper ignored the satellite height contribution to the Doppler shift. I was in fact very reluctant to do the same (as you can read in my paper), but its inclusion made both ‘branches’ fit significantly worse. This gave me the impression that it was somehow included in the INMARSAT numbers already. This issue (together with others we (partly) discussed before) was my reason to put very low weight on the BFO data.

  687. TBill says:

    @Victor
    Dual flameout worst case- maybe 80nm glide, not necessarily straight away from Arc7, so maybe +60 nm outisde Arc7 search zone. Inside Arc7 not impossible. Not good, but better than my worst case scenario 300-nm glide with some fuel left.

  688. Andrew says:

    @David

    In my view, Tydeman’s paper argues that investigators need to carefully consider the limitations of simulators before drawing any conclusions from an accident simulation. That seems reasonable to me.

    Unfortunately, I didn’t get to see the cricket.

  689. David says:

    @TBill. An observation. For a glide it could be that his altitude at LOR, 00:19:29 was much lower than cruise altitude, the result of his dropping the nose earlier; at, or possibly before, MEFE. His descent rate at LOR on a southerly course could have been up 13,600 fpm while remaining consistent still with that BFO.

    He might have been able to recover some increased kinetic energy, maybe not (e.g. spoilers deployed, APU inlet drag while running).

    I realise that you are looking at the likely maximum range but achieving that depends on assumptions as to his aim.

  690. David says:

    @Andrew. “In my view, Tydeman’s paper argues that investigators need to carefully consider the limitations of simulators before drawing any conclusions from an accident simulation. That seems reasonable to me.”
    I agree.

  691. Richard says:

    @Viking

    My apologies, the statement on PDA should read, a nominal +1.5%, not a nominal -1.5%. I got the sign wrong:

    “The PDA is -0.16%, compared with a nominal +1.5%, which implies there was 304 kg too little fuel to reach a MEFE at 00:17:30 UTC. The Right Engine flamed out at 00:07:42 UTC and the Left Engine flamed out at 00:12:43 UTC”.

  692. TBill says:

    @David
    Re:
    I did a quick simulation starting at FL250 and cutting power at 00:17 and let the aircraft slow down until 00:19 and then a dive to approx. FL170 and then glide. I could have dived to lower altitude but I assumed the pilot was either (1) trying to lower flaps, or trying to pressurize aircraft to comfortable range possible with O2 bottle so as (2) to maintain some altitude for glide.

    Nornally we say could glide over 100-nm if you were at FL350. Of course possibly max glide was not the goal, could have gone in water earlier at higher speed. Perhaps probablilites could be developed.

  693. TBill says:

    @David
    PS- I set tailwind of 30 knots from the west and tried to glide with the wind perpendicular away from Arc7 using an old 180 South CTH case.

  694. Sid Bennett says:

    Forgive my laziness in searching the blog:

    What is the issue regarding the effect of satellite height on the Doppler shift?

  695. Richard says:

    @ Sid

    No need to search the blog! My comment was only yesterday.

    I did not say “satellite height”. I stated “change in satellite height”.

    I did not say “Doppler shift”. I stated “Ignoring the change in the satellite height, because it didn’t fit @Viking’s model.” Both BTO and BFO.

    BTO is dependent on distance (which can change if the satellite height changes).

    BFO is dependent on velocity aka change in distance (which can change if the rate of change of satellite height changes).

  696. Victor Iannello says:

    @Viking said: It is clear from the numbers in table 9 that the Ashton paper ignored the satellite height contribution to the Doppler shift.

    I’m not sure what you mean. Inmarsat does not ignore any component of the satellite’s velocity vector when computing the C-band and L-band Doppler shifts. They do ignore the vertical component of the aircraft’s velocity vector, i.e., the rate of climb (or descent). We know the BFO is very sensitive to vertical velocity because it is completely uncompensated by the AES. For level flight, that simplification is justified. During the initial climb and during the final descent, it is not.

  697. Sid Bennett says:

    Earlier this week I undertook to collect some information on BFO offset and while I have not found all of it, the following link should be useful.

    https://www.dropbox.com/preview/Public/BFO%20documents.pdf?role=personal

  698. DrB says:

    @Sid Bennett,

    Viewing your Dropbox requires me to sign in to Dropbox and also gives Google access to all my contacts (which I won’t do). I suggest you provide your file in a different manner.

  699. DennisW says:

    @DrB

    +1

  700. Sid Bennett says:

    I understand. Can you suggest another service I can use?

  701. Victor Iannello says:

    @Sid Bennett: Dropbox is fine. If you are using a Windows operating system, right click on the filename in your Dropbox folder, and select “Copy Dropbox link”. Then paste that link here.

  702. Andrew says:

    @Sid Bennett

    RE: “If you are using a Windows operating system, right click on the filename in your Dropbox folder, and select “Copy Dropbox link”. Then paste that link here.”

    The process is the same in OS X.

  703. Andrew says:

    @George Tilton

    RE: “Curious….What is the One-Engine Inoperative Cruise Speed (ETOPS) for a B777 and at what Flight Level will the crew fly the plane?

    That decision depends on a number of factors, including weight, temperature, terrain and distance to the nearest suitable airport. If range is an issue, the crew would normally fly at the Engine Inoperative Long Range Cruise speed (EO LRC), typically about M0.65 and around FL210-FL240 at medium weights. The crew might fly faster (and lower) if range is not an issue.

  704. Viking says:

    @Victor

    When I recalculated the numbers in table 9 long time ago, I found that INMARSAT had also ignored the vertical component of the satellite velocity.

  705. Viking says:

    @Richard

    I suspected that sign error, but I am extremely glad you found it yourself. It is interesting that the flameout of the right engine calculated by your model fits exactly with the contrail. By reading off from the weather satellite photo one I got 00:08, just 18 seconds later.

  706. Viking says:

    @Sid

    The issue on the satellite height is that INMARSAT ignored the first term in my formula (5) (DeltaH/Deltat) in their table 9.

  707. Sid Bennett says:

    Well, I can use the link that I posted and get the document.
    If anyone wants it sent another way, please suggest it.

  708. Sid Bennett says:

    @Viking
    What is the magnitude of the error?

  709. Richard says:

    @Viking

    You are missing the point. The PDA for your path does not fit a MEFE at 00:17:30 UTC and your BFO calculations are completely wrong.

  710. Richard says:

    @Paul Smithson

    Some time ago you proposed a flight path starting at 18:32:30 UTC with a start latitude of 7.113°N and start longitude pf 95.2520°E at FL 340 (35,439 feet) in LNAV mode on an initial bearing of 190.55°T and at a speed of CM 0.8455.

    My apologies, that it has taken me so long to get around to running your flight path through my model, you probably thought justifiably, that I had forgotten you.

    Here are the results:

    https://www.dropbox.com/s/w016njiendcb87o/MH370%20Flight%20Path%20Model%20V19.8%20RG%20LNAV%20CM%200.8455%20FL340%20190.55%20Paul%20Smithson%2003082019.png?dl=0

    You calculate the flight path crossed the 7th Arc at 39.72°S. I calculate 39.51°S, which is not far off your result. There are quite strong headwinds toward the end of flight, which might explain the difference in our calculations.

    The BTOR standard deviation is 34.2 µs, which is good (nominal 29.7 µs). The BTOR at 22:41:22 UTC is -85.1 µs, which is high.

    The BFOR standard deviation is 5.98 Hz, which is a bit high (nominal 2 to 5 Hz). The BFOR at 00:11:00 UTC is 6.36 Hz, which is high.

    The PDA required for a MEFE at 00:17:30 UTC is +2.52% (nominal +1.5%), which means the fuel consumption does not quite fit. There is around 159 kg too little fuel, and the REFE would occur at around 00:07:55 UTC and the LEFE would occur at around 00:15:20 UTC.

    The RMS GSE is 3.19 knots and peak GSE is 3.265 knots, which is a high (nominal is < 1.5 knots).

  711. Victor Iannello says:

    @Viking said: When I recalculated the numbers in table 9 long time ago, I found that INMARSAT had also ignored the vertical component of the satellite velocity.

    That’s wrong, as Richard and I have been telling you. All components of the satellite’s velocity vector are included in the calculation.

  712. Victor Iannello says:

    @Sid Bennet said: Well, I can use the link that I posted and get the document. If anyone wants it sent another way, please suggest it.

    I already did. Please read my previous comment.

    In addition to Windows, @Andrew said this procedure works on OS X.

  713. Don Thompson says:

    @Viking said: When I recalculated the numbers in table 9 long time ago, I found that INMARSAT had also ignored the vertical component of the satellite velocity.

    I will ‘upvote’ both Richard and Victor’s comments. Inmarsat describes using ECEF geocentric coordinates in their work. The Perth GES antenna, the satellite ephemeris, and the aircraft positions were all dealt with as ECEF coordinates. To do so is both accurate and pragmatic, although ECEF coordinate space appears to remain a mystery to many casual observers.

    In your paper, you have chosen to demonstrate that a solution can be worked in the geodetic coordinate system despite the complexities of accounting for ellipsiodal models, oblateness, and local geoid variations.

    The ephemeris for I3-F1 was calculated in 3 dimensions, stepped over time, by a number of analysts who have contributed here, not only Duncan Steel.

  714. George Tilton says:

    @Andrew

    Thank you!

    This one will be right up your alley…

    Does the FMC automatically initiate step climbs as the weight burns off or must the crew manually initiate them?

  715. Andrew says:

    @George Tilton

    RE: “Does the FMC automatically initiate step climbs as the weight burns off or must the crew manually initiate them?”

    The FMC will ‘recommend’ a cruise flight level based on the cruise speed schedule, the size of the ‘step’ to the next higher cruise altitude (eg 1,000 ft, 2,000 ft, 4,000 ft), weight, wind and temperature. The FMC will calculate the optimum step point for a climb to the next higher level, and the crew can also enter planned step points at specific waypoints. However, the aircraft will not climb unless it is manually initiated.

  716. Sid Bennett says:

    https://www.dropbox.com/s/1d7ioamhqbzogi1/BFO%20documents.pdf?dl=0

    Oh! There were a few more steps. I haven’t done this in a few years…
    Thanks for your patience.
    Sid

  717. Victor Iannello says:

    I did some simulations and calculations using the PMDG 777 model to see how the tilt angle of the free surface of the fuel in a fuel tank might change as the aircraft enters into an increasing steep descent with a nose-down pilot input to the column. The results are highly dependent on initial conditions and the amount of force on the column. Nonetheless, I think the results are illustrative. I used standard atmospheric conditions with no wind.

    Time 0: Alt=39,230 ft, GS=441 kn, TAS=442 kn, VS=-2650 fpm, Pitch=-1°
    Time 12.5s: Alt=36,920 ft, GS=419 kn, TAS= 473 kn, VS=-22,250 fpm, Pitch=-30.5°

    (The plane loses about 1046 ft of total energy over this interval.)

    Assuming a constant acceleration over this time interval, I calculate:
    a_vert=-0.81g, a_hor=-0.092g

    where the positive horizontal and vertical directions are defined by the direction of travel over ground and up relative to ground, respectively.

    Treating the acceleration as a fictitious body force in the accelerating frame (the aircraft), the effective g-field is:
    g_vert=-0.19g, g_horiz=0.092g, g_tot=0.21g, g_dir=-63.9° (63.9° below the horizon and in the direction of travel)

    which means that an object moving with the plane would feel 81% lighter and be pushed forward (parallel to the ground).

    The tilt up of the free surface relative to the body of the plane would be: (g_dir-Pitch+90), or 27.1° at 0s, increasing to 56.6° at 12.5s.

    The bottom line is that for a fuel inlet at the rear of the tank, the inlet would likely be uncovered at low fuel levels.

    A ball placed on the floor of the plane would roll forward.

  718. Victor Iannello says:

    @Andrew said: However, the aircraft will not climb unless it is manually initiated.

    I can think of several ways that automated climbs can be obtained. Your thoughts are appreciated:

    1) With autothrottle engaged and the autopilot engaged, set a very high altitude that is unobtainable (like 45,000 ft), and as fuel burns off, the altitude will slowly increase. The thrust will remain at maximum

    2) Same as (1), but with the autothrottle disengaged, and the thrust set to a lower value, like 90%.

    3) With autopilot engaged in VNAV, and the MCP altitude set to a high altitude, set altitude constraints associated with waypoints along the route to replicate the manual step climbs.

  719. TBill says:

    @Victor
    Would a pilot realize that action could uncover the APU fuel inlet? It was a surprise to me…I was thinking down pitch would force more fuel supply.

    I would say in PSS777 the dual flameout intentional descent scenario seems reasonable explanation from FL250.

  720. Andrew says:

    @Victor

    RE: ”1) With autothrottle engaged and the autopilot engaged, set a very high altitude that is unobtainable (like 45,000 ft), and as fuel burns off, the altitude will slowly increase. The thrust will remain at maximum

    2) Same as (1), but with the autothrottle disengaged, and the thrust set to a lower value, like 90%.”

    I assume you have VNAV SPD or FLCH SPD engaged? If so, the aircraft will continue climbing very slowly as the aircraft weight decreases.

    RE: ”3) With autopilot engaged in VNAV, and the MCP altitude set to a high altitude, set altitude constraints associated with waypoints along the route to replicate the manual step climbs.”

    As far as I’m aware, that would only work while the FMC is in the climb phase.

    https://www.dropbox.com/s/gj6eucf96wdmk2m/B777.NAV.FMS%202%20-%20Altitude%20Step%20Points.pdf?dl=0

  721. Victor Iannello says:

    @Andrew: Yes, For (1) and (2), VNAV SPD or FLCH SPD is engaged.

    And yes, for (3), I am assuming that the FMC is in the climb phase.

    By the way, in the PMDG 777, while in VNAV, I often encounter problems transitioning to a climb phase by entering a new cruise altitude after a descent phase. I’m not sure whether that is a fault of the PMDG 777 model, or I am doing something wrong, or the real FMC has restrictions about multiple climb phases in a flight.

  722. Niels says:

    @all

    Through the following link a revision to my last paper can be accessed. The most important changes are related to the incorporation of “Case C”, with changes to the results and discussion sections accordingly. The main conclusions / recommendations remain the same.

    https://www.dropbox.com/s/mzclfjn5knhqb7b/WGS84_LRC_TT_report_v1_5.pdf?dl=0

    The most relevant cases in terms of results are B and C. If we take their 00:19 latitudes as limits, a region from S33.7 to S34.8 is defined (near the 7th arc). So 33.3 – 35.2 should imo be the mimimum range to be in focus. It is clear that a wider latitude range is allowed by the data, even under the same main assumptions (CTT/LRC/level flight in the final hours of the flight). I’m very curious for the outcome of the towed sonar data “meta-study” that Don Thompson is carrying out, as well as for the follow-up paper on Richard’s last publication.

    For a possible future search, I currently tend to advise to re-scan all or most of the “Phoenix” area (the area that OI went “around” / the “yellow” area in fig. 73 of the 3 Oct. 2017 ATSB report), before extending wider based on a “pull-up and glide” assumption. My opinion may change depending on the results presented from both studies mentioned, as well as the ongoing discussions on end-of-flight scenarios.

  723. Andrew says:

    @Victor

    RE: ”By the way, in the PMDG 777, while in VNAV, I often encounter problems transitioning to a climb phase by entering a new cruise altitude after a descent phase. I’m not sure whether that is a fault of the PMDG 777 model, or I am doing something wrong, or the real FMC has restrictions about multiple climb phases in a flight.”

    IIRC, you can force the FMC back into the cruise phase by entering a new cruise altitude, but there can only be one climb phase per flight unless the aircraft executes a go-around. In that case the FMC transitions back to the climb phase when TOGA mode is engaged or, in the case of a manually executed go-around, when the MCP is set to a higher altitude, the rate of climb is 600 ft/min or greater and the flaps are retracted from a landing position.

  724. DennisW says:

    @TBill

    I find it difficult to embrace the notion of rescanning areas that have already been scanned. The expectation that the “comprehensive survey” of flight paths will yield anything useful is fading pretty fast.

    Have we seen this movie before? Did it have a happy ending?

  725. Barry Carlson says:

    @DennisW;

    The expectation that the “comprehensive survey” of flight paths will yield anything useful is fading pretty fast.

    Just think of the exercise as if you’re digging up potatoes; the more you ‘turn-over’ the ground, the more you’ll get!

  726. Richard says:

    @DennisW

    We have now simulated well over 1,300 possible flight paths. I call that a comprehensive survey.

    … and yes … we are still carrying on looking for “potatoes” … as @Barry Carlson describes our work.

    A LNAV180 LRC FL390 flight path is still the only flight path simulated to date, that has a standard deviation BTOR < 40 µs and a standard deviation BFOR < 5 Hz and a RMS GSE < 1.5 knots and a PDA of 1.5% that fits a MEFE at 00:17:30 UTC (albeit with the air bleed system switched off for 38 minutes after diversion).

    I think the only thing fading fast is the basis for your continued unfounded criticism and impatience.

  727. David says:

    @Victor. About your downwards acceleration simulation, the below might be of some interest. I posted it 2 years ago to persuade another blogger, hoping that the signature would help. https://www.dropbox.com/s/80snphi23vc54wq/Forward%20forces%20from%20aircraft%20internals%20in%20aircraft%20acceleration.pdf?dl=0

    Another thought on the consequences of residual fuel flowing forward is that at the right descent angle it might cover the left main tank draw-down outlet and fuel a relight through that, though by that time the left engine might well have shut down anyway

  728. David says:

    @TBill. I hypothesised that the pilot might have poled forward before left engine fuel exhaustion. If fuel forward movement in the tank did not cover the left main tank draw-down outlet, and do that before boost pump inlets were exposed, the left would have flamed out. Besides, were the APU DC fuel pump inlet also exposed a relight (more likely at the higher atmospheric pressure) would have been prejudiced by vapour lock still. The higher atmospheric temperature would tend to reduce that risk though, depending on the aircraft acceleration, the nose down attitude could offset that.

    In summary, unless the pushing the nose down I speculated about occurred when there was enough fuel to keep either boost or APU DC fuel pump inlets covered once the nose was down, there would have been no log-on.

  729. David says:

    line 5, ‘pressure’ for ‘temperature’ please

  730. Victor Iannello says:

    @David: That analysis (implicitly) assumes the lift and angle of attack are zero, i.e., the flight path angle equals the pitch angle. Those effects are included in the calculations I provided.

    I think we have persuaded ourselves that the left engine did not re-start, as it would have occurred before the APU, and starved the APU of fuel.

  731. Richard says:

    @Niels

    Many thanks for your paper. It makes interesting reading, especially Case C.

    You note that your proposed Case C flight path passes close to waypoint BEBIM and that @ventus45 noted the subsequent track heads towards waypoint EKUTA. I also note that the earlier track passes directly over waypoint ISBIX and fits a track coming from waypoint AGEGA. In general a track via VAMPI – MEKAR – NILAM – AGEGA – ISBIX – BEBIM – EKUTA fits the satellite and fuel data quite well.

    I have run a LNAV path using these waypoints from your start point at 2.3600°N 93.7503°E at 19:41:03 UTC in LRC speed mode at FL335.3. My end point near the 7th Arc is 33.66°S 94.70°E (you calculate 33.67°S).

    The results are in the link below:

    https://www.dropbox.com/s/dubdj9ekaq7uy9f/MH370%20Flight%20Path%20Model%20V19.8%20RG%20LNAV%20LRC%20FL335.3%20Waypoints%20Niels%20Case%20C%20Full%20Report.png?dl=0

    The standard deviation BTOR is 49.5 µs (nominal < 40 µs), the standard deviation BFOR is 2.8 Hz (nominal < 5 Hz), the PDA is 1.6% (nominal 1.5%) and the RMS GSE is 2.9 knots (nominal < 1.5 knots).

    An interesting result.

  732. DennisW says:

    @Richard

    The basis for my impatience is your continued use of BFOR as a path qualifier.

  733. TBill says:

    @DennisW
    At the moment I feel like Victor’s new proposal for double-flame out is possible game changer. One implication is MH370 may have gone beyond Arc7 so yes that suggests rescan is less important than wider search. I am re-open on where MH370 crossed Arc7, believing the dual flame out assumption might lead to a new calculation of end point location(s). Working to understand that.

  734. Richard says:

    @DennisW

    You stated “The basis for my impatience is your continued use of BFOR as a path qualifier.“

    You had better get used to it.

    I am not throwing the BFORs out the window.

  735. Sid Bennett says:

    @Niels
    @Richard

    NILAM EMRAN ISBIX +186 FMT18:40 M=0.832 is another excellent fit in Barry’s model.

    The turn to EMRAN could be consistent with the BFO behavior at c18:22 and the first telephone call. This path results in a bearing of 186 between EMRAN and ISBIX.

    I am still trying to find a understandable discriminator between the various options after ISBIX.
    1)ISBIX + 186 is the simplest
    2) ISBIX +181 or so seems reasonable but requires more pilot input and therefore assumes more “arbitrary” parameters.

    Once one pins the FMT time at c18:40, adjustment of M brings at least several hypothetical paths into a very low error regime. At the moment, I do not think that the BFO can separate relatively similar paths after the FMT in the current situation.

    I am going to revisit the study of the azimuth distribution after the FMT that I circulated several years ago.

  736. Niels says:

    @Richard
    Thank you for your feedback, and for further exploring a possible route by waypoints related to case C.
    What I notice in your sheet is the rather high 19:41 BFO error, and the rather high BTO errors from 20:41 onwards. I see two possible ways to especially reduce the BTO errors (mean and RMS):

    a) If you prefer to keep constant FL and LRC setting for the whole 19:41 – 00:19 interval it is perhaps possible to shift the 19:41 latitude slightly northwards along the track.

    b) The errors could also be indicative for lower speed than LRC/FL335 in the 19:41 – 20:41 interval; so perhaps a lower average FL in this interval? This is in fact what my fig. 5 (right panel) is suggesting, however that is partially forced through satisfying the 19:41 BFO as my procedure is programmed for.

    I’m especially curious if a) would work; could you perhaps give that a try?

  737. Richard says:

    @Victor

    @TBill stated “At the moment I feel like Victor’s new proposal for double-flame out is possible game changer.”

    Are you now proposing a double flameout?

    You stated earlier “There’s another interesting aspect of allowing pilot inputs. When the fuel starts running low, the crossfeed fuel valves can be opened, which means both engines will flameout at nearly the same time, and the fuel range will be maximized. That means the double flameout would have occurred around 00:17.”

    Are you now agreeing there was an active pilot until the end?

    When I previously proposed the idea, you response was:

    “@Richard said: Can we assume an active pilot at the end of flight?

    I think it is a possibility. I’m not sure it is our base case. Let’s continue to discuss and think about it.”

  738. Richard says:

    @Sid Bennett

    You stated “NILAM EMRAN ISBIX +186 FMT18:40 M=0.832 is another excellent fit in Barry’s model.”

    Barry’s model only includes the weather data to the nearest integer latitude and longitude, does not interpolate for time between 18:00, 21:00, 00:00 and 03:00 UTC and does not interpolate for pressure altitude between 350, 300, 250, 200 and 150 hPa. My model performs a quadrilinear interpolation every minute precisely in 4D.

    Barry’s model only includes LRC mode for the Rolls Royce Trent 892 engines. My model includes CM, LRC, MRC and ECON speed modes.

    I do not believe that Barry’s model is accurate enough for our purpose.

  739. Victor Iannello says:

    @Richard: I have not concluded that there was pilot input, or a controlled glide. I think these are possibilities that we should discuss and think about. My “base” recommendation is still to search within a radius of 30 NM from our best estimate of the last transmission (area A1) and then extend to the south if that search fails (area A2). Since A1 << A2, that still seems the best way to proceed. My concern was if we concluded that there was no dive-recovery-dive scenario (as some here suggested), the search would end after completing A1.

  740. Sid Bennett says:

    @Richard
    As I have stated before, the version of Barry’s spread sheet that I use is constant M. I also use a constant altitude. I am not depending on the fuel consumption details.

    I have asked you to try one of my cases to see what the practical difference is. IGOGU at 18:40, GCP at 186T, M=0.82 FL=350.

  741. David says:

    @Victor. Your 6:56AM. My analysis approximate? True. Indicative of the principle? I thought so. Proved persuasive? Well, no.

    On left engine fuel draw down on pitching. As you imply, a restart including the relight I alluded to, that including an acceleration towards climb thrust, would deny the APU enough fuel for it to power the SDU reboot.

    Including the effects of pilot pitch downs earlier than otherwise, manual APU selections and the alternate electrical configuration (though that is not now favoured) might disclose feasibles. I will let you know should by chance I make any headway with that.

  742. Richard says:

    @Sid Bennett

    I have run a LNAV path using from your start point at waypoint IGOGU 7.516940°N 94.416667°E at 18:40:00 UTC in CM 0.82 speed mode at FL350 (36,503 feet). My end point at 00:19:37 UTC near the 7th Arc is 37.19°S 89.09°E (Barry calculates a model position at 00:19:30 UTC of 37.33°S 89.10°E).

    The results are in the link below:

    https://www.dropbox.com/s/5blk2oy4eoz4ulw/MH370%20Flight%20Path%20Model%20V19.8%20RG%20LNAV%20CM%200.82%20FL350%20186%20Sid%20Bennett%20Full%20Report.png?dl=0

    The standard deviation BTOR is 67.3 µs (nominal < 40 µs), the standard deviation BFOR is 4.2 Hz (nominal < 5 Hz), the PDA is 2.3% (nominal 1.5%) and the RMS GSE is 3.2 knots (nominal 1.5 knots, which are too high.

  743. TBill says:

    @Richard
    I like that you show BTO/BFO at the Arcs for the above path.
    Here is how I would currently look at that above data. I don’t know if maneuvers are in progress on at Arc2 and Arc6, so I do not weigh BFO too heavily. BFO match Arc3 to Arc5 is perfect. I am guilty of assigning priority to BFO over BTO, assuming BTO delta has possible explanation due to wind uncertainty etc. But if you say Arc5 BTO delta cannot be fixed, then the other way to get perfect BFO match above is active pilot on that route or alternate route with dogleg turn towards southeast.

  744. Sid Bennett says:

    @Richard
    Thanks for the prompt reply. I will study it. I am currently handicapped by being on vacation and having a single screen and no printer as well as such distractions as bridge, numerous cultural events etc. etc.

  745. sk999 says:

    Richard,

    Occassionally I will compare my route generator against yours just to see how well they compare. (Our BTO and BFO calculations are extremely close, so I stopped checking those.) For your most recent route, IGOGU @ 18:40, 0.82m, initial track 186, FL350, LNAV, you reach the 6th arc (24:11:00) at location -36.229, 89.251 whereas I reach it at -36.597, 89.226. There is a discrepancy of nearly 0.4 degrees in latitude. How significant is that?

    1. If I compute the ground track at your 6th arc position, starting from IGOGU, using the Vincenty inverse formula, I get a value of 187.24 deg, where you have a value of 186.88, a difference of about 0.4 degrees. My ground track at my 6th arc position is 187.403, 0.002 smaller than the true Vincenty value. Note that I do not enforce compliance with the Vincenty formula during the numerical integration.
    My BTO rms is 32 microsec, BFO rms 4.4 hz.

    2. If I modify certain parameters to get a route close to yours, all I have to do is increase the FL from 350 to 360 and change the initial heading from 186 to 186.02, I get a route that is very close to yours – deviation of 0.02 degree in 6th arc latitude, BTO rms of 77 microsec (stdev of 57 microsec), BFO rms of 4.2 hz.

    What does it mean? Simply computing a route with all input parameters set to predefined values and then determining a figure of merit at the end is not really meaningful. Small changes in the input parameters, and particularly those that are set to arbitrary values, can have a significant impact on the figure of merit. Small errors in the route propagator (whether yours or mine) can be compensated by small changes to those same input parameters.

    Not sure if you would agree with the above assessment, but I’d be happy to answer questions.

  746. DennisW says:

    @sk999

    What does it mean? Simply computing a route with all input parameters set to predefined values and then determining a figure of merit at the end is not really meaningful. Small changes in the input parameters, and particularly those that are set to arbitrary values, can have a significant impact on the figure of merit. Small errors in the route propagator (whether yours or mine) can be compensated by small changes to those same input parameters.

    I agree. It is an excerise in futility. I have several well founded objections to the basic approach (as you know). Have we seen this movie before?? I look forward to the long awaited final report. I will not go easy on it.

  747. Ventus45 says:

    @Richard

    When I plotted your path of 05Aug2019 (August 5, 2019 at 6:29 am above) in Google Earth, I notice, that it pretty much hits Cocos Island approach waypoint CC02 (300nm north west of Cocos Island at -9.378333° 92.623333°) almost on the button, a few minutes after 20:41:05.
    https://www.dropbox.com/s/t9xiy9vmfzba7eh/CC02%20waypoint%20on%20Richard%20Godfrey%20Parth%2005Aug2019%20%282%29.jpg?dl=0

  748. David says:

    @Victor. Returning to William Langewiesche’s article in The Atlantic, what I highlight from that are the accounts of his first hand interviews with friends of Zaharie Shah and others knowledgeable of him. Those portray him as troubled emotionally, adding more weight to the view that the Malaysian police assessment is unsound.
    Langewiesche coupled these with Zaharie’s home flight-simulations and with his final in-flight messages, which uncharacteristically included extraneous information yet excluded other, in suggesting that Zaharie was the likely culprit in a murder-suicide.

    Oddly though, he makes no mention of various accounts that Zaharie was politically active, being concerned in particular about Malaysian Government corruption. Moreover while his theme is critical of the Malaysian investigations and information concealment, he does not address whether or not that political aspect was investigated by the Malaysian police; and if not why not.

    Assuming this to be an oversight I now conflate Zaharie’s mental condition as above with possible motivation from this apparent concern about political corruption, offering for discussion some speculation (not entirely loose) on what might have been in his mind in sum.

    The Malaysian Opposition had been stymied and he was a ‘fixer’. To this end he might have expected that disappearance at his hands of such a large aircraft and people aboard would have a major consequence for the Malaysian Government: not being able to explain that would make it appear incompetent at home and abroad. His hope therefore might have been that this could be made into a catalyst in bringing it down.

    Reconstructing what may have been his plan, it was to confuse air traffic controllers as to where the aircraft was, once well abroad, and then to approach Malaysia as an unidentified aircraft – knowing he would be tracked by military radar – demonstrating that as such an aircraft could fly to the major air force base at Penang with impunity, finally disappearing out of radar range, apparently heading north west.
    What he would picture is that he would be tracked not just by military radar but that ATC could trace him via primary radar records; and that other nearby countries might trace him also. Thus there would be a reasonable prospect that the incident could not be concealed; for long anyway.

    In his mind, adding to this public embarrassment there would be the opprobrium of not having alerted other countries to this rogue aircraft at the time.
    In particular he would be foreseeing that those to the north west surely would have expected timely warning of a potential terrorist attack heading their way.

    Also, he would expect other governments whose nationals were aboard the aircraft would join families of those missing and the media at home (perhaps) and abroad in expecting an explanation for the disappearance together with action to find the passengers.

    The intention of his southern Indian Ocean destination was to minimise the prospect of flotsam giving away that there had been a crash or at worst the crash location, thereby sustaining the embarrassment while minimising the risk that his responsibility would be uncovered, to the detriment of family and friends and to his plan’s success.
    He might have expected that any wreckage that floated for some time would circle the earth down south and in any case any that drifted ashore with time would pass unnoticed or at the least would give little clue as to where the main wreckage was located, and what had happened.

    He would stay and watch the final leg to its end in case some addressable problem was encountered on the way and to ensure by a final dive that there was no ELT signal at impact and that the aircraft flotsam was pulverised, (though the latter was not realised).

    In the above scenario he would not have known that his turn and leg south could be deduced later by a novel and highly skilled analysis of some characteristics of routine SATCOM confirmations that the aircraft was operating still, and of unanswered enquiries to it. Quite possibly he would not even have known that the operator of that particular satellite had started recording these characteristics.

    While I do not maintain the above presents a strong case I prefer it to its current alternatives.

  749. Richard says:

    @sk999

    You stated “Simply computing a route with all input parameters set to predefined values and then determining a figure of merit at the end is not really meaningful.” @DennisW added “futile”.

    In the paper linked in this post Bobby, Victor and I ran 828 flight paths without any prior constraints. None of the input parameters were set to predefined values. All 7 input parameters below were allowed to vary without constraints.

    1. Start Time.
    2. Start Latitude.
    3. Start Longitude.
    4. Flight Level.
    5. Lateral Navigation Mode.
    6. Initial Bearing.
    7. Speed Control Mode.

    The outcome of the comprehensive survey of possible MH370 flight paths shows only one flight path has a unique fit to the data we have, a LNAV180 LRC FL390, which passes close to waypoint BEDAX.

    In the current and previous posts, 7 contributors each with their own flight path model have asked me to run 16 different flight paths to compare with their own results. Similarly Bobby, Victor and I have run a comprehensive test program comparing our independent models. I have found both sets of comparison very helpful and have helped correct a number of errors in both my model and the models from others. This is not meaningless or futile.

    A key argument of many contributors has been that we have only considered flight paths requiring a passive pilot. Many of the flight paths proposed by contributors require an active pilot. You can find a table containing a list of all 16 flight paths proposed by various contributors in the link below:

    https://www.dropbox.com/s/yv9a8ma3cpubfrv/Contributors%20Flight%20Path%20Proposals.png?dl=0

    You will see from the results that none of the flight paths proposed by various contributors result in a standard deviation BTOR < 40 µs and a standard deviation BFOR < 5 Hz and a PDA < 1.5% and a RMS GSE < 1.5 knots.

    The only flight path that comes close is the @Niels Case C.

    For the benefit of @DennisW, you will also see from the results that none of the flight paths proposed have a standard deviation BTOR < 40 µs and a PDA < 1.5% and a RMS GSE < 1.5 knots, even ignoring the BFOR results.

    I do not consider that either the comprehensive survey or the subsequent requests from @TBill, @Niels, @Hank, @Nederland, @Viking, @Paul Smithson and @Sid Bennett are meaningless or futile.

    In fact I find your and @DennisW's attitude speaking out against such experimentation and model comparison as meaningless and futile.

  750. Richard says:

    @DennisW

    You stated “I look forward to the long awaited final report. I will not go easy on it.”

    The first positive thing you have said in weeks.

  751. George G says:

    Dennis W
    sk999

    I, too, look forward to the long awaited report forthcoming from:
    Dr B;
    Richard; and
    Victor.

    But, Dennis W, I do not necessarily see it as “an excercise in futility”.
    More to the point, it makes one stated assumption so that some groundwork can be utilised. That is simply that at “Arc #2” the aircraft was already on a single path in the general southwards direction, and that it flew through “arcs” 2 through 6, -possibly-even-7- unninterrupted.
    If so, Then by progressively, i.e. incrementally, proceeding through a range of potential conditions a (very large) set of paths could be inspected for the possibility and/or probability that that was the actual path of the aircraft.

    And, sk999, those small changes in input parameters (i.e.initial conditions) are just why the progressive and incremental stepping through the (repeat, very large) set of paths is required.

    And, of course, Dennis W, no-one expects you to “go easy on it”. More the better.

    __________________________________________________

    There are other questions remaining. See post to follow.

  752. George G says:

    All,

    Following on from the above:
    1. If we do not consider only that after some initial short time into the flight generally southwards the aircraft was sent on a single path, then the potential for finding the likely final resting point is diminished. If the aircraft even took one change in direction after what has been termed the “FMT” then all bets are lost. The number of possible, and potential, and variable, and conspirational, flightpaths increases “exponentially” in the figurative, not literal, sense. Even, with only one turn sometime after the “FMT”, the potential for determining the final resting place is dramatically reduced.
    2. So, re-inforcing 1. above, the only likely and “worthwhile” search criteria is to define a “point” where the aircraft remains may lay presuming that the aircraft was flown, or flew, on a single path. This may well be worth a look-see. (i.e. Search)

    3. A serious question remains as to how the aircraft descended to it’s final demise into the ocean. The “worst” case would be to assume that the aircraft somehow was able to glide in a straight direction (repeat “straight” direction) after loss of engine thrust. Other considerations exist, as have been explored on this blog of Victor’s. So this remains a serious point of discussion and resolution of probability.

  753. Richard says:

    @Ventus45

    You stated “When I plotted your path of 05Aug2019 (August 5, 2019 at 6:29 am above) in Google Earth, I notice, that it pretty much hits Cocos Island approach waypoint CC02 (300nm north west of Cocos Island at -9.378333° 92.623333°) almost on the button, a few minutes after 20:41:05.”

    An interesting observation! Many thanks!

    I calculate that the flight path proposed by @Sid Bennett passes 2.147 km to the East of Waypoint CC02 at 20:46:23 UTC.

    I have often thought that Cocos Island (YPCC) might have been an initial target. As you say, for an approach to Cocos Island from the North West, CC02 at a range of 300 NM would have been the first waypoint to set, followed by CC01 at a range of 150 NM and finally PCCNE and PCCNI to line up with Runway 15 to arrive at around 21:23 UTC, where the surface wind was 16 knots from 130°T. There would have been plenty of fuel to make it to YPCC.

    Unfortunately it does not fit a MEFE at 00:17:30 UTC.

  754. Richard says:

    @sk599

    I find your latest comment a bit schizophrenic. On one hand you show how the satellite data is very sensitive to small changes in initial bearing or flight level, on the other hand you reject BTORs or BFORs within an appropriate range as a meaningful differentiator.

    You can’t have it both ways.

    We have amply demonstrated that the BTORs, BFORs, fuel constraints and GSE are all good differentiators of the actual MH370 flight path.

  755. Victor Iannello says:

    @sk999 said: Note that I do not enforce compliance with the Vincenty formula during the numerical integration.

    Bobby, Richard, and I spent some time comparing our numerical methods. I generate paths by numerical integration, probably similar to what you do. Bobby uses the Vincenty formula for great circle paths. I believe Richard has at times used both. The bottom line is that if properly implemented, including an adjustment for altitude, the results yield essentially equal results for great circle paths. (We’ve used on-line calculators as the final arbiter for the great circle paths, recognizing that the distances are slightly off because those on-line calculators implicitly assume an altitude of zero.) If you are seeing a large discrepancy, I suspect there is an error in your method.

  756. Victor Iannello says:

    @DennisW said: It is an [exercise] in futility.

    It is not certain that this effort will result in a search that finds the plane. However, our choices are to call it quits, or to determine if there is something that is objectively unique about this path that would support a search in this area. I have not seen another analysis that defines a different area that is manageable in size, is objectively of high priority, and has a reasonable chance of success.

  757. Victor Iannello says:

    @David: What you propose might have transpired.

    Although I think William Langewiesche did a respectable job at summarizing the incident, and his piece might be the narrative that most people reference in the future, we have to acknowledge what it is. He spent several weeks, conducting long interviews with people like Mike and me, traveled to Malaysia to meet and talk with more people like Blaine, Grace Nathan, and friends of the captain, and then he took what he learned, combined it with his knowledge of aviation and common sense, and assembled the piece into an engaging story. There is nothing new in that story. Even the quotes from the pilot-friend of the captain are similar to statements made previously by retired Malaysia Airlines chief pilot Nik Huzlan.

    I’ll add that there is another version of the story that some knowledgeable people believe. In that version, the captain expected to be shot down, first by Malaysia (as evidenced by a path that passes near Butterworth), and then by Indonesia. When neither country responded, his third option was the dump in the SIO. For a number of reasons, I’m not a fan of this theory.

  758. TBill says:

    @David
    @Victor
    I agree with David’s point that MH370 might have been trying for many countries radar to see MH370 flying over Malaysia, hence FL400 at IGARI. if alive and active, I do not know what style crash was being tried for (I am still thinking sinking in Broken Ridge or other deep spot).

    I agree with with Victor’s alternate version, that intercept/shoot down by Malaysia, although maybe not Plan A, might have been a concern or possibility that was planned around. My personal feeling is the perp was trying for deniability in Malaysia, so I am thinking a number of steps might have taken to avoid definitive ID of pilot as the perp (DFDR management etc) in the intercept case.

  759. vodkaferret says:

    Re: rescanning areas

    as there is discussion about rescanning areas already checked, albeit presumably with newer (OI) technology – how confident should we be in terms of the analysis of the scans already made? There must be certain assumptions made during analysis of scan data (minimum target size etc) that may be invalid in the final result. Does anyone know if the data from the initial scans has been reanalysed some time later (say, following lessons learned from the discovery of the San Juan) or not?

    it would be shit, but also very human, if we already had the information we needed to solve the problem but we’re just looking at it in the wrong way.

  760. Sid Bennett says:

    To be clear, I asked Richard to run the case approximating IGOGU+186T so as to be able to compare exact like cases computed by his model and Barry’s model. I expected slightly different results and they are slightly different. As time permits I am studying Richard’s results and will comment when I have some conclusion to state.

    The case I proposed is a simplified version of the complete path from 18:22 to splash which satisfies other constraints such as the FMT time at a waypoint and an azimuth that corresponds to a minimum total error in the combined normalized errors of the BFO and BTO.

    Due to the differing atmospheric models, adjusting the value of M slightly to achieve the minimum BTO error is merely using a “variable constant” to tweak the model. One could have chosen a slightly different cruise altitude etc.

    Patience.

  761. Viking says:

    @Sid, @Richard, @Victor, @Don

    Concerning: Satellite height contribution to BFO.

    When I did the calculations I found deviations from 0Hz (around 19:41) to 7Hz (last points). Up to 40% of this may in principle be due to higher order effects which I ignored by using perturbation calculations of the satellite non-ideal effects. However, it would be strange that the size of the deviations should fit the size of the satellite height contribution (when compared to the Ashton paper, table 9).

    On the other hand it is a relatively small effect compared to the effect an open door would have, so it is not possible to make a rigorous disentanglement.

    Effects from cycling temperature and pressure of the SDU oscillator are most likely somewhere in-between these two in magnitude, so that alone makes rigorous disentanglement impossible.

    Finally, there are some minor effects of not using a proper inertial system and (even smaller) effects from ignoring the gravitational red shift.

    At the time I did the original calculations all these effects made me conclude that the absolute BFO uncertainty was most likely higher than 7Hz. I handled this by giving the BFO a low weight in the optimization. This means that any errors below 15Hz are unlikely to have significantly affected the end result. However, the disadvantage is that one must use other information to choose the correct end point (around 34.4S or around 13.4S).

  762. Sid Bennett says:

    @Ventus45

    CC02 is directly on track beyond ISBIX and would be an acceptable waypoint it seems. CC02 does not show on the SkyVector chart, but SkyVector will accept it as an input.

    I presume that it is a local waypoint associated with an approach to Cocos.
    Would this be known to a person using a computer-simiulated flight?

  763. Don Thompson says:

    @vodkaferret asked “how confident should we be in terms of the analysis of the scans already made?

    Fugro’s delivery to ATSB involved a range of measures that assessed confidence for their work. The work is described in the ATSB Operational Search report.

    Phoenix International, contracted by Malaysian parties, operated separately from Fugro. It is my understanding that the products from this work were submitted to some, but not all, of the quality assurance procedures that Fugro, ATSB, and Geoscience Australia applied to the data Fugro acquired.

    And “Does anyone know if the data from the initial scans has been reanalysed some time later.

    I’m not aware of any project that has undertaken wholesale re-analysis of the data acquired during the seafloor search. That would be require a significant effort.

    At present, I’m working to re-evaluate some of the ‘data holiday’ categories.

  764. DennisW says:

    @Victor

    @DennisW said: It is an [exercise] in futility.

    It is not certain that this effort will result in a search that finds the plane. However, our choices are to call it quits, or to determine if there is something that is objectively unique about this path that would support a search in this area. I have not seen another analysis that defines a different area that is manageable in size, is objectively of high priority, and has a reasonable chance of success.

    The probability (90%) is that the aircraft wreckage is in an area that has not been scanned. Lower it to 80% if it makes you feel better. How on earth can anyone propose searching at 34S unless you believe in the dive-glide-dive scenario?

  765. Victor Iannello says:

    @Sid Bennet asked: Would this be known to a person using a computer-simiulated flight?

    CC2 is the “transition” waypoint for the VORY15 and VORY33 approaches. If YPCC was selected as the destination airport, and either the VORY15 or VORY33 approaches was subsequently selected, then CC2 can optionally be chosen to be part of the route.

    That said, I don’t see a reason for doing this if a landing at YPCC was anticipated, nor do I see a reason for faking an approach to an airport with no approach radar.

  766. Victor Iannello says:

    @DennisW: How do you explain the discovery of the San Juan after it was missed the first pass? In that case, the first pass was using AUVs; in this case, the first pass for much of the area was using a towfish, which cannot be used to scan steep slopes.

    We are prioritizing among options, none of which are “unblemished”. We would be making a recommendation with the understanding that there is no guarantee of success. OI seems to have some interest in pursuing a new search, and we owe it to them to provide them with data and analyses to help them make their decisions.

  767. Richard says:

    @DennisW

    Great science! “Attack! Attack!”

    You stated “How on earth can anyone propose searching at 34S unless you believe in the dive-glide-dive scenario?”

    The answer is because (unlike you), we have done our homework.

  768. Victor Iannello says:

    @DennisW: I’ll give you another example of how success probabilities might be incorrectly applied. In the case of AF447, an area of the seabed was “cleared” after neither the pinger from the FDR or CVR was detected in that area. Metron’s main contribution to the search was to consider that there was a finite probability that either both pingers failed, or one or both was operating but not detected (rather than assigning these probabilities as zero). On this basis, the previously cleared area was scanned, and the debris field was found.

  769. DennisW says:

    @Richard

    Great science! “Attack! Attack!”

    I am not attacking you. I am simply looking at probabilities. A 10% (or 20%) probability is not zero. It simply makes more sense to dedicate resources to an 80% or 90% probability. Hey, you are doing the work. Far be it for me to criticize you for that, although applying Gaussian stats to a non-stationary and non-ergodic process is unforgivable.

    @Victor

    Same general theme in response to your examples. Of course, it is possible that the wreckage was missed.

  770. Richard says:

    @DennisW

    You stated “applying Gaussian stats to a non-stationary and non-ergodic process is unforgivable.”

    I left out the BFO result, just for you and still proved my point!

  771. ArthurC says:

    @DennisW

    I look forward to the long awaited final report. I will not go easy on it.

    Looks like you already made up your mind…
    How is that constructive or helpful?

  772. Sid Bennett says:

    @DennisW

    Can you give us some relevant stationary and ergodic processes?

  773. DennisW says:

    @ArthurC

    A great deal has been written on positive and negative feedback. The studies always conclude that positive feedback is more beneficial when dealing with beginners, and that negative feedback is more beneficial when dealing with experts. I regard Richard and Victor as experts.

  774. Andrew says:

    @Victor

    RE: ‘CC2 is the “transition” waypoint for the VORY15 and VORY33 approaches. If YPCC was selected as the destination airport, and either the VORY15 or VORY33 approaches was subsequently selected, then CC2 can optionally be chosen to be part of the route.’

    It’s interesting that CC02 is included in the FMC database as a transition for the VOR approaches to YPCC. CC02 is not on an airway and it is not part of any published approach procedure for YPCC. I wonder where it came from?

  775. DennisW says:

    @Sid

    Most things are stationary and ergodic. A simple example is the roll of a pair of dice.

    If you roll one pair of dice 100 times or 100 pair of dice once, the statistics will be the same – the process is ergodic. It is also stationary. You will get the same result on Sunday that you did on Saturday. The distribution of height for 1000 men in California will be the same as 1000 men in Pennsylvania. Although the California men will probably have a higher average IQ.

  776. DennisW says:

    @Sid

    On the other hand 10 random walks of 20 steps will not have the same mean and variance as a single random walk of 200 steps.

  777. Victor Iannello says:

    @Andrew: I made a mistake. There is no CC02 in the FMC database, as far as I can tell. The CC2 transition I was looking at refers to an approach with a hold at fix CC. Here’s the language, including the go-around:

    APPROACH VORY15 FIX FS15Y AT OR ABOVE 1300 RNW 15 FIX CC AT OR ABOVE 510 TRK 151 UNTIL 1500
    TRANSITION CC2 HOLD AT FIX CC LEFT TURN INBOUNDCOURSE 308 ALT 1500 SPEED 210 LEGTIME 1 TRK 316 UNTIL 7.5 FROM FIX CC

    where CC is S 12 12.0345 E 96 50.45382
    and FS15Y is S 12 6.92298 E 96 47.208

  778. Sid Bennett says:

    @DennisW

    It is some 50 years since that I first dealt with this subject, reading Weiner’s little book.

    The average height of the population of California has increased over the past century. As such the mean height is a function of time and the process is not stationary.

    In reality, the mathematics of sadistics only approximates reality and it is the judicious application of the math that helps us reach practical conclusions.

    Most of the math blows up at the short and long time scales.

  779. DennisW says:

    @Sid

    Yes. BMI has changed a lot over the last century as well.

  780. DrB says:

    @DennisW,
    @Sid Bennett,

    The Inmarsat BFO values contain contributions from at least 3 independent sources of error.

    On short time scales, less than a few minutes, OCXO frequency noise is negligibly small and is immaterial. The dominant sources of noise are two: (1) a random, stationary, and ergodic electronic read noise of about +/- 1 Hz, and (2) systematic transmit frequency errors in the AES which are caused by the use of limited precision trigonometric functions used in computing the compensation frequency value. These computational errors typically produce a sawtooth error pattern of several Hz peak-to-peak. You don’t see these when the aircraft is at the gate because the aircraft is stationary with respect to the assumed satellite location, and therefore the compensation frequency is constant. When the aircraft is flying, the sawtooth period and amplitude vary along a route in a way which we cannot predict because we do not have the software used in the AES (although DSTG did). The impact of these transmit frequency errors is to induce an identical pattern in the observed BFO values. Thus, all BFO values during the MH370 flight period contain both the small random electronic read noise and the transmit frequency errors caused by trig function errors.

    We also know something about the OCXO frequency drift, which becomes important on time scales of hours and longer. Whenever the observed frequency drifts more than 16 Hz from its nominal value (and when on the ground and after at least 24 hours), a transmit frequency offset correction is made of 16 Hz in the AES. Reportedly, this happens occasionally, but not every day. So, the typical OCXO drift is less than 16 Hz per day. During the period of interest in fitting routes from 19:41 to 00:11, the elapsed time is 4.5 hours. Then,with a typical drift of less than 16 Hz/day, over 4.5 hours the typical drift would be less than 7 Hz assuming the drift varies as the square root of time.

    In their MH370 paper, Inmarsat says: “Figure 15 shows the measured BFO for flight MH21 that travelled from Kuala Lumpur to Amsterdam at the same time as the MH370’s final flight, overlaid with the upper and lower BFO predictions (± 7Hz) using the refined BFO model. This suggests that ±7Hz is a conservative estimate of the typical accuracy BFO calculation achieves, as well as illustrating the BFO versus time characteristics for a flight moving along a path close to the Northern route. While the validation demonstrates the general accuracy of the BFO technique, it is important to note that agreement is only achieved with ±7 Hz accuracy during this flight, and to assume better accuracy for the measurements taken on MH370 would be unrealistic.” So, Inmarsat says the BFO errors on a single flight should be within +/- 7 Hz.

    In their book on MH370, DSTG said: “Empirical statistics of the residual measurement noise w^BFO/k were determined using the previous 20 flights of 9M-MRO. Data points corresponding to when the aircraft was climbing or descending were excluded. . . . The statistics show that even when outliers are discarded a standard deviation of about 4.3 Hz is applicable. As discussed above, to be conservative and allow for potential variation in the d f bias k (xk; sk) value on the accident flight, our model assumes a noise standard deviation of 7 Hz. Section 5.5 illustrates the sensitivity of the BTO and BFO measurements to variations in the aircraft state. Figure 5.5 shows a histogram of the 3392 in-flight BFO errors. On-tarmac BFO errors were excluded due to the pre-biasing described above. A Gaussian fit to the distribution is shown as a black line. It can be seen that the distribution shows some non-Gaussian features and the tails of the distribution.”

    In my analysis, I use the same PDF shown in Figure 5.5 by DSTG. Roughly only about 10% of the values in the tails lie outside of the +/- 7 Hz Inmarsat criterion. So, I would say that the Inmarsat (+/- 7 Hz) and the DSTG (4.3 Hz standard deviation) criteria of BFO errors are generally consistent with one another.

    What I do is different than what DSTG did in one way. They broadened the standard deviation of the BFO errors to accommodate an unknown shift in the bias frequency. When evaluating a single trial route, I find the mean BFO error but I don’t use it as a figure of merit, because the mean might have been significantly affected by a cold soak immediately after diversion for up to an hour. Typically I see a mean value of about -3 Hz when I use the bias frequency which best fits the KLIA data before take-off. I do use the standard deviation (about the mean) of the BFO error during the flight as a figure of merit, just as Inmarsat and DSTG did. It is a perfectly valid method since we have data from at least 20 flights informing us of the probability density function. Trial routes which have BFO errors inconsistent with the PDF in Figure 5.5 are suspect. It’s a silly notion to discard the BFO residuals altogether. We have exactly what we need in Figure 5.5 to do an appropriate comparison and evaluation to see if the BFO errors are larger than typical values, in which case systematic route parameter errors may be contributing in a significant way. Some trial routes I have fitted show BFO residuals similar to Inmarsat’s Figure 15, being within about +/- 4-5 Hz.

    So, let’s not throw out the baby with the bath water. The BFO data are actually well understood and their dispersion about their mean is a useful figure of merit for comparing MH370 trial routes.

  781. Andrew says:

    @Victor

    RE: “The CC2 transition I was looking at refers to an approach with a hold at fix CC.”

    Thanks – there are two variations of the VOR-Y 15 approach, one for Cat A/B aircraft and the other for Cat C/D aircraft. The ‘Cat’ is the ICAO PANS-OPS aircraft category, which is based on the aircraft’s approach speed at its maximum landing weight. In this case, ‘CC2’ refers to the Cat A/B approach; CC2 is not a waypoint.

    FWIW, the B777 is a Cat D aircraft and would therefore fly the Cat C/D approach, which is the CC1 transition.

  782. Ventus45 says:

    @Sid Bennett

    Re your August 6, 2019 at 10:36 am

    Some time ago, Warren Plats compiled a kmz file of waypoints (which is what I have in my Google Earth) in the SIO (and elsewhere) presumably sourced from the on line aviation sim community, of which we know Z was an avid member.

    I think it is more than likely that Z had similar (perhaps even the same) data in his home simulator.

    On further investigation, I have discovered that Waypoints CC01 and CC02 are two out of ELEVEN (CC01 to CC11) on what appears to be a custom SIM gamer’s air route, between YPCC (Cocos Island Airport) and VRMM (Velana International Airport). See skyvector links below.

    https://skyvector.com/?ll=-3.7984839775168275,89.85937500369423&chart=304&zoom=8&fpl=N0480F350%20YPCC%20CC01%20CC02%20CC03%20CC04%20CC05%20CC06%20CC07%20CC08%20CC09%20CC10%20CC11%20VRMM

    https://skyvector.com/api/navlog?94581

    So the plot perhaps thickens somewhat.

    Any sim pilot planning a flight to Cocos Island would have that information, or access to it.

    I wonder what TimR might make of this ?

  783. Victor Iannello says:

    @Ventus said: Any sim pilot planning a flight to Cocos Island would have that information, or access to it.

    I don’t know who defined those waypoints, but they are NOT fixes included in my “AIRAC Cycle : 1309 (22/AUG/2013 – 18/SEP/2013) – Ver.1”, nor are they part of any SID or STAR. I really doubt they are supplied as part of the navigation database of 9M-MRO.

  784. DennisW says:

    @DrB

    You say It is a perfectly valid method since we have data from at least 20 flights informing us of the probability density function.

    Figure 5.5 of the DSTG book was created using almost 3400 BFO measurements from 20 different flights. The fact that the mean BFO error reported by the DSTG is near zero is indicative that the DSTG treated the data as an ensemble of measurements. Do you really believe the mean BFO error of any of the 20 flights was on the order of 0.20 Hz (as reported in Table 5.1)? The mean value of an ensemble of random walks is indeed near zero. The mean value of any one random walk is very unlikely to be close zero. The PDF of the ensemble is not the same as the PDF of any single member of the ensemble for non-ergodic processes.

    You say The BFO data are actually well understood and their dispersion about their mean is a useful figure of merit for comparing MH370 trial routes.

    That is good news. Can you explain the results shown in figure 5.4 of the DSTG book? I do agree that the BFO data is well understood, and cannot be used as a flight path qualifier. I don’t suggest tossing the BFO data out. It is useful for indicating the aircraft flew South after the FMT, and descended very rapidly at the end of the flight.

  785. Sid Bennett says:

    @Victor
    SkyVector accepts and plots CC02.

    @DennisW
    As DrB rightly points out the hardware-related errors have different behavior at different time scales. DrB’s limits on “measurement noise” are, in my judgment, quite conservative.

    I have used a more empirical approach that, once having established the bias offset and any adjustment at each ping ring (mostly at 20:41), the same settings are kept for each path study. While this may have some weakness in that it results in an excellent fit in the best case, the variation of BFO error when other parameters are “adjusted” is smooth and consistent.

    The telephone call data qt 18:40 is consistent with the FMT and the BFO error for the second telephone call is quite reasonable. Whether used or not, the result is essentially the same.

    It is a mistake not to use the BFO. It is an independent contribution to the determination of position.

  786. Victor Iannello says:

    Sid Bennet said: The telephone call data [at] 18:40 is consistent with the FMT

    It’s also consistent with an ongoing descent at a constant vertical speed. I don’t think we can assume it was one or the other.

  787. Richard says:

    @DennisW

    You ask “Can you explain the results shown in figure 5.4 of the DSTG book?”

    That is easy for a relatively short 3 hour 20 minute flight in a straight line from Mumbai to Kuala Lumpur on 2nd March 2014.

    More challenging is the longer 7 hour 50 minute flight from Kuala Lumpur to Amsterdam on the evening of 7th March 2014.

    In the Inmarsat paper Fig. 15, we are given a plot of 30 BFO values taken from flight MH16 from Kuala Lumpur to Amsterdam on the evening of 7th March 2014.

    In the DSTG book Fig. 9.9, we are given the flight route, altitude, heading and Mach of the flight MH16 from Kuala Lumpur to Amsterdam on the evening of 7th March 2014.

    From the DSTG data in their book, I have reconstructed the flight path of MH16 and compared the predicted BFOs to the actual BFOs from the Inmarsat data in their paper.

    I find the BFO data to be a reliable indicator of the flight path and not an inconsistent unpredictable erratic non-ergodic data set, as you keep trying to make out.

    There are no random walks starting after about 100 seconds into the flight as you claim.

  788. TBill says:

    @Victor
    For the dual flame-out case, I suppose that implies a passive MH370 was going full speed over Arc6, and almost full speed over Arc6-to-Arc7. As a first approximation, Arc6-to-Arc7 average speed is 95% of full speed. Which allows the following simple calcs:

    QUICK and DIRTY ARC6 BFO STUDY (Looking for BFO=252 as Measured)
    Arc6 at -10 South BFO=233 Heading=90 Speed(Calc)=319 “Xmas Island Case”
    Arc6 at -21 South BFO=238 Heading=139 Speed(Calc)=346 “Zenith Seamount”
    Arc6 at -30 South BFO=248 Heading=170 Speed(Calc)=415 “Home Sim Path approx”
    Arc6 at -34 South BFO=253 Heading=180 Speed(Calc)=456 “180 South Case”
    Arc6 at -37 South BFO=257 Heading=188 Speed(Calc)=467 “38 South Case”

    Notes:
    (1) Assuming straight True Track passive flight
    (2) All flight paths radiate from a 10S 93E central focal point (near BEBIM)

    SUMMARY
    For a passive flight, assuming Arc6 BFO=252 exact, the 180 South case is the best fit. Speed fit is very good considering the uncertainties. However, keep in mind at Arc5 the 180 South match is not the BFO best (38 South is better there for BFO).

    There are two ways to “fix” the BFO to the east if they are too low. If the active pilot changes heading to the North Pole (360 deg North) you can add +10-15 units to the BFO. The is how the Captio (Xmas Island) case gets a more respectable Arc6 BFO.

    Secondly, a relatively minor ascent of 600 ft/min also adds about +10 units to the BFO. I feel it is possible an active pilot was ascending into Arc6 in preparation for a longer glide.

  789. DennisW says:

    @Richard

    It is not correct to say the random walk “starts” after any particular time. If I gave that impression I am sorry. The random walk component is always present. It becomes dominant in the Allan Variance of quartz oscillators after about 100 seconds or so.

    https://photos.app.goo.gl/QM95HPLQGQv3x6GE8

    I don’t get any sense of convergence in our discussions. Probably best at this point to let it lie.

  790. Ventus45 says:

    @Victor,

    I did not suggest, or mean to suggest, that those waypoints were in 9M-MRO’s FMC, only that they are possibly in Z’s home sim.

    Z was known to have numerous other aircraft on his sim as well, not JUST the 777.

    Z was also (apparently) known to practice landings at “remote islands”.
    Practice landing “what”, on those “remote islands” – a 777 – I doubt it.
    It had to be some other aircraft – surely – but what ?

    However, going back to all those waypoints themselves, I found it very curious, “very” in fact, that all of those waypoints:-
    (a) are in Skyvector in the first place, and
    (b) that they are all exactly 150 nm apart.

    Why exactly 150 nm apart ?
    There has to be a reason for that, don’t you think ?

    Skyvector is not just for sim gamers, it has real world use for lightie punters.
    So why are those waypoints in Skyvector in the first place ?
    Someone put them in, for an obvious reason, to be used in real world.
    So who, and why ?

    I think I may have a plausible answer for the “why”.

    The middle of oceans certainly qualifies as a “designated remote area” for lighties for SARTIME requirements.
    Typically, SARTIME reporting protocols for lighties in remote areas are (usually) hourly.

    So, what typically cruises at 150 knots, have to report hourly, and could plausibly use the route, and did it often enough to justify making a route template in Skyvector ?
    Regular airline service(s) – no way.
    But ferry flight(s) – of small stuff – possible / probable.

    The de Havilland Canada DHC-6 Twin Otter cruises at 150 knots.
    There is a whole fleet of float equipped twotters operating all over the Maldives.
    They all had to get there somehow.
    Plenty of twotters in Australia also.

    Digging around, I found a web site devoted to the twotter with some interesting ferry flight data.
    In it I found a few FERRY flights between VRMG and YPCC – IN BOTH DIRECTIONS.
    See http://twinotterspotter.blogspot.com/p/dhc-6-ferry-gallery.html

    Eg-1: VH-AQU DHC-6-300 SN 557 YSBK Sydney, Australia to Bern, Switzerland.
    http://2.bp.blogspot.com/-pBVqDE6p4IM/VGogtkpvRNI/AAAAAAAAIg8/ElOrK5DkUsM/s1600/MAP-VHAQU.gif
    http://1.bp.blogspot.com/-xpPblNvfbD8/U2o64QpFrMI/AAAAAAAAGXU/Q2-LV-pyGD4/s640/557_VH-AQU_ECA_SPLIT_04-MAY-2014_EJC_1024A.jpg

    Eg-2: F-OIAY DHC-6-300 MSN 507 Gap/Tallard, France to Noumea, New Caledonia
    http://2.bp.blogspot.com/-4ou4uYp_qto/VGpC24qs-6I/AAAAAAAAIls/B0DaTmM1vaU/s1600/MAP-F-OIAY.gif

    Eg-3: F-OIJI DHC-6-300 MSN 277 Altenrhein, Switzerland to Noumea, New Caledonia
    http://2.bp.blogspot.com/-yMAX7dlqz90/VHS2lUAO7xI/AAAAAAAAI3c/ndIss6KrmJo/s1600/MAP-FOIJI.gif
    http://1.bp.blogspot.com/-HxDeFYgnKGQ/VHS3U7_7DII/AAAAAAAAI3k/yVEs5eYeRFs/s1600/277_F-OIJI_ERICH_KUNZ_ALTENRHEIN_2009_1024.jpg

    Now, do we know if Z had the DHC-6 (or very similar) set up on his home sim ?

  791. Richard says:

    @DennisW

    You stated “The key takeaway from Figure 2 is that all quartz oscillators are characterized by a random walk in frequency at measurement intervals greater than ~100 seconds. The bounds of the quartz oscillator random walk span two orders of magnitude depending on the crystal cut and oscillator environmental control. The Inmarsat data we have is sampled at ~3600 seconds – well into the random frequency walk region of the quartz oscillator used in the AES.”

    There are no random walks starting after about 3600 seconds into the flight as you definitively and reportedly claim.

  792. DennisW says:

    @Richard

    Random walks are just that, random. Still no response on Figure 5.4 of the DSTG book? It is a bit of an “elephant in the room”.

    I would really like to see the corresponding plots for the 20 flights referenced by the DSTG. Looking at Figure 5.5 it is clear that variations up to 3 sigma and 4 sigma are present in the data. Not that I endorse the DSTG use of sigma. Nonetheless your setting a threshold for discarding routes near 1 sigma seems restrictive to me when there examples of actual flights significantly exceeding that threshold.

  793. Richard says:

    @DennisW

    Both @DrB and I have answered your point on Fig. 5.4 in the DSTG book.

    You are not listening.

  794. Sid Bennett says:

    https://www.dropbox.com/s/8d88laf7dsbnkcj/report%20draft%20for%20comment-turn%20time-r2.docx?dl=0

    Finally, after browsing my shambolic file system, I came across the summary report that Geoff and I wrote on the studies of the FMT time and azimuth. This was circulated to the IG on or about 15 March 2015, but I do not believe has had wider distribution. So, in reading the report, please remember that it was written 4 years ago.

    The data was obtained by running Barry’s model for turns originating along N571 to determine the best resulting azimuth and BTO+BFO error for each turn time (which included a specific best M for each case) and then using the most likely turn time of 18:40, which occurs at IGOGU, to determine the most probable azimuth (186.4T). Turn times after 18:40 are not compatible with a simple path.

    I accept that the model currently being used by Richard and similar models have a better model for the met data, but I can see little likelihood that the gross behavior is any different than we previously computed.

    Let the debate go on.

  795. Victor Iannello says:

    @ventus45: I’m really losing you. There is one particular reconstructed path that falls close to a Skyvector waypoint that isn’t part of the navigation database supplied to airlines. I think trying to draw conclusions from this observation is a big stretch.

  796. George Tilton says:

    @Victor – Sorry about the length of this post

    @All

    The exchange between Richard, DennisW, DrB and Sid Bennett brings to mind:

    “It was six men of Indostan, to learning much inclined, who went to see the elephant (Though all of them were blind), that each by observation, …” (from a poem by John Godfrey Saxe)

    Four years ago I found the Jeff Wise forum while sitting for three hours a week in an oncology treatment center having a cocktail of poisons pumped into my veins. When Jeff’s forum degenerated into a loony bin of conspiracy theorists I came over to Victor’s site where the original IG members, ALSM/Mike Exner, DrB and other clear headed rational individuals had migrated.

    Since retirement the only thing I have used Excel for, was to track my retirement investments and my medical expenses.

    I have downloaded Barry Martin’s, YappFF’s, DrBobby’s, one of Richard Godfrey’s and several of Victor’s.
    I am pretty damn competent with spreadsheets but I have learned something from each of the spreadsheets.

    I am not a script monkey and refuse to copy somebody else’s work without taking it apart and understanding it enough to know it is bug free and can rewrite it myself. I was paid big bucks to fix other people’s code beginning with my employer’s (Sperry Univac 1980-1990).

    DrB and Richard are virtuoso’s with Excel. I would not approach the solving problems the way they do tho…300 hours to do an run on a Laptop with an 8-core Intel chip??
    Okay…Nature abhors a vacuum…I bought my son a Gateway 386 at 25Mhz that had more horsepower than the 1100/80 main-frame that I had to do software support on for a classified government customer with a 3 letter name (take your pick…I actually did work for both of them). The Radio Shack CoCo2 and CoCo3 computers that he had started on were not enough.

    Last year I replaced the mother-board on my circa 2009, HP Compaq 6730b 2.4Ghz laptop when I started getting errors on the disk drive that were actually caused by the interface chip on the mother board.
    This also gave me the opportunity to move from Windows XP to Windows 7 and from Office 97 to Office 2003. My Excel models are no longer resource limited. I have always been forced to find the most efficient way to solve problems quickly…I don’t have the patience to wait for Excel Solver to crank through a spreadsheet model if it takes more than 10 seconds!

    I have written a number of worksheets to test different hypotheses and path scenarios posted here over the past 3 years.

    Richard has 7 parameters to define a path.
    Victor’s “Possible MH370 Paths Along Great Circles” tells us we can combine 3 of the parameters into single parameter.
    In my model the start latitude and start longitude are set by selecting an azimuth from the 19:41:03 UTC satellite sub-point to a point on the 2nd Arc.
    A Scroll Bar control in my sheet varies the azimuth between 88° and 98° which varies the start latitude between +2.5° and -2.5° in 0.001° increments.

    A second scroll bar control varies the parameter: crossing bearing at the 19:41 ping ring between 160° and 200° in 0.05° steps.

    The third parameter scroll bar varies the height-above-ellipsoid WGS84 from 0 to 40000 feet.

    For a survey I only need a great circle…LNAV modes can be determined after a dip is found in the GS residual (GSE)

    That’s it…a minimalist approach…

    I loaded DrB’s GDAS data from 2017…Excel has a 65k row limit so I had to put the HGTS, TEMP, UWND, VWND data into separate columns…

    I decoded the SU data from the unredacted logs and extracted The ACARS position reports…others may have also…

    I tested my ISAT calculations against the MH371 ACARS data then had a go at MH370.

    Voila I got a hit!

    Lat@19:41 -1.2430deg/Lon 93.5829deg
    Bearing@19:41 188.18°T
    Lat@00:19:37 -38.3291deg/Lon 87.2416deg
    ISA h 25712 feet
    Geo h 27037 feet
    GSE 0.5797 knots RMS
    BTOR 22.1 ms RMS
    BFOR 2.9942 Hz RMS

    Why was MH370 flying at 27037 feet?
    I won’t speculate…there are better qualified people here to pursue that question…

    @Richard – Look at the BTORs…
    From one blind man to another, these BTO residuals are deltas between the great circle test path and the actual path…they are clue to the LNAV mode.

    Slant Slant
    Range predicted Range
    RAES(µs)BTO(µs) BTOR(µs) residual(µs)

    122770.0 11870 0.0 0.0
    122569.7 11500 0.1 0.0
    122702.8 11737 -2.7 -1.3
    123262.1 12771 -9.4 -4.7
    124204.4 14521 -18.7 -9.3
    124695.6 15413 -23.6 -11.8
    126082.8 18008 -32.1 -16.1
    126266.1 18346 -34.1 -17.1
    126266.2 18346 -34.4 -17.2

  797. George Tilton says:

    Sorry about the formatting on the previous post…tabs were causing a problem…

  798. vodkaferret says:

    @Don Thompaon

    Thank you very much for your answer, and best of luck with your work
    on the data holidays. That angle of attack seems far more likely to yield useful results than some of the other approaches being followed and I hope it bears fruit

  799. vodkaferret says:

    @victor @Richard @DrB

    without prejudice, fully acknowledging the effort and knowledge that you guys have applied to your latest model (and with full thanks for that), understanding that 90% of a large area can mean 10% of a smaller zone and accepting that Victor at least seems to accept the possibility of a piloted ending (not to single Victor out, I’m just not sure what @Richard and @DrB believe)

    if (as seems to be the case) your current studies lead you to an area that has been searched to a 90% coverage, which do you believe to be more likely? That MH370 is in the 10%, .. or that your assumptions are flawed??

    I’m not trolling, I’m trying to understand the thought process. is it “if a glide started anywhere it most likely started here”, is it “I dont care about the scans, my math says it’s at 34 deg”..?? Because from the outside, and openly admitting I cant understand your level of maths
    ..if you finish in the 90% zone then the first thing to re-examine would be the assumptions behind the work. At some point Occam’s Razor has to come into play…

  800. Richard says:

    @George Tilton

    There was a seventh man from Indostan
    Who approached the Elephant from behind
    The Elephant thought of something quite profound
    And dumped shit on all around.

    The Elephant was a computer whizz
    And soon got everyone in a tizz
    But no soooner than he shouted Eureka
    A mouse blew him up with a bazooka.

  801. George Tilton says:

    @Richard

    I have always been impressed with your work.

    And now I see you are quite a wit…

  802. DennisW says:

    @Richard

    Both @DrB and I have answered your point on Fig. 5.4 in the DSTG book.

    You are not listening.

    That is simply not true.

  803. DrB says:

    @DennisW,

    You said: “Do you really believe the mean BFO error of any of the 20 flights was on the order of 0.20 Hz (as reported in Table 5.1)?”

    I never said that. Why do you ascribe beliefs to me that you make up out of whole cloth? DSTG never said that, either.

    The DSTG Table shows an ensemble average value near zero based on 20 flights, where the bias was adjusted on the tarmac at the beginning of each flight. DSTG never said the what the mean was for the in-flight BFO errors for any of the single flights. They only give the ensemble mean. They also did not adjust the bias so that the in-flight mean BFO error was zero. You are misunderstanding and misrepresenting their table. The expected value of the ensemble mean is zero, assuming the OCXO drift is a random walk, so I am not surprised by their result. I would expect the mean of the in-flight BFO errors for a single flight be be within +/- 3.5 Hz, using the tarmac bias, based on my estimate of +/-7 Hz maximum drift in 4.5 hours. Undoubtedly this contributes to the broadening of the ensemble PDF in DSTG’s Figure 5.5. The PDF for a single flight is narrower than the ensemble PDF and has a mean generally displaced from zero.

    That DSTG figure uses the tarmac bias. That is exactly what I am doing for MH370. I find the standard deviation about the mean (which is not zero in the case of MH370), but I do not use the mean BFO as a statistic. I do not force the mean to be any particular value. So, that BFO error ensemble PDF from prior flights shown by DSTG is a valid (and conservative) comparison for my BFO standard deviation statistic. The typical mean values I do find are consistent with the expected OCXO drift after take-off. Thus, there is no mystery here, and, while the BFO error value is only predictable within (at most) +/- 7 Hz, the dispersion about the mean is bounded by a well-defined ensemble PDF of errors, as shown by DSTG, which is broader than the PDF for a single flight. Indeed, many flights will have noticeably smaller dispersion about the in-flight mean than predicted by the ensemble average, and this is borne out by the significantly smaller dispersions I (and others, including Inmarsat) have obtained from a number of trial route fits using the MH370 data as well as BFOs from prior flights.

    The bottom line is that the ensemble PDF of BFO errors is broader than the expected PDF for a single flight. This reduces the discrimination effectiveness of BFO residuals when fitting trial routes, but still the BFO residuals are useful in that regard.

  804. DennisW says:

    @DrB

    DSTG never said the what the mean was for the in-flight BFO errors for any of the single flights. They only give the ensemble mean.

    I thought that is what I said. I also pointed out that the mean of any single flight would depart significantly from zero which is why the use of an ensemble mean as representative of the PDF of any given flight is flawed. I also never attributed any of that to you. I merely asked what you thought about it. Why are you and Richard so sensitive and quarrelsome? I am raising sincere questions not trolling. BTW, I could find nothing relative to an explanation of the DSTG book 5.4 data other than a vague speculation that it had a geographic dependence. No physical explanation.

  805. Ventus45 says:

    @Victor,

    A big stretch – perhaps – but how long is a piece of string ?

    My interest in this idea was simply sparked by the initial observation that Richard’s path crossed CC02, nothing more.

    A number of people proposing northern crash locations on the arc have floated the idea of there being a need for a “hold” somewhere.
    If a hold was needed, where would he do it. Near Cocos Island would be a sensible location to do it, iff the initial intention was an evental safe landing, presumably at YPCC, but you would do it away from active airways.

    If you look at the charts, you will see that the CCxx twotter waypoint route lies between routes T41 and M641.
    BEBIM is on T41 at 200nm from Cocos on the 300 radial.
    M641 outbound Cocos is on the 320 radial.
    CC01 is 150nm from Cocos on the 307 radial, and CC02 is 300nm also on the 307 radial.
    Cocos also has a VOR and an NDB.

    If you were going to set up a hold anywhere reasonably close to YPCC, simply select that airport in the FMC (and I am presuming that YPCC was in 9M-MRO’s database) and then, at some point after 20:41, (and perhaps “knowing” from his sim that CC001 was at 150nm on the 307 radial) simply turn off the southern track when the FMC showed that the heading towards YPCC was 127.
    Then, proceed inbound to 150nm, within range of the navaids, tune an aid, and then simply set up the hold at 150nm.
    You don’t need a waypoint in a database to do that.

    Now, how long do you hold here ?
    I got interested in the idea, when I observed that CC01 is almost bang on the 21:41 arc.

    In the image below:
    T41 is the purple airway.
    M641 is the red airway.
    Richard’s path is light brown (at left) with position at 20:41 near the 20:41 arc (green at left).
    The airway between YPCC and the Maldives containing the “CCxx” waypoints is the Light Blue line.
    https://www.dropbox.com/s/4zvj5olgxuperw5/Possible%20hold%20at%20CC01%20-%20Dropbox.png?dl=0

    The proposed possible hold is at CC01 on the intersection of the light blue airway and the green 21:41 arc.

  806. Victor Iannello says:

    @ventus45: I disagree with much of what you wrote, including the description of how a B777 pilot would likely “tune in a VOR”, as well as the likely location for a holding pattern, but we are all just hypothesizing, so I’ll just leave it at that.

  807. Viking says:

    @DennisW

    Great that you take the fight on the statistics of the oscillator drift. That is the key to understanding the problems with the BFO.

    @All

    Please do not forget that the oscillator was almost certainly cycled violently before 18:21. If that had not happened I would be willing to accept the pragmatic approach from Inmarsat to use a (normal) uncertainty of 7Hz. It is conceptually wrong but works in most practical cases.
    However, after a violent cycling the problems brought up by @DennisW increase dramatically and become dominant. Therefore it is simply wrong to use a 4Hz value, and highly questionable to stick to 7Hz.

    There are many scientific papers describing the fundamental problems involved. Take for instance a look at this Physics Today paper from 1991:
    H. Scher, M.F. Shlesinger and J.T. Bendler, Phys. Today volume 44, page 26 (1991)

  808. airlandseaman says:

    “violent cycling “? 100% Rubbish. The OCXO is designed to start cold from any temp between -55C and +55C, and achieve the nominal stable frequency within a few minutes. No problem. Nothing violent about it. Where did you get that shit?

  809. TBill says:

    @Richard
    I’d be interested in your critique of the older IG 186 deg path by Yap FF. It seems to have quite good BTO/BFO fit but I am not good at judging BTO fit.

    A few of us still use the YAPFF spreadsheet version of the path. I use it mainly as a prototype of a good path to 38 South.

    Link in reference #4 below:
    http://www.duncansteel.com/archives/1366

  810. Richard says:

    @DennisW

    You asked “Why are you so sensitive and quarrelsome?”

    Complete nonsense! I am presenting facts, which you are ignoring.

    For example, yesterday I presented an analysis of the 30 BFO values taken from the MH16 from Kuala Lumpur to Amsterdam on the evening of 7th March 2014. I compared the Inmarsat paper Fig. 15 with the DSTG book Fig. 9.9.

    I found the BFO data to be a reliable indicator of the flight path and not an inconsistent unpredictable erratic non-ergodic data set, as you are claiming.

    The standard deviation of the BFOR for this flight is 4.5474 Hz.

    I agree that BFO is a bit sensitive when the aircraft is turning or climbing, which produced the BFOR peaks ± 7 Hz observed by the DSTG, but in a stable cruise the BFOR is ± 3 Hz.

  811. Richard says:

    @DennisW

    You stated “I could find nothing relative to an explanation of the DSTG book 5.4 data other than a vague speculation that it had a geographic dependence. No physical explanation.”

    The DSTG book Fig. 5.4 shows a flight on 2nd March 2014 from Mumbai to Kuala Lumpur. You will find more details on this flight in Fig. 9.3.

    For the majority of the flight the BFORs were between 0 Hz and -10 Hz. The 16 Hz deviation limit had not been reached on the ground in Mumbai and the BTO Offset did not require re-calibration.

    At around 22:20 UTC Fig 5.4 shows the BFORs were between +10 Hz and -20 Hz. From Fig. 9.3 you can see the aircraft was turning at this time. We know that the BFO data is very sensitive to the aircraft turning, climbing or descending. We also know that the BFO can be affected by the proximity to buildings or when taxiing on the ground. In the cruise BFOR can be affected by turbulence.

    The BFOR was around 0 Hz at 18:15 UTC on the ground in Mumbai.

    The BFOR in the cruise from 19:20 UTC onwards varies between 0 Hz and -10 Hz because there were several turns and a step climb at 21:20 UTC. The aircraft started descending at 22:45 UTC and you see the variation in BFOR between 0 Hz and -10 Hz again.

    The BFOR was around -10 Hz at 23:30 UTC on the ground in Kuala Lumpur. It is possible that the aircraft was taxiing or being towed at this time.

    In my view, this is a physical explanation of the BFOR peaks.

  812. Sid Bennett says:

    @Richard
    Thank you for your data-driven comments on Fig. 5.4 of the DSTG paper.
    I feel certain that the DSTG team decided that it was too difficult to model the details of the actual flight dynamics for the additional cases. it was easier to choose a large SD since it was apparently not inconsistent with their conclusions. It is a pity that the data sets they used are not publicly available. As I understand it, some peer-reviewed scientific journals require the background data to be publicly available.

    @airsealandman
    Agreed. Who would buy a OCXO that was so erratic in its characteristics? The power to the unit is often cycled over the live of the equipment. Also, if memory serves, the oscillator error is one of the criteria for enabling the transmissions, to ensure that warm up has completed. I can’t remember the details, but I am sure you do.

  813. Viking says:

    @airlandseaman

    Cycling the oscillator to a temperature of potentially -60 degrees, and simultaneously a low pressure is not a normal treatment.

    I am convinced it will survive, and that the furnace will still be able to reach its normal temperature within only moderately extra time. However, did anyone ever measure how the offset of the oscillator behaves after this kind of treatment?

    I have personally done it to other types of oscillators, and found them trapped in metastable states for hours afterwards.

  814. airlandseaman says:

    Viking: “However, did anyone ever measure how the offset of the oscillator behaves after this kind of treatment?”

    Are you kidding? Manufacturers and independent researchers have spent decades cycling and perfecting these devices. I have personally tested OCXOs and TCXOs hundreds of times. Some devices do exhibit a very, very small offset following a deep power off thermal cycle, but describing the event as violent is grossly misleading. It is simply not true.

  815. TBill says:

    @ALSM
    Re: OXCO I think many people over-react to the depressurization/temperature scenario.

    While it is true that the air temperature can dip drastically after a rapid dpressurization, there would not be many Kilograms of cold air left in the aircraft to cool down the metal/interior components by heat transfer. So the remaining air will start to heat back up right away. The interior solid materials could slowly cool if no bleed air. Presuming the OXCO has some insulation presumably cooling would be gradual and I am not sure how cool it would get.

  816. TBill says:

    PS- I am thinking if MH370 depressured, perhaps by Penang the pressure is coming back possibly bleed air, so we coud be talking 20-min cooling down.

  817. DennisW says:

    @Richard

    The DSTG figure 5.4 data between ~22:00 and ~22:30 is difficult to reconcile with the constant speed and altitude of the flight at that time. There was a gentle turn, but apart from some work I did quite some time ago based on sensor lag, the affect of turns on BFO has not been characterized to my knowledge. I concluded the affect was quite small even for standard rate 3 degree/second turns.

    The fact that the BFO data is non-ergodic and non-stationary is not debatable. It is the reason that Dr. Allan developed the Allan variance used by every oscillator manufacturer.

  818. DrB says:

    @DennisW,

    I have made a refinement in the way I use the BFOR statistical behavior as one indicator of many in evaluating various trial routes. Previously I ignored the mean and used the standard deviation. However, there is a better way, and that is to simply use the RMS instead of the standard deviation. This method captures the both impact of the non-zero mean of a single flight (caused by potential random walk of the OCXO frequency) as well as the dispersion due to all three sources of BFO error.

    From DSTG’s Table 5.1, for the case of in-flight data with outliers excluded, the mean of the ensemble BFOR PDF is 0.18 Hz and the standard deviation is 4.32 Hz. I modeled this distribution using 500,000 random values (matching the gaussian envelope in Figure 5.5), and the resulting average value of the RMS is 4.33 Hz. Obviously, the near-zero mean value produces a RMS expected value very close to the standard deviation. Next, I took 100,000 trials of 5 random values following this same distribution and found the RMS value for each set. As one should expect, the average value of the RMS derived from 5 samples is smaller than the true RMS, being 4.12 Hz. Thus, as expected, the RMS statistic is biased low with only 5 samples. In addition, the standard deviation of the 5-point RMS is 1.33 Hz. So, the 5-point BFOR RMS has an expected value of 4.12 Hz and a standard deviation of 1.33 Hz. To assess a trial route, I find the Z-statistic where Z = (RMS – 4.11)/1.33 and then I find the probability using the normal PDF.

    Otherwise good fits of trial routes have 5-point BFOR RMS values near 3.8 Hz, within 0.22 sigma of the expected value of 4.11 Hz. For comparison, the mean is near -3.4 Hz and the standard deviation is near 2.0 Hz.

    Using the RMS BFOR statistic in my route fitter, and using a BFO bias determined from the tarmac BFOs before take-off on the same flight (as I do), is consistent with DSTG’s Figure 5.5 ensemble PDF. To the extent that random-walk excursions are contained in the BFO errors, they are characterized statistically by the ensemble PDF, and the corresponding probability may be estimated.

  819. Richard says:

    @DennisW

    Once again you are in denial, even when evidence is presented to you.

    You deliberately twist what I stated. You focus on altitude and speed, when I was NOT talking about altitude and speed at 22:20 UTC, I was taking about a significant turn.

    You deliberately try to call the turn “gentle”, as if it was of no influence on the BFO.

    You continue to cherry pick and ignore hard facts such as a standard deviation BFOR of 4.5 Hz.

    You continue to raise Allan Variance as if it had an all encompassing effect on the BFO data.

    Listen to the experts like Mike, that have been in the aircraft satellite communication business for years.

    Your experience is in mobile telecommunications. Viking’s experience is in completely different engineering fields. But you both know better than the experts. And you are both buried in your pre-conceived ideas.

  820. DennisW says:

    @Richard

    Would you please explain how turns at constant atitude affect BFO? The reason I called the turn gentle is because it was gentle (10 degrees). I have not seen anyone on this blog attempt to attribute BFO change to a turn except some work I did in September of 2015.

    I thought the DSTG reference, “structured bias”, to the errors shown in figure 5.4 was appropriate. The errors are clearly not “white”. Quote from DSTG paper below.

    The mean bias is different between flights and even within a single flight there is evidence of structured variation.

    The variations in bias shown in Fig. 5.4 happen over a timescale of minutes rather than hours. In the accident flight the available BFO values are generally at least an hour apart. This is a relatively long time compared with the correlation structure of the error, so the model does not use a coloured noise model for the BFO. However, the drift of the BFO bias means that it is not sufficient to assume that δ f bias
    k (xk ,sk ) will be the same in flight as on the tarmac before takeoff. The potential variations were incorporated by modeling the BFO bias as an unknown constant with a prior mean given by the tarmac value and a standard deviation of 25 Hz.

    Two interesting things (to me) about the quote are:

    1> The BFO error was calculated using the tarmac bias for each flight. Not the ensemble mean bias, and it is not sufficient to assume the bias is the same thoughout the flight as it was measured on the tarmac.

    2> A standard deviation of 25Hz for was used for the bias drift in the Bayesian model.

    My feedback to you is based on the well known Sagan standard “extraordinary claims require extraordinary evidence”. You are proposing searching an area that has already been searched with at best a 20% probability that the debris was missed (I doubled it from the 10% probabilty estimated by Metron to be generous). I am not endorsing a dive-glide-dive scenario. The use of a 4.5Hz BFO route rejection criterion is simply not justifiable. Even if one accepts the 4.5 Hz standard deviation bound on BFO, that implies a 32% possibility that a valid route is rejected.

    It has norhing to do with being “buried in any pre-conceived ideas”.

  821. Richard says:

    @DennisW

    You asked “Would you please explain how turns at constant altitude affect BFO?”

    I am surprised by your question.

    I will try to explain simply, but I do not want to insult your intelligence.

    If you want a more complicated answer, then please let me know.

    The BFO is the doppler shift between the velocity of a moving satellite and the velocity of a moving aircraft.

    The satellite has a velocity in the x, y and z planes, which I will call VSx, VSy and VSz.

    The aircraft has a velocity in the x, y and z planes, which I will call VAx, VAy and VAz.

    If the altitude is constant, then VAz is zero.

    If the aircraft is turning, then VAx and VAy are both changing, hence the sum of VAx-VSx, VAy-VSy, VAz-VSz is changing.

    This effects the Aircraft Compensation Doppler and Aircraft Satellite Uplink Doppler, which in turn effects the overall Doppler calculation. The Perth Satellite Downlink Doppler, EAFC Eclipse Effect are not affected.

  822. DennisW says:

    @Richard

    Of course aircraft and satellite motion affects the Doppler. I am assuming your VAx, VAy and VAz are referenced to the aircraft local tangent plane and not ECEF (otherwise VAz is not zero at constant altitude). The BFOR exists only because the AES uses a stationary satellite position over the equator and does not use actual satellite position and velocity (nor changes in aircraft altitude) to calculate the Doppler compensation.

    A small ground track change over a short time period has a neglible effect on the DIFFERENCE between actual satellite-aircraft Doppler and the Doppler compensation performed by the AES. Our calculation of descent rate based on BFO at the end of the flight does not even consider changes in aircraft heading as the plane spirals to the sea. It is assumed that the Doppler compensation takes care of that.

    The work I performed earlier considered sensor update rate (latency in the information provided to calculate Doppler compensation) during turns.

  823. Sid Bennett says:

    A thought question.

    Can a broadly stationary process convey any information in a Shannon sense?

  824. Viking says:

    @airlandseaman

    Concerning the cycling to low pressure and temperature and back, I am pretty sure manufactures test if the devices survive this. The problem is how they behave for the first hours to days following such a treatment. If the oscillator is just 5 ppb off its center frequency we are in trouble.

    Many things can cause such a deviation. A possible scenario which is often overlooked in testing is:
    (1) Slow cooling to low temperature and pressure
    (2) Relatively rapid reheating to room temperature and return to normal pressure at significant humidity
    (3) Shortly after this repower the device.
    During reheating and return to normal pressure water vapor may have been driven into the device and condensed to ice. Repowering causes a ‘lake’ of water around it. I have seen exactly that effect casing severe problems.