MH370 Family Member: Give Us the Truth

Narendran (Naren) KS, who lost his wife Chandrika on MH370, on Twitter says: Yes, give us the truth. Not too many people we know are convinced that 459 pages of the 2018 report is the sum total of all that is known regarding the disappearance of MH370. The following note accompanies his Tweet.


My family and friends in India send you their greetings.

Unlike previous years, I have been ambivalent about making this trip. I have searched hard to understand this. It is almost 5 years since I saw off my wife, Chandrika, and the last we heard from Malaysia Airlines flight MH370, the flight she boarded en route to Mongolia.

It has been a long haul dealing with loss and rebuilding a life, a task that remains incomplete. What became clear is that lately the cloud of sorrow and despair, the listlessness with life in general, and the restlessness with the MH370 search and investigation has become less intense.

What remains is the solidarity with the MH families, their loss and struggles to cope, to envision a future and reconstruct their lives. This only grows as each nuance in a shared language of loss, grief and reconstruction becomes more widely shared and understood.

5 years is a long time.

Among the MH families, the old have grown older. Some whose will to live was broken by irreconcilable loss believe they have nothing left to live for.

Among the young, some have by now moved on to pursue studies, take up jobs, moved home, find partners, have babies, … Somethings to cheer about and remind ourselves that winter is not the only season, each day isn’t always drab, and the sky isn’t only grey.

However, many among us continue to struggle while applying ourselves to the chores of daily existence. For all of us, knowing what happened to MH370 remains the key to unlock a part of our lives, our energies.

Our prayers have remained unchanged: Find the plane. Find the passengers. Give us answers to ‘what’, ‘why’, ‘how’ and if it comes to it, ‘who’.

Give us the truth.

Yes, give us the truth. Not too many people we know are convinced that 459 pages of the 2018 report is the sum total of all that is known regarding the disappearance of MH370. Those who know more but have chosen silence, if indeed there are some, will eventually die a thousand deaths each day, for guilt is a latecomer but an unforgiving squatter. It is the order of things and not what we would wish for them.

In this search for the truth, there isn’t a Malaysian truth, an Australian truth, a British truth or an American one. Or (even) an Indian and a Chinese one. There were 239 passengers from 14 countries. An international mix of nationalities. A Boeing 777, an American company’s product. The incident is believed to have occurred in the Indian Ocean, in international waters. The investigation is governed by the conventions written in by the International Civil Aviation Organization. Experts from across the world when consulted or otherwise, have weighed in with their analysis and recommendations regarding the search. It is an event that concerns, every airline, every passenger and perhaps almost every family across the world exposed to civil aviation.

To suggest as some do that it should be left to Malaysians to script the story and the end game does not cut ice. If anything, the fact that it was a Malaysian airline’s flight under Malaysian air traffic control and Malaysia’s leadership of the investigation places an unshakable burden of responsibility to the world at large….to persist, to mobilize the funds necessary, and to hunt for the credible evidence for further search that it never tires of reminding us as being a requirement. It also has responsibility to test the claims of those who proclaim new knowledge, new evidence and fresh coordinates, and offer a public, transparent well-argued refutation if indeed that is what will end misguided or false narratives.

The best tribute we can offer to those we have lost is through demonstrating the will to find credible answers, fix the issues and assure the world that more lives will not lost in future to similar incidents.

A new Malaysia under this new Government has been the source of new fledgling hope.

To those who have fallen silent travelling on MH370, we remain respectful and eternally indebted for the time they shared with us. To those who have offered silent support, those who have lent their voice and those who have sent their prayers, we remain grateful. We draw strength from your little acts of kindness, words of solidarity, and your quest for the truth – whichever corner of the earth you seek it from.

Questions have a way of persisting and even outliving you and me till satisfactorily answered. I go back to pick up the threads of my life in Chennai secure and with faith that one day we will know.

Thank you.

386 Responses to “MH370 Family Member: Give Us the Truth”

  1. DrB says:


    A crawler on Fox News just reported Malaysia MOT is now interested in re-opening the MH370 search. Ocean Infinity was also reported as being interested in resuming the search using new technology developed during the last year.

  2. DennisW says:


    Fox News story linked below along with the Malaysian fisherman’s claimed GPS location of crash.

  3. Andrew says:

    Pipped at the post!

  4. TBill says:

    I sympathize and agree with with the NoK on “give us the truth.” However, I would say ICAO and international law basically gives sovereign right to countries like Malaysia to control airline investigations. In so many words, public does not not have right to know unless the responsible country so desires. Sorry but there is a conflict of interest between what the global public wants and what a country needs to do in its own self-interest. In the case of MH370, many of us feel it may strike to heart of Malaysian politics, so it is probably not an event that Malaysia has a realistic option to be completely open about.

  5. TBill says:

    Re: drag of open window, I was not thinking the cockpit window is open for long. Maybe toss out the co-pilot cell phone, maybe put an external wire out to get reception on a Iridium sat phone.

  6. Victor Iannello says:

    The current Malaysian administration seems much more interested in finding the plane than the previous one. However, there are still the obstacles of crafting a credible proposal that is acceptable to Malaysia and the willingness of Ocean Infinity to take the financial risk.

  7. Victor Iannello says:

    @TBill: I can point to a handful of instances where important information was either officially or unofficially released, despite the initial opposition by the Malaysian government. I think there is always hope that we learn more.

  8. TBill says:

    As you probably know, both Atlas Air “black” boxes have reportedly been found as of today. Fingers crossed that the Data/Voice was recorded to show what happened.

  9. Victor Iannello says:

    @TBill: Yes, in short order, we’ll know more about 5Y3591 from both the FDR and CVR. The CVR was found first and will be easier to analyze, so perhaps they will provide some information before a more in-depth analysis is performed on the FDR data.

  10. David says:

    Since the ATSB’s photo here,

    the rear of the rib at the broken outboard end of the outboard part-flap has been removed, presumably for inspection and access, together with a variety of fasteners in that vicinity:

    Quite likely there has been further investigation of the character of the break. It would be interesting to learn of the exact reason and results. (I have not yet watched the show: possibly noted and addressed there.)

  11. David says:

    The second URL does not open as I sought.

    That photo is No.5 at:

  12. oddball says:

    [From previous thread]


    My apologies! You were correct:

    MRO was carrying just over 1200kgs of water. So you could go to the galley and jam open a tap

    @Don Thompson

    Many thanks for that info and correction.

    Would you please forgive me if I am a little obtuse? I don’t quite understand your closing paragraph, beginning:

    The present objective is “where“: best to deal with facts, simple compliance to Occam’s Razor, …

    I don’t want to fight with anyone, but I’m keen to debate and get facts clear. I’m also very keen to contribute my ideas, if they are of any value. I don’t know if I’ve said anything new so far, but I’d like to keep trying.

    I agree with you: the current objective is to figure out where.

    However, in my mind, to do that, we need as far as we can to establish the outer limits of that where, then narrow down to most likely, and to do that, I think we need to look at how, why, and I would strongly argue, who. If we can get an idea of who, we might get a better idea of why, then how and where. I know I’m hopping around those like a flea, which may be annoying! My apologies, but that’s how my mind likes to operate. When doing a jigsaw puzzle, the best way to start is usually to find the four corner pieces first.

    As Einstein reportedly said: “Imagination is more important than knowledge. For knowledge is limited, whereas imagination embraces the entire world, stimulating progress, giving birth to evolution.”

    Regardless, imagination is of little use without facts to build from.

    As Victor suggested earlier, progress seems to have slowed, but without more data and/or ideas to work with, that must be expected.

    Do you mind if I continue my hopping about all four corners?

  13. oddball says:


    maybe put an external wire out to get reception on a Iridium sat phone.

    Ahhh!!! Now we’re talking. Has anyone followed that line of thought? Is there any evidence of an Iridium sat phone pinging (or calling / being called) in the SIO at the relevant times? Or do those satellites not relay any of the relevant info? “Privacy concerns” maybe…

  14. Brian Anderson says:

    “maybe put an external wire out to get reception on a Iridium sat phone”

    Come on guys ! Connect the wire to what? Iridium satphones operate in L-Band, i.e. about 1.6GHz. An external wire out the window is not going to help. Probably degrade the signal actually.

    As for opening a window at a few hundred knots . . . get real.

  15. Ventus45 says:

    The Malaysians had no interest in the Pemba flap when it was discovered.
    Some time went by, before it was obtained by the ATSB, and subsequently shipped to Canberra.
    It is now in Malaysia.
    The French have retained the flaperon.
    When, how, and why, was the the Pemba flap, sent to Malaysia ?

  16. David says:

    Ventus45. The ATSB spokesman told me some time ago, after I raised questions about that item, that it had been despatched to Malaysia. Presumably that was when the ATSB wrapped up its work.

    ATSB debris investigations appear to have been limited to establishing provenance, whether the accompanying mollucs, barnacles, minerals etc were consistent with drift track estimates and recovery sites, and whether damage might help with the search for the underwater wreckage.

    I suspect a reason for sending it to be that any further analysis that might help with cause was up to the Malaysians as the acccident investigators.

    The French judicial system led to their retention of the flaperon as you know, though its release to Malaysia would allow a closer comparison of the flaperon/Pemba flap mutual damage.

    While the french reports on the flaperon’s condition were comprehensive, relevant information has come to light subsequently so some review is needed.

  17. oddball says:

    Further to your very helpful answers re Emergency O2 in the previous thread:

    The passenger oxygen system installed in 9M-MRO used chemical oxygen generators with a nominal duration of 22 minutes, regardless of altitude

    may I ask some further questions, please?

    Would I be correct in thinking that, once activated, the passenger oxygen generators simply run to exhaustion? They cannot be stopped, then resumed again later, as in depressurization, then restoration, and later, another depressurization?

    Do you know what releases the face masks? Any way to activate the oxygen generators without masks dropping?

    Similarly, what of the pilot’s oxygen supply? Can that be turned on and off at will, i.e. under manual control?

    I’m thinking that even for the pilot, extended flying depressurised at high altitude is going to be pretty unpleasant, if not uncertain with respect to continued consciousness. Would you agree?

  18. Tim says:

    Here is my calculation for the extension of endurance if the aircraft was 1200kgs lighter. A B777 burns 3%/hour of any weight change. So…
    1200×0.03x 7hrs=252kgs less fuel. Therefore as burning 6000kgs/hr that’s an extra 21/2 mins endurance.

    As for the O2 systems, the pax O2 once activated cannot be turned off, and if the aircraft doesn’t descend will probably not keep the passengers conscious for the full 22 minutes. It is activated automatically or by a switch in the flightdeck, there is no stop switch. The pilot’s O2 is an on demand mask, so can be stopped. It is not designed for continuous use at high altitudes either, so there is a real risk if used above 40000ft it would render the user unconscious. Hope this helps

  19. airlandseaman says:

    Re ATSB Flap analysis: In Feb 2018, I visited ATSB in Canberra. The Flap Segment had already been sent to Malaysia at that time, but I discussed the flap analysis at length with Pete Foley and the people that conducted the detailed analysis. They walked me through their process, analysis and results. Their conclusion was:

    • The right outboard flap was most likely in the retracted position at the time it separated from the wing.
    • The right flaperon was probably at, or close to, the neutral position at the time it separated from the wing.

    Note that, in both cases, they say: “…at the time it separated from the wing…”, NOT “at the time of impact”. We discussed this important distinction in the context of the theory that the flap and flaperon may have separated in flight, seconds before impact. Separation before impact is possible given the speed implied by the BFO derived descent rate. Pete was reluctant to draw any firm conclusion, but agreed it was a possibility in light of the BFO data, physical evidence and analysis. Their analysis can be found on pages 19-26 of the following document: published 2Nov2016.

  20. TBill says:

    The pilots have special pressurized O2 masks, that allow operation at much higher altitudes. Yes it could be uncomfortable and risky, but in theory it unfortunately facilitates intentional depressurization as a nefarious tactic.
    My understanding the PAX O2 mask are low pressure and essentially useless above about FL350.

  21. airlandseaman says:

    TBill: Passenger O2 masks/systems are “constant flow” 100% O2 devices. They provide a constant flow rate for ~20 minutes. Constant flow O2 systems in general are suitable for continuous use up to ~20,000 feet. In an emergency at FL400, the passenger masks provide adequate supplementary O2 to allow for a rapid descent (~5 min) to a safe flight level, typically below 14,000 feet. The Pressure Demand masks/systems available to the cockpit crew will work up to FL450 (some are only rated to FL400) continuously.

    Personal experience: Because nearly all sailplanes are un-pressurized, those that are flown to altitudes above 25,000 feet are typically fitted with Pressure Demand systems. I had A14 systems in my Standard Libelle, Twin Astir and Ventus-B sailplanes. During wave flights (one to 39,500 feet), I always switched to 100% and pressure mode above ~30,000 feet. I did not find it particularly difficult to breath even over longer wave flights. (-70C OAT is a bigger issue!) It feels a bit odd having air forced into your lungs, but it would not a problem to fly with a mask on for an hour or two. Maintaining the straps tight enough to form an airtight seal is where most of the discomfort comes from, not the pressure.

    We do not know for if 9M-MRO was depressurized. However, we do know:
    1. depressurization for the period circa 17:21-18:22 would be consistent with other facts and observations,
    2. the PF could easily have flown for that length of time using the Pressure Demand system on board,
    3. and the passengers and crew could not have survived on constant flow O2 alone at 40,000 feet for an hour.

  22. oddball says:

    @airlandseaman, @TBill & Tim,

    Many thanks to you all for your very helpful answers. Exactly what I was after.

    Yes, thanks, I understand we don’t know that this happened, but it’s part of the hypothesis I’m considering.

    In line with Occam’s Razor and KISS principles, this confirms there would be no need to try to surreptitiously bleed off the emergency cabin oxygen supply, as the pilot could easily sit this out without needing to descend, (which would be counter to his intentions, in my model).

    This leads to my next questions, if anyone can help, please.

    1. Is there any credible / reliable evidence of any changes in altitude prior to exiting radar coverage? I know the media raved about all kinds of ascents and descents but I believe this was later discredited… Where did that end up?

    2. Could someone please help me understand the relationship between bleed air supply and cabin pressurization? If bleed air is turned off, will the cabin inevitably be depressurized? Or can the outlets (& inlets?) be closed to maintain cabin pressure? I guess there is no way to maintain cockpit pressurization independently. Presumably, it’s only the cockpit side windows that can be opened (which would not be helpful!)

    3. Does anyone have any ideas how far a skilled pilot could manually glide the plane following flame out, versus how far it would go on AP as per Victor’s earlier description, from say 40,000′?

    4. Does anyone know if ZAS was fanatical about physical fitness, like he was about other things? Did he keep himself extra fit, does anyone know?

  23. airlandseaman says:

    oddball: Re #3…given all other conditions equal, the distances would be the same from any altitude.

    Reminder: There is ZERO evidence the plane glided, but several lines of evidence it did not.

  24. David says:

    @ALSM. Was it the ATSB that removed the rear end of that rib?
    They say that besides the damage that led to their retracted conclusion (I prefer the conclusion that they were retracted when they collided to retracted on separation), “Numerous other discrete areas of flap damage were analysed.”
    Do you know what those analyses were?

    If the Malaysians removed it there is a report out there somewhere.

  25. airlandseaman says:

    The following flap photo was taken in the ATSB Lab. This should answer the question:

  26. David says:

    Thanks but that is the other, seal pan, (inboard) end

  27. oddball says:

    @ALSM, Many thanks. Interesting. Boeing designed the Autopilot that way, I guess? So the idea it would perform phugoids is eliminated? Uhhh… Isn’t that still gliding (i.e. not performing phugoids) and descending slowly without power?

    Less need for the pilot to “hang about” for the whole trip then. But I still think he would want to, to execute the “Final Act”. Of course I’m not saying it glided into the ocean… Is that what you are referring to?

    What evidence is there (could there be, even) that it didn’t glide during the early final minutes?

  28. airlandseaman says:

    0019 BFO, no IFE logon, debris size & origin…

  29. Mick Gilbert says:


    Re: We do not know for if 9M-MRO was depressurized. However, we do know:
    1. depressurization for the period circa 17:21-18:22 would be consistent with other facts and observations,
    2. the PF could easily have flown for that length of time using the Pressure Demand system on board,
    3. and the passengers and crew could not have survived on constant flow O2 alone at 40,000 feet for an hour.

    Mike, point 2 is highly debatable particularly if you assume that the Captain was the PF.

    You’ve only got to look at the Connie [Kalitta] Flight 861 incident (NTSB Identification: NYC94LA062) to understand that even when using the crew oxygen system the effects of depressurisation vary considerably. In that particular instance you had a crew member of similar age to the Captain become incapacitated before the cabin altitude had even reached 35,000 feet, leave alone 40,000 feet.

    Based on the Captain’s December 2013 Medical Examination Report we know that he was overweight, had Stage 1 hypertension, had a resting heart rate that placed him in the ‘Poor’ fitness range and that he was a smoker. He also had a family history of heart disease and high blood pressure. They are all risk factors for hypoxia and evolved gas decompression sickness.

    Going back to your first point, what other facts and observations are consistent with the aircraft being depressurised?

  30. airlandseaman says:

    odball: Re ” Boeing designed the Autopilot that way, I guess? So the idea it would perform phugoids is eliminated? Uhhh… Isn’t that still gliding (i.e. not performing phugoids) and descending slowly without power?”

    I have no idea what you are talking about. I understood your question to be: Could an autopilot control the aircraft the same way a pilot could, to descend at best L/D speed (max glide range), after FE, in both cases assuming electrical power was available, and obviously in the first case, with the autopilot engaged. Under those conditions, the descent could be comparable. Neither of those cases would produce phugoids. Phugoids can and often do occur when the autopilot is NOT engaged AND no one is flying manually. Does that clarify?

  31. Ventus45 says:

    In your photo linked above, you might also notice, that green oxygen cylinder, on the top shelf……..
    It presumably is “a left over trophy” from tests conducted during the investigation of the pax O2 cylinder that failed on QF30 and shows evidence of a distinctive “incipient failure”.
    The crew O2 cylinder in the MEC is still a “possible” cause IMO.

  32. DennisW says:


    Depressurization is not consistent with the accuracy of the BFO data at 18:25. It would be a lucky happenstance if depressurization occurred.

  33. airlandseaman says:

    Mick Gilbert: Sorry, but you are wrong about #2. My statement is correct. It is not “highly debatable”. It is highly likely anyone in reasonably good health could go an hour using a Pressure Demand mask, properly fit. Both MH370 pilots were in reasonably good physical condition.

    You neglect to point out that the captain of Flight 861 was the only crew member that suffered any problems. The others had no problems with their Pressure Demand masks. The Captain’s physical condition may have been a contributing issue, but he may also have not sealed his mask properly (likely). Whatever the reason for the 861 Captain’s problem, it is a rare case, not the norm, and not relevant to MH370.

  34. airlandseaman says:

    Dennis: Hope your recovery is proceeding well and you are feeling better.

    I do not understand your comment: “Depressurization is not consistent with the accuracy of the BFO data at 18:25.” Please explain. There is no relationship between the BFO accuracy and the cabin pressure as far as I know. Regardless, if there was deliberate depressurization circa 17:21, presumably MH370 would have been re-pressurized circa 18:20.

  35. DennisW says:


    My assumption is that depressurization would affect the temperature in the aircraft. A single oven oscillator (again my opinion based on experience) is that a significnat offset would occur. We have been over this ground before. We just have divergent views. Likewise with a depowering of the AES. It won’t be settled until the FDR is recovered.

  36. Ventus45 says:

    In the MSM news here, it was reported, that OI’s Oliver Plunket had sent a video to the NOK at the 5 year memorial, in which he apparently offered to return to the search, but there were no details. Has anyone here seen it, and / or got a link ?

  37. airlandseaman says:

    Ventus45: Video of the IG and OI presentations can be found in the following link:

    IG presentation at -2:03
    OI presentation at -1:18

  38. oddball says:

    Sorry, don’t follow:

    I understand that the thinking is that the 00:19 handshake suggests a high speed descent, but IFE Logon? I guess you mean In-Flight Entertainment system? I don’t understand the significance. Why would it try to logon at that time?

    Also not sure which debris you mean – flaperon & flap? Origin being 9M-MRO, or are you referring to the Drift Analysis PoO indications?

    Can anyone tell me where the incoming RAT supply connects to, please? Across the batteries?

    Is it true that the RAT is released automatically? On what condition?

    I’m guessing the batteries continue to power the avionics etc after flameout, until the RAT kicks in. Is that correct, and if so, what would the (relevant) flameout trigger? Similarly, what would RAT “Power Good” trigger?

    Could it be that the 00:19 handshake was triggered / generated as a result of actions relating to AP disconnect and/or the pilot taking control? That perhaps occurring upon the gliding descent having reached the pilot’s “Trigger Altitude”?

    Are you thinking said pilot didn’t bother with gliding and took over right after flameout? Wouldn’t a glide down, say 30,000 feet, take longer than 8 minutes? Sully got 90 seconds from a slow climb to 3,000 ft in an A320.

    If nose vertical from >= FL250, how many more surfaces would have parted company before sealevel? Would the windshields even take it? ==> B777-B (Balloon model)

  39. airlandseaman says:


    The IFE system normally establishes virtual channels with the GES ~90 seconds after the AES logs on. That did not happen after the 0019 AES logon. Either (1) the plane hit the water by 0021, or (2) it was in such an unusual attitude that the AES antenna could not be pointed to the s/c at 0021. Take your pick. The implications are the same.

    The origin of the debris on the aircraft includes pieces from the interior (seat back display, floor panel, several others), which indicate a very high energy impact, not a “water landing”.

    You will need to do your homework on the rest. It has all been discussed in detail.

  40. Ventus45 says:

    Thanks for the link.
    The file is 705Mb and nearly 3 hours long.
    Downloading it, but it is dead slow !
    I only have 25Mb so far, and my computer is showing two hours to go !
    May as well go and do the shopping !

  41. airlandseaman says:

    Ventus45: You should be able to stream it and FF to the points of interest.

  42. airlandseaman says:

    Dennis: OK. OCXO Temperature coefficient, not the pressure.

    First, the OCXO has a double oven, not a single. And as previously posted here:

    …the measured tempco is about -0.3 Hz/°C. An ambient temperature reduction from +30C to -30C only causes a change in BFO bias of ~-20 Hz. That is significant for path reconstruction, but not for the two primary conclusions that the plane went south (not north) and the descent rate at 0019 was very high. I agree that, if the cabin temp was lowered by say 60°, it would likely lower the transmit frequency at L band by about 20 Hz.

  43. DennisW says:


    We are on the same page.

  44. Andrew says:


    RE: ‘It is not “highly debatable”. It is highly likely anyone in reasonably good health could go an hour using a Pressure Demand mask, properly fit. Both MH370 pilots were in reasonably good physical condition.”

    I’d say that really depends on the altitude. Positive pressure breathing (PPB) becomes increasingly difficult above FL400, because the system keeps increasing the pressure differential to ensure sufficient oxygen reaches the circulatory system. Breathing becomes increasingly difficult (and unnatural) against the higher pressure differential, and there comes a point where physiological problems start to occur, particularly in pilots who are not specifically trained in PPB techniques. The following is a quote from RM Harding, Pressure changes and hypoxia in aviation:

    “The cardiovascular effects of PPB include peripheral pooling, impaired venous return, and reduced central blood volume; if there is a loss of peripheral arteriolar tone, then tachycardia and a gradual fall in blood pressure will lead to a collapse resembling a simple vasovagal syncope.”

    The crew oxygen masks found on the B777 are typically rated to FL450, but they are only intended as emergency “get-me-down” equipment, not for continuous use at very high altitude. The aircraft itself is only certified to FL431.

  45. Andrew says:


    RE: “Could someone please help me understand the relationship between bleed air supply and cabin pressurization? If bleed air is turned off, will the cabin inevitably be depressurized? Or can the outlets (& inlets?) be closed to maintain cabin pressure? I guess there is no way to maintain cockpit pressurization independently. Presumably, it’s only the cockpit side windows that can be opened (which would not be helpful!)”

    If the bleed air is selected off, the pressurisation system closes the two outflow valves in an attempt to maintain pressurisation. However, there is still some leakage and the cockpit/cabin will slowly depressurise. The cockpit cannot be pressurised independently of the cabin. Only two cockpit windows can be opened; one either side.

    “Can anyone tell me where the incoming RAT supply connects to, please? Across the batteries?”

    The RAT supplies the standby electrical system, which includes the main battery.

    “Is it true that the RAT is released automatically? On what condition?”

    Yes, the RAT is automatically deployed under the following conditions:

    – both engines fail and centre hydraulic system pressure is low, or
    – both AC transfer buses are unpowered, or
    – all three hydraulic system pressures are low.

  46. oddball says:


    Sorry for the crossed posts – an interrupted, delayed reply, without refreshing the thread.

    Anyway, many thanks for the clarification. Yes, you understood almost exactly. I was surprised that an Autopiloted, unpowered descent and a well piloted glide would (could?) produce the same range, but guessed that Boeing must have designed for that, and it sounds like that’s true. And why not? It’s good design. And thanks also for clarifying about the phugoids. That makes perfect sense too. Like one of those square-nosed paper darts, it’s just aerodynamics.


    The IFE system normally establishes virtual channels

    Sorry, but I think your assumption there is unjustified, and your choice of just two options too narrow, even if your conclusion may happen to be correct.

    My reasoning: what happens “normally” is probably not relevant here. This is anything but a normal situation. If my hypothesis is anywhere near correct, the pilot switched off the IFE system hours earlier, to conserve fuel. I mean, if the passengers are all dead, why not? So, very near the end of a flight into oblivion, why would he turn it back on?

    Besides, if the plane is out of fuel and all the engines have flamed out, I’m sure there won’t be power to it anyway.

    Anyway, I have taken your suggestion to heart, and done some more reading on older threads.

    I noted quite a bit of discussion in earlier threads about certain AES logons being different from others, and from normal. This can probably be explained by certain items, such as the IFE, being unpowered. Most often, with properly written software, when something expected isn’t appearing at the other end of the wire like it should, the program starts a time-out loop, to give whatever it is expecting to hear from, time to finish booting up (or get turned on, or whatever). While it might normally already be on, booted up and ready, and so handshake immediately, the software doesn’t usually “learn” this, so it waits, causing other processes to be delayed. And, if it isn’t on, or is otherwise dead, disconnected or whatever, eventually the software will (ok, should) give up waiting and move on. However, depending on what the missing device is supposed to do, other things down the track may also fall over or go off the rails.

    It’s at times like this that software bugs tend to rear their ugly heads. Typically, when everything is working normally, it works normally. But when things are unexpectedly unavailable, the software sometimes falls into a hole (though that may be entirely irrelevant here). There may be no bug, just the effect of device timeouts, with the software handling things correctly, or at least, as best it can. Airbus (in particular, I believe) has a history of poorly written software causing inappropriate actions under unusual conditions, so it’s not like this has never happened before.

    In general, these bugs can be either unknown or unrecognised (due to inadequate testing or lack of thought during design or implementation), or known but not corrected due to time and/or budgetary constraints. An IFE, being a non-critical device, is more likely to have such software bugs, but if it’s off, it’s the AES’s job to figure that out and adjust itself accordingly.

    That said, some of the incorrect, or undecipherable data observed in the satellite data may have occurred because an expected device was off-line. I have no idea what cables, connections or protocols apply between the AES and the IFE or other devices, but it is possible that poor design has allowed unpowered data cables to pick up electrical noise from who knows where, and that has been received by the AES.

    I also note in the data that the Flight Number is recorded as MH0000 somewhere. This is probably because the device (possibly the IFE??) was off so didn’t send anything, and the AES filled it in with zeroes. This is acceptable programming, especially if there is no such flight number (as I’m sure there isn’t) and would indicate to those who know what they are looking at that the IFE (or whatever device it was) was off-line.

    I’m sure the satellite logs contain lots of clues like this, if only one can find out where all the data items come from and what they mean.

    Hope this is helpful.

  47. airlandseaman says:

    Andrew: I agree it depends on the altitude. It does become increasingly difficult to last an hour above FL400. But would you not agree that there exists an altitude between say FL350 and FL400 where a pilot could go 30-60 minutes with the plane depressurized, killing everyone else on-board, while breathing from a Pressure Demand mask? We don’t know the exact altitude over that hour, other than the estimate at KB, which now looks like about FL380 (40,200 ft). Do you have any doubt that a depressurization scenario is feasible at FL380?

  48. TBill says:

    Sounds like you may have revised the altitude estimates at KB down a bit to FL380?

  49. airlandseaman says:

    Tbill: Yes, same analysis but new KB PSR data files became available in Jan and Feb 2019. The new data is from the same radar, but it has accurate millisecond level time stamps and raw az & range data. The observations are essentially the same, but they are more accurate, especially near the CoS (less processing). The new data produces essentially the same family of speed profiles, but the CoS average speeds changed slightly, resulting in a slightly lower estimated altitude.

  50. Ventus45 says:

    Returned from shopping, full video obtained, and segments IG, OI and Minister viewed.

    Oliver was very upbeat on the difficulty of finding a submarine, by definition designed to be difficult to find by sonar, and their success in doing so, as being proof of OI’s capability (as if any was needed). He then effectively landed the proposed prospect of a new search, fair and square, in the Minister’s lap, who was sitting front row right, and on camera. Will be interested to see what comes of that.

    Re the new radar KB data in your 12:56 post. I am perplexed. The previous data was presented to us at the time as the raw data. Now this new data relegates the previous data to “processed” data, does it not ? Why was it “processed”, in the first place, and by whom ? What is the providence of this new data ? Are the files available ?

  51. Andrew says:


    RE: “But would you not agree that there exists an altitude between say FL350 and FL400 where a pilot could go 30-60 minutes with the plane depressurized…while breathing from a Pressure Demand mask?”

    It’s difficult to say with any certainty. A young, fit individual probably could, but ZS was 52 years old, overweight (based on his BMI), was a smoker and had blood pressure that was borderline acceptable for a pilot. They are all significant risk factors for hypoxia and decompression sickness (DCS), as Mick said. DCS becomes increasingly likely above FL250, especially with coexisting hypoxia.

  52. Ventus45 says:

    What is the “provenance” of this new data ?
    Are the files available ?

  53. oddball says:


    i>If the bleed air is selected off, the pressurisation system closes the two outflow valves in an attempt to maintain pressurisation. However, there is still some leakage and the cockpit/cabin will slowly depressurise.

    Excellent information, thank you. 😉 The plot thickens.

    Of course, always another question to follow on…

    Is it a big deal to turn the bleed air on and off at intervals to maintain acceptable pressurisation, and does it take long to repressurise the cabin, do you know?

    There is a low cabin pressure [high altitude] alerting system, isn’t there?

  54. Don Thompson says:


    I was surprised that an Autopiloted, unpowered descent…

    Unpowered means no autopilot and the PFCs operate in secondary mode.

    The IFE system normally establishes virtual channels […] I think your assumption there is unjustified […] what happens “normally” is probably not relevant here.

    In the context of the AES performance at 00:19, the behaviour is entirely normal: it was repowered (the Previous Satellite ID parameter it transmitted in the Log On Request is consistent with prior loss of power), the AES received & processed ADIRU originated position and attitude data so as to operate the HGA, the GES Log On completed successfully. A L Main AC Bus power loss prior to 00:19 is consistent with multiple engine fuel exhaustion and auto-start of the APU which runs for a limited period to supply power to the Main AC busses.

    … the pilot switched off the IFE system hours earlier, to conserve fuel.

    Selecting the IFE/PASS switch to OFF does not impact the PIIC & EPESC components of the IFE system. The PIIC and EPESC are involved in datalink comms and integration with other aircraft systems such as the CSS. The PIIC and EPESC are located in the MEC, not the IFE equipment centre in the cabin. The Log On session initiated at 18:22 shows that the expected two IFE switched virtual circuits were established over SATCOM logical channels. Those SVCs were not cleared prior to 00:19. Had the IFE system ceased low-level status exchanges with the SDU I expect the SDU would have cleared the logical channels. The two X.121 destination addresses defined for the IFE switched virtual circuits are consistent throughout all the Log Ons recorded in the log.

    Replying to your @all comments…

    …undecipherable data observed in the satellite data […] Flight Number is recorded as MH0000 somewhere […] the satellite logs contain lots of clues

    You’re off in the realms of uninformed speculation, take care not to misrepresent that speculation as anything else. Previous conversations here will inform you.

    The entire satcom log is understood.

    It’s necessary for the Flight ID, MAS370 signalled as part of the Log On to the GES, to be communicated to the SDU. It isn’t signalled by the IFE, it originates from the FMC, normally communicated to the SDU via functions of the Data Comms Manager.

    It is possible that the Flight ID field had been cleared on the FMC screen.

    During the 18:22 Log On no ACARS correspondence was exchanged over the SATCOM link it is assumed that, sometime prior to the Log On restore at 18:22, SATCOM was deselected as an available datalink for the Data Comms Manager (DCM is not related to the IFE datacomms operation).

    With SATCOM deslected it is also possible that no session related configuration data was communicated via DCM to the SDU, including the Flight ID.

  55. Victor Iannello says:

    An impact more than 25 NM from the 7th arc requires pilot inputs for the dive-glide-dive sequence and the APU to run out of fuel before the expected IFE log-on (which some here believe is possible). I haven’t been a fan of this scenario, but I also thought the plane would have been found already.

  56. Mick Gilbert says:


    Mike, I am confident that the readers of this forum are intelligent enough to form their own view as to whether the case of the very rapid incapacitation of a 51 year old, obese smoker while on crew oxygen on a flight at a moderately high altitude in an unpressurized airplane is relevant to the likely outcome for a 53 year old, overweight smoker while on crew oxygen on an hour long flight at a very high altitude in an unpressurized airplane.

    Peter Foley of the ATSB thought that the Kalitta incident was relevant enough to offer it in evidence to a Senate Budget Estimates hearing last May. As he said,

    Most of the people sitting in their armchairs looking at MH370 are saying that the aircraft was depressurised for up to an hour. The pilot of this particular aircraft [Kalitta 861] was 51 years old and overweight, and, if you look at the pilot in command of MH370, he was 53 and overweight.

  57. TBill says:

    You know that Kalitta flight it was the smoker old pilot was the one who stayed awake. That flight took off a few miles from here in Manassas where I fly all the time (on flight sim). Was he on O2? High altitude? No thinking so.

  58. airlandseaman says:

    Re Pressure Demand breathing endurance: I think this thread is off track due to a hidden assumption. It seems Andrew and Mick assume Z is the only possible PF. We do not know the PF. I make no assumption about the PF in the context of Pressure Demand use endurance. I am simply saying that it is possible that the PF, whoever it was, could have depressurized the plane and flown for 30-60 minutes at an altitude high enough to kill all the other people on board. I have not heard any valid arguments that contradict that statement.

    If one is inclined to believe Z was in such poor health that he could not have been the PF in this scenario, then I suggest in that case, the PF might have been someone else. Personally, I do not buy the theory that Z was in such bad health that he could not have done this.

    And from in person conversations with Foley, I know he agrees that the plane was likely depressurized, not withstanding his public comments. I’ll leave it there.

  59. Andrew says:


    RE: ”Is it a big deal to turn the bleed air on and off at intervals to maintain acceptable pressurisation, and does it take long to repressurise the cabin, do you know?”

    It’s no big deal to turn the bleed air on and off (or turn the a/c packs on and off) to keep the cabin pressurised, but what would be the point?

    It shouldn’t take too long to re-pressurise the cabin.

    ”There is a low cabin pressure [high altitude] alerting system, isn’t there?”


  60. airlandseaman says:

    RE the Jan 2019 KB PSR data, I can summarize the difference with the April 2018 data this way:
    1. The paths are essentially the same
    2. The recorded observations are nearly the same over the same time period.
    3. The absolute UTC timestamps differ by about 1 second
    4. The range numbers are nearly identical, except close to the COS, where they differ by up to ~0.3nm (new data values are less)
    5. The CoS time gap is 272 milliseconds less in the new data

    4 and 5 above lead to the slight reduction in estimated altitude at KB.

    It is believed that the April 2018 and Jan 2019 data sets both originated from the same radar head. However, the new data has true “observation timestamps”, where the April 2018 data had only approximate timestamps added by the Malaysian radar data communications network. Recall, we had to reconstruct the estimated “observation times” from these approximate network times by fitting the observations to a clock derived from the known antenna rotation speed and phase (az). In addition, it is believed that the April 2018 range and az values were probably not discrete independent observations. They appear to have been values derived by a process down stream from the radar head, “smoothed” for ATC displays. This explains why the values are essentially identical for longer ranges, but show a slight difference close to the radar head where the azimuth is changing rapidly.

    The new data has not been been made public at the request of the source. However, a redacted file including all the basic data (Time, Az, Range) may be made available in the near future.

  61. oddball says:


    Thanks again for your most helpful answers!

    No time for a lengthy reply, but another couple of quick questions, please.

    1. Does the APU have its own, separate fuel supply, or does it draw from the main tanks?

    2. Can it be disabled / controlled so as not to Autostart when the power fails?

    Will answer your question later today.


  62. Andrew says:


    RE: “If one is inclined to believe Z was in such poor health that he could not have been the PF in this scenario, then I suggest in that case, the PF might have been someone else. Personally, I do not buy the theory that Z was in such bad health that he could not have done this.”

    Isn’t there still the possibility that Z did do it, but became incapacitated at some point?

  63. Andrew says:


    RE: “1. Does the APU have its own, separate fuel supply, or does it draw from the main tanks?”

    The APU draws fuel from the left engine feed manifold. It does not have a separate supply.

    “. Can it be disabled / controlled so as not to Autostart when the power fails?”

    The autostart sequence can’t be disabled from the cockpit, but it can be interrupted by selecting the APU selector on the overhead panel to ON and then OFF.

  64. Paul Smithson says:

    @ALSM. Interesting to hear of the updated radar dataset. I don’t know if you are at liberty to share the data or your derived speed/altitude curves analogous to the ones previously posted? Considering the method by which we had corrected timestamps and smoothed speed estimates, I’m puzzled that the new result should be any different even with more accurate timestamps. Can you elaborate?

  65. Mick Gilbert says:


    Bill, you’re thinking of Kalitta flight KFS-66, a cargo flight flying from Manassas, Virginia to Ypsilanti Airport, Michigan back in 2008. That was a Learjet 25. In that case neither the PIC or FO were on oxygen; it was a hypoxic flight incident subsequent to a failure to pressurise. Astoundingly, there was no NTSB investigation of KFS-66 so the ages of the PIC and FO aren’t known.

    Connie [Kalitta] Flight 861 was a 1994 incident involving a DC-8 flying from Ypsilanti back to Atlanta. It is one of the very few (perhaps the only) documented instances of an unpressurized flight continuing to fly on at high altitude (they reached FL330) rather than initiating a rapid descent. The fact that the PIC on 861 was of a similar age and physical condition to the PIC on MH370 is at the very least interesting, if not instructive.

  66. Andrew says:

    NTSB press release re Atlas Air Flight 3591:

    The CVR shows that crew communications during the last 18 seconds of flight were consistent with a loss of control.

  67. Brian Anderson says:

    Just happened to see copies of Ean Higgins latest book on MH370 in a local bookshop today. Had a quick read of a few pages. Oh dear! Yet another collection of mis-information, conjecture and speculation.

    I am surprised that these authors can’t even get basic facts correct. He continues to present the view that “pings” originate from the aircraft, and in particular can originate from the RR engines. Doesn’t mention ACARS in this context because he can’t distinguish these messages from the BTO measurements collected by Inmarsat. Seems that he is confused about what the BTO numbers represent. Tries to explain but explanation is very deficient and misleading. Conflates the BTO derived information with BFO numbers and offers a confusing explanation as to how they might be used.

    Nothing new here, and hardly worth reading.

  68. haxi says:

    So Jeff Wise released his new book days before the 5th anniversary? Hmm…

  69. oddball says:


    Brilliant! Thank you once again for your help.

    To answer your question for a change: It’s no big deal … to keep the cabin pressurised, but what would be the point? A couple of points:

    1: On, the boost cabin air pressure to levels sufficient for (the pilot’s) continued consciousness;
    2: Off, to save fuel.

    As TBill said earlier (previous thread):
    Cutting cabin bleed air allows 2-3% more range

    I don’t know if you agree with those figures, but based on that, if he can keep bleed air off for 95% of the time, he can increase his range by ~1.9-2.8%. I don’t know what range benefit he could gain by cutting off all the cabin power supplies, A/C and IFE, but I’m sure it all adds up. Of course, he’d want to make sure the APU stayed off too.

    On a different subject, I was disappointed (but not really surprised) to hear that ZAS was not really a very healthy specimen. But it fits.

    Today was an interruption to my train… I’ll expound more a bit later.


  70. oddball says:

    1a: … and stave off DCS.

  71. Victor Iannello says:

    @Brian Anderson: Thanks for the review. You’ve persuaded me to pass on buying the book.

  72. Victor Iannello says:

    @haxi: For amusement, I purchased and read Jeff Wise’s book. For readers here, there is nothing new. It’s his same old Kazakhstan/BFO spoof theory that has failed to gain traction. His account of how certain events came to pass is both entertaining and self-aggrandizing.

    In the “Speculative Scenario” he saves for the appendix (his title, not mine), he describes the trek to Kazakhstan, ending with an automated landing by unskilled pilots at the Yubileyniy runway at Baikonur. He acknowledges that at 00:11, fuel reserves were “very, very low”, but adds that Yubileyniy is “nearly within gliding range”. Wow. Obviously, he believes that his readers are both uninformed and gullible.

    What is not amusing is his insinuation that the recovered debris was planted, and Blaine Gibson is a Russian agent that was complicit in this activity. He devotes an entire chapter to Blaine’s business dealings in Russia, and ends the chapter by saying does not know if Russia planted the debris, but the man that found most of the debris has ties to Russia.

    The recovered debris represent a tremendous blow to Jeff Wise’s Kazakhstan theory, so he spends considerable effort attempting to discredit this evidence. As Blaine discovered so much of the debris, bolstering the Kazakhstan theory comes at the expense of unfair treatment of the man that is singly responsible for organizing the efforts that have led to the discovery of much of the debris. This paints Wise as an opportunist that has written a sensational book to make a buck.

  73. TBill says:

    OK Thank you…I was not aware the other Kalitta flight incident, I’ll take a read on that one. Unrelated to flying, I saw a recent documentary about body acclimation (to altitude sickness). Interestingly it appears the human body acclimates over time to high altitude, and that capability seems to get soft-coded into you DNA, such that your children can inherit resistance to altitude sickness.

  74. DennisW says:


    Debris planting is only part of the disconnect associated with the Wise theory. Making the planted debris appear to have damage consistent with a water entry is also not trivial. Taking pieces from an aircraft and modifying them in a way to resemble crash damage is virtually impossible.

  75. haxi says:


    Oh my, he actually called Gibson a “Russian agent” in the book? That’s not decent.

  76. Victor Iannello says:

    @Dennis: The Kazakhstan theory hinges on an omnipotent, omniscient, malicious actor with numerous accomplices. If you believe this is true, it is impossible to refute the theory because whatever practical objections you raise are countered with the extraordinary capabilities and knowledge of the bad actor. In this context, the issues you raise are just small details that can easily be overcome. But despite the capabilities of this bad actor, it left small clues that only Jeff Wise was astute enough to detect and pursue. The whole scenario quickly becomes absurd if examined with critical eyes.

  77. Victor Iannello says:

    @haxi: Jeff Wise prefers innuendo and implication rather than direct accusation.

  78. TBill says:

    Re: Atlas Air CVR
    Obviously the NTSB open communications are fantastic, if you are a concerned citizen, you get briefed on what they know on a timely basis. But it does not sound like the CVR is giving them the clear cause yet. My understanding is their techique is to establish an initial hypothesis and then establish facts to prove or disprove. So we are not hearing the hypothesis yet, unless it is just loss of control.

  79. Don Thompson says:

    @TBill wrote “My understanding is their techique is to establish an initial hypothesis and then establish facts to prove or disprove.

    The process is to first gather facts, the investigative groups each prepare a factual report, if probable cause can be determined it will be described in the summary report.

    You are correct to mentioned the open communications of the NTSB, it’s detailed here: NTSB Investigative Process.

    Hypotheses may be necessary only after the facts are gathered.

  80. Andrew says:


    I agree insofar as the CVR does not necessarily provide the investigators with a cause for an accident. However, in this case it seems to have given them valuable information about the context, ie loss of control. The investigators will try to determine why that occurred by examining the FDR data, debris, aircraft maintenance records, etc. My guess is they will be particularly interested in recovering the tail components such as the horizontal stabiliser, the elevators and their actuators. Finding those components might prove very difficult, given the nature of the impact.

  81. Mick Gilbert says:


    Re: Interestingly it appears the human body acclimates over time to high altitude …

    Yes, something like 2 per cent of the world’s population live above 9,000 feet with the highest permanent settlement up around something like 15,000 feet. Back when I worked for Rio Tinto I visited a JV copper mine at Escondida in Chile. It’s up around 10,000 feet. The locals have no problems whereas quite a few visitors do not fare well up there.

  82. airlandseaman says:

    Re 2019 KB PSR data: Publication of the new KB PSR Data in a redacted form was approved by the source. The following paper provides some background and updated results. A link to the file containing the 2018 and 2019 data can be found at the end of the paper. The original files contained a large amount of other data unrelated to MH370. Nothing relevant to the MH370 KB PSR data of interest has been omitted or changed.

  83. airlandseaman says:

    In the #5Y3591 Sheriff video, the descent appears to be nearly vertical. However, the actual descent angle was ~49 degrees below the horizon. This chart explains the context:

  84. oddball says:

    Hi again Andrew,

    Would you mind me asking for some more information, please?

    If not:

    What Emergency Oxygen supplies were available to the Cabin Crew on MH370? Was there a fixed (built-in) supply, as well as portable oxygen bottles?

    In the Helios Airways Flight 522 crash,
    one of the cabin crew was still alive and conscious about 2 1/2 hours after the flight crew apparently succumbed to hypoxia, apparently surviving on one or more portable oxygen bottles.

    How long would the cabin crew oxygen last in terms of total person hours, please? Could you also break it down by supplies, please? I believe there was a cabin crew of 10, all Malaysian.

    As I see it, ‘dealing with’ the passengers would have been no problem at all, but the other pilot, and the cabin crew are a different story altogether. To my mind, if ZAS was able to pull this off single-handedly, is was absolutely masterful, if not miraculous. Would you agree, or can you see an easy way to also knock out the cabin crew?

  85. Paul Smithson says:

    @ALSM. Thank you for sharing your analysis of the new version KB primary radar data and the new range/azimuth series. Is there also a new dataset available for the Butterworth section?

  86. airlandseaman says:

    Paul: Yes, I have new from BU. But it is not materially different from 2018 data since BU was further from the target. But I’ll put a file together.

  87. Paul Smithson says:

    @ALSM, many thanks.

  88. Paul Smithson says:

    @Victor, Dr B. One thing that bothers me a little is the way in which Victor’s range/azimuth to lat/long “engine” treats horizontal distance to radar head and lat/long. We’ve used a “flat earth” for purpose of calculating horizontal range where measured range (data) is hypotenuse, assumed altitude is opposite and horizontal distance is adjacent. And then obtain lat/long using arc distance on an oblate sphere. Does it matter, or are the effects too small to make any material difference over the distances in question? Secondly, I don’t know whether the radar’s “range” is actually raw, or if itself is a processed “horizontal range” that takes account of target elevation and earth’s curvature. You’d imagine that it must be processed to obtain reasonably accurate position at larger distances.

  89. airlandseaman says:

    Paul: Re “…I don’t know whether the radar’s “range” is actually raw, or if itself is a processed “horizontal range”. Why not? I told you it was the true range.

    The new KB radar data contains the true range and az data from the radar head. It is not processed (as was the 2018 data).

    As for Victor’s “engine”, I’ll let him elaborate, but we went through this last April. It is good to a few meters.

  90. TBill says:

    The Helios flight the air outflow valve was partially closed and bleed air was on, such that the flight maintained approx. 25000-ft altitude inside the aircraft, which is survivable especially with bottle air. If MH370 was intentionally depressured, it could have been much higher altitude, but of course we do not know. It takes time to depressure, but if you open the outflow valves and cut bleed air, it would not take too long.

  91. Paul Smithson says:

    @ALSM. “I told you it was the true range”. Yes, you said the same about the first dataset 😉

    On another note, I wonder if the other sample data of other fly-bys that you have access to for this radar head gives us an alternative method of altitude estimation. Can the COS diameter/elevation characteristics allow us to estimate altitude if we can make a reasonable estimate of CPA for MH370? eg If COS boundary occurs when horizontal range is double the target’s height…

  92. airlandseaman says:

    Paul: Not true. We said all along there were questions about the level of processing (coasting, etc.) on the April 2018 data set. I guess you missed that. In any event, the 2019 data was extracted from a very large file containing both data directly from the radar head (unprocessed) and data that had been processed for display on ATC screens. This new data is the best we are going to get.

  93. PaxLambda says:

    @ALSM : ” The original files contained a large amount of other data unrelated to MH370.”

    Did these other data are related to other planes? If yes, have you been able to compare some of them to known paths?


  94. airlandseaman says:

    PaxLambda: Yes. We found a good candidate for calibration of the KB radar and the method used. AIQ2932 (SQK 2254/8) passed very close to KB radar at FL380 at 15:44UTC. Unfortunately, all we have extracted (so far) is SSR records, which includes RHO and THETA. But these values appear to be “processed values”. We will do a deep dive in the big file to see there is any PSR data for this flight.

  95. airlandseaman says:

    Here is a note on the status of efforts to calibrate the KB radar RHO/THETA errors and validate the altitude determination method.

  96. Paul Smithson says:

    @ALSM. Thanks for these “notes”. Another systematic anomaly exacerbated by proximity to radar head. It will be very interesting to see how this relates to the PSR data if you can get hold of it. I wonder has the software (or our assumed radar location) got the radar lat/long wrong?

  97. airlandseaman says:

    Paul: No, the KB Radar lat/lon is correct, and the radar software is doing what it is programmed to do. The problem is that the track for ATC display purposes is calculated (an X-Y plot) based on assumptions that work when the target is farther away, but not close to the radar.

  98. airlandseaman says:

    I looks like the radar network assumes the geo altitude is the FL altitude for X-Y plotting purposes. That is what is causing (most of) the error close to the radar.

  99. DennisW says:


    Thanks for your continued perserverance relative to this issue (and others).

    I took a quick pass at the new data, and did not see a significant change in the results. I think that reflects on how the analytics relate to the physics and measurables. I know you and DrB were never fans of my approach. I am not going to belabor it and open old wounds with a new publication.

  100. TBill says:

    The thing I notice is 39,000 to 40,000-ft is the altitude that pilot Ewan Wilson mentioned in his 2014 book Goodnight Malaysian 370. He also feels the aircraft could have done 45000-ft but Wilson sticks with 39000 to 40000-ft for his proposed intentional depressurization scenario. That book has always been one of my MH370 templates.

  101. David says:

    @Victor. Kok denies withholding key information, notes that none of his 7 counterparts dissented to the Safety Investigation Report and describes what might prompt a new search.

  102. Victor Iannello says:

    @David: Thanks. Based on past behavior, Malaysia has a credibility gap that is difficult for them to overcome.

  103. Victor Iannello says:

    @Paul Smithson said: One thing that bothers me a little is the way in which Victor’s range/azimuth to lat/long “engine” treats horizontal distance to radar head and lat/long… Does it matter, or are the effects too small to make any material difference over the distances in question?

    The approximation that I have used in the spreadsheet is to linearize the curvature of the WGS84 ellipsoid at the radar site. This produces a linear relationship between X and longitude and between Y and latitude, and the solution for latitude and longitude can be obtained non-iteratively.

    The more exact solution is to transform from Latitude-Longitude-Altitude (LLA) coordinates to Earth-Centered Earth-fixed (ECEF) coordinates and then transform again to East-North-Up (ENU) coordinates. The calculated range and azimuth (and elevation) between the radar and target is directly obtained from the ENU coordinates, and these values can be compared with the measured range and azimuth. If the calculated values of the range and azimuth do not match the measured range and azimuth, the latitude and longitude of the target are iteratively adjusted until the calculated range and azimuth match the measured values.

    Is the extra work of the iterative solution worth it? I looked at two targets corresponding to the first target of the inbound segment (range = 58.0 NM) and the last just before the Cone of Silence (range = 10.7 NM). For an assumed target geometric altitude of 39,542 ft, the errors introduced by the linearization were 0.068 NM (413 ft) for the first and 0.0080 NM (48 ft) for the second.

    [These values were corrected as units were km, not NM.]

  104. flatpack says:

    @Don Thompson

    Sorry for the tardy response, restricted internet access.

    “The call priorities are known, they’re evident in the SUs.” You are right, my bad, I had missed them.

    I guess it is not relevant to the matter in hand but just out of curiosity, what was the 4th call attempt all about?

  105. Niels says:

    For clarity: the radar data you obtained is still from the civilian radar installations only? So “BU” is the one at WMKB? Is there any development regarding “Western Hill” data?

  106. Victor Iannello says:

    @Niels: Malaysia has not released the military radar data. According to the Chief Investigator Kok, there are no plans to do so because the data is classified.

    At this point in time, there is no excuse for not releasing the data, as it poses no security risk. Whatever deficiencies in radar capabilities there might have been have long been attended to, or deemed not critical.

  107. airlandseaman says:

    Neils: Yes. All Civil PSR data. We have some for BU too. Still working on the military radar data. Slow, but steady progress with that effort.

  108. Victor Iannello says:

    @Paul Smithson: I corrected an error in my previous comment. The error values were reported in km, but the units were incorrectly stated to be NM. The units and values now agree.

    The errors introduced by the linearization are acceptably small, as even at a range of 58.0 NM, the error is only about the length of two B777s. At a range of 10.7 NM, the error is only about one-fourth the length of a B777.

  109. Niels says:

    @ALSM, Victor
    Thank you for clarifying. Yes agreed there is no excuse to keep the military radar data classified. I understand this is done by “tradition”, but capabilities must already be well known to relevant players in the region. Given the gravity of the case it is unacceptable if it would not be released.

  110. Andrew says:


    RE: “Would you mind me asking for some more information, please?”

    What Emergency Oxygen supplies were available to the Cabin Crew on MH370? Was there a fixed (built-in) supply, as well as portable oxygen bottles?

    The cabin crew would have had access to both the passenger oxygen system (ie the ‘fixed’ system) and the portable oxygen bottles. There are oxygen masks at the flight attendant stations and spare oxygen masks at every seat row.

    In the Helios Airways Flight 522 crash…one of the cabin crew was still alive and conscious about 2 1/2 hours after the flight crew apparently succumbed to hypoxia, apparently surviving on one or more portable oxygen bottles.

    As @TBill mentioned, the cabin remained partly pressurised in the Helios Airways accident. The investigation report states:

    “Based on calculations performed by Nord-Micro on data from the NVM, the cabin altitude during cruise was 24 000 ft. According to calculations by the Boeing Company, the cabin altitude was between 20 500 ft and 28 200 ft.”

    A portable oxygen bottle would have been sufficient at those cabin altitudes. The other cabin crew members presumably relied upon the passenger oxygen system, which in this case only provided oxygen for about 12 minutes (B737).

    How long would the cabin crew oxygen last in terms of total person hours, please? Could you also break it down by supplies, please? I believe there was a cabin crew of 10, all Malaysian.

    Once activated, the B777 chemical oxygen generators that supply the passenger oxygen system have a duration of about 22 minutes. There are multiple generators throughout the cabin (eg every seat row, every flight attendant station, every toilet) and they are activated independently by pulling on an associated oxygen mask lanyard.

    However, please note that the passenger oxygen masks are only designed to provide protection during an emergency descent. They are NOT designed for continuous use at high altitude. It is highly likely the occupants would have been incapacitated before the oxygen generators were depleted if the cabin was depressurised and the aircraft remained at its cruising altitude.

    The Safety Investigation Report states the aircraft had 15 portable oxygen bottles, each of which had a capacity of 310 litres. According to the report, the oxygen flow could be controlled from 0 to 20 litres per minute, providing a minimum supply of 15.5 minutes at full flow.

    As I see it, ‘dealing with’ the passengers would have been no problem at all, but the other pilot, and the cabin crew are a different story altogether. To my mind, if ZAS was able to pull this off single-handedly, is was absolutely masterful, if not miraculous. Would you agree, or can you see an easy way to also knock out the cabin crew?

    I think the biggest obstacle would have been the other pilot. However, if the other pilot had left the cockpit for some reason and Z was alone, he could have easily locked the cockpit door and denied access to the cockpit. If he also disabled the SATCOM, he would have prevented anyone in the cabin from alerting the outside world. Of course, he would have had to orchestrate some reason for the other pilot leave the cockpit for that scenario to work.

  111. Louis says:


    Another crash and disaster for the families, this morning (March 10 2019), of a B737-8 MAX, 6 minutes after take-off from Addis Ababa. Another “JT610” situation ??
    Link :
    1st flight : Oct 2018. Delivered : Nov 2018
    Link :

  112. vodkaferret says:


    the guardian is reporting “unstable vertical speed after takeoff”… which would be similar to JT610 also.

  113. Victor Iannello says:

    The ADS-B data from FR24 for ET302 shows some interesting things:

    1. The vertical speed was unstable and low for a takeoff.
    2. The groundspeed was high for this part of the flight.

    This does not appear to be a problem with engine thrust. Rather, it appears to be a problem with controlling pitch during the climb.

    The QNH was 1029 hPa at the airfield altitude of 7625 ft. At this QNH, the altitude correction is -434 ft, i.e., the standard altitude of the airport was 7191 ft. Therefore, 434 ft must be added to the pressure altitudes to convert to geometric altitudes.

    From the ADS-B data, the final position and pressure altitude were (9.01602,38.98415) at 8600 ft. From Google Earth, at this position, the altitude of the terrain is 8155 ft. The geometric altitude of the plane at that position was therefore (8600 + 434) = 9034 ft. The height above the terrain was (9034-8155) = 879 ft, and even lower before that.

    The interruption of the ADS-B data was probably due either to the loss of line-of-sight of the receiver at the low aircraft height, or due to the crash.

    [This comment was edited to more accurately calculate the geometric altitude.]

  114. Julia Farrington says:

    From what I have read the Captain was given permission to return to Addis Ababa after reporting technical problems with the aircraft.This will have to be verified

  115. airlandseaman says:

    Here is a preliminary cut at the FR24 image data converted to text (by Don) and cleaned up by me. I agree with Victor’s observations.

  116. Julia Farrington says:

    You will all know more about this technical issue than me.
    Why haven’t Boeing grounded all 737 Max aircraft after the Lion Air crash instead of issuing pilots with instructions/ directive to override a switch on the computer when the stalling fault occurs?
    To put it another way, Boeing is relying on remedial action by the Pilots to correct a potential fatal fault in its aircraft.
    Ryanair has just ordered over 100 X 737 Max aircraft which will be in service next month

  117. David says:

    @Andrew. You said, “The Safety Investigation Report states the aircraft had 15 portable oxygen bottles, each of which had a capacity of 310 litres. According to the report, the oxygen flow could be controlled from 0 to 20 litres per minute, providing a minimum supply of 15.5 minutes at full flow.”

    The CAL Training Manual 31-35-00 of 1995 describes another bottle that supplied via 2 regulators at 2 or 4 litres/min. If this was sufficient the 20 seems excessive except perhaps for medical emergencies. I imagine there are instructions as to how to adjust the flow to the proper rate though I suppose that varies with cabin altitude. If 2 does the trick inactive/low altitude and 4 is needed active/high, assuming the latter then each bottle could be expected to last five times as long?

  118. DennisW says:


    Typical human respiration at rest is about 8 liters of air per minute. Air is 20% oxygen. My assumption is that 2 liters of oxgen per minute would be sufficient at rest.

  119. David says:

    @DennisW. Thanks

  120. DennisW says:


    Things are not looking good for Boeing right now. I would expect their stock to take a real beating on Monday. I too felt that the response to the JT610 incident was very poor on the part of Boeing. The reluctance to ground the jet is based on purely economic consderations. Frankly, I would not fly in one at the moment.

  121. Andrew says:


    RE: “If 2 does the trick inactive/low altitude and 4 is needed active/high, assuming the latter then each bottle could be expected to last five times as long?”

    The portable oxygen bottles are primarily for therapeutic/first-aid use. My understanding is the 2 & 4 L/min settings you mentioned are based on the first-aid oxygen requirements stipulated in 14 CFR 23.1443 (d), which states:

    “If first-aid oxygen equipment is installed, the minimum mass flow of oxygen to each user may not be less than 4 liters per minute, STPD. However, there may be a means to decrease this flow to not less than 2 liters per minute, STPD, at any cabin altitude…”

    I don’t know where the information in the SIR originated, but on further investigation it’s probably wrong. The MAS Safety Emergency Procedures Manual indicates the airline’s B777 aircraft were fitted with SABRE portable oxygen bottles, which have a capacity of 310 L and two selectable flow rates: HI (7 L/min) and LO (2 L/min). The duration of each bottle is 44 minutes at HI and 155 minutes at LO.

    As Dennis mentioned, the typical resting respiration rate is about 8 L/min, which requires an oxygen flow of about 2 L/min at sea-level. The required flow rate increases with both activity and altitude. The efficacy of the portable oxygen is also constrained by the mask. The typical ‘orange cup’ continuous flow mask fitted to the portable bottles (and the fixed oxygen system) is only effective for continuous use up to about FL250.

  122. DrB says:


    You said: “I took a quick pass at the new data, and did not see a significant change in the results. I think that reflects on how the analytics relate to the physics and measurables. I know you and DrB were never fans of my approach. I am not going to belabor it and open old wounds with a new publication.”

    I have no fundamental problem with your approach except that it assumes a straight path with no turns. I think that assumption may be violated in this case by a slight turn near the entrance to the COS.

  123. Peter Norton says:

    @DennisW, @Andrew RE: MAX8
    What do you think about this analogy?
    “I must say that if i bought a car where the steering might go wrong under certain conditions and tbe solution was to pull a CB, i would pass”

  124. Louis says:

    From AFP ( Agence France Presse ) and FWIW :

    ” The plane was already on fire when it crashed to the ground. The crash caused a big explosion,” said witness Tegegn Dechasa at the site, littered with passenger belongings, human remains, and airplane parts.

    “The plane was in flames in its rear side shortly before the crash. The plane was swerving erratically before the crash.”

    Farmer Sisay Gemechu, said: “The plane seemed to be aiming to land at a nearby level open field, but crashed before reaching there. “

  125. Louis says:

    Black box from crashed Ethiopian Airlines flight recovered

    Reuters :

  126. Louis says:

    Latest info : both boxes recovered

  127. Andrew says:

    @Peter Norton

    RE: “I must say that if i bought a car where the steering might go wrong under certain conditions and tbe solution was to pull a CB, i would pass”

    So you’ve already bought the car?? In that case it’s probably a bit late to ‘pass’!

  128. Andrew says:

    CNBC: China and Indonesia ground all Boeing 737 MAX-8s after Ethiopia crash

    “Boeing shares slid almost 10 percent in early trading on Monday. The move, if maintained through normal trading hours, would be the biggest fall in Boeing’s stock in nearly two decades, halting a surge that has seen it triple in value in just over three years to a record high of $446 last week.”

  129. Peter Norton says:

    @Andrew: I didn’t say that, the quote isn’t from myself. I sourced it.
    I merely wanted to ask your thoughts about it.

  130. vodkaferret says:


    I’d be very interested in your opinion on this… and understand you might prefer not to answer. for sure there is not enough info yet… at least one eyewitness says Et302 was on fire before impact, which is very different to JT610. However it would appear there are some similarities between the behaviour of the 2 flights and for 2 brand new planes of the same model to fall out of the sky shortly after takeoff is unusual.

    If you were a max 8 pilot right now… would you be concerned?

    (again, understand if you don’t want to answer that. )

  131. Victor Iannello says:

    @Andrew: I would have been amazed if there was not a slide in Boeing stock price considering the financial importance of the B737 MAX and the growing movement to ground the plane. After the Lion Air JT610 crash, I would not have been hesitant to fly on a B737 MAX. Now, I would avoid it.

    The ADS-B signal was lost due to the low height of the plane above the terrain. It’s hard for us to know exactly what occurred without the later ADS-B data. The reports are the plane was experiencing problems and received ATC clearance to return to the airport. Whether or not there were problems with MCAS or runaway stabilizer trim, the pilots seemed to be battling a problem with pitch and speed. That is eerily similar to JT610.

  132. Andrew says:


    I agree, it’s unusual for two brand new aircraft of the same type to suffer an accident in what seem to be similar circumstances. However, I think it’s unwise to jump to conclusions when so little is known about this latest tragedy. There might have been a flight control problem, however, the FR data seems to show the vertical speed excursions occurred almost immediately after take-off, at a time when MCAS should not have been active. That’s somewhat different to what happened in the earlier Lion Air accident, where the flight control problems did not occur until after the flaps were retracted.

    The eyewitness reports of a fire are interesting, however, such reports are notoriously unreliable.

    Would I be concerned? I’d be lying if I said no, especially when we don’t know what caused the second accident. That concern might be tempered in the coming days, depending on the information that emerges.

  133. vodkaferret says:


    thank you very much as always

  134. TBill says:

    Boeing stock had soared to new highs after the low of 304 just after Lion Air, for what it’s worth.

  135. Victor Iannello says:

    @TBill: Boeing is in for a rough ride in the short term. The impact will be much greater than after JT610 because now there is a higher probability that the deficiency is systemic. However, I think it is unlikely that the stock price won’t recover and do well in the long term. The company is financially strong with a solid order book, and Airbus has had it’s own issues. Any deficiency it finds in the design should be addressable.

  136. Andrew says:

    @Peter Norton

    I realise it wasn’t ‘your’ quote, I just think it’s a silly analogy. I don’t wish to underplay the seriousness of the apparent problems with the 737MAX, but new aircraft often enter service with teething problems. The aircraft type I currently fly entered service a little over four years ago and it still has a number of ‘temporary abnormal behaviours’ and several operations engineering bulletins that cover problems that could be quite challenging if they occurred.

    Regarding your analogy and the 737MAX, this latest accident will no doubt have an effect on sales of the aircraft in the immediate future. However, those who have already contracted to buy the aircraft might find it a bit more difficult to back out, especially while the investigations of these accidents are pending.

  137. Victor Iannello says:

    @Andrew: As always, thank you for all your insights, and for giving us all reality checks.

  138. DennisW says:


    you said:

    I have no fundamental problem with your approach except that it assumes a straight path with no turns. I think that assumption may be violated in this case by a slight turn near the entrance to the COS.

    Yes, I also get the slight turn, but it appears to have little effect on the range data conclusions. I will try to take a closer look at it.

    As an aside, I find apects of the new data just as troubling as the previous data set.

  139. TBill says:

    While we await possible findings of the recent air accidents, for MH370’s 5th anniv. I had prepared a new thought experiment essay below. I’ve struggled with this write-up probably 6 months, this is what I ended up with. Also posted on my Twitter page @HDTBill

    MH370: What if Plausible Deniability was the Pilots Whole Strategy?

  140. David says:

    @Andrew. Thanks for the extra digging on portable oxygen.
    With access to unused passenger/lavatory etc oxygen generators and portables a crew member might last some time were the cabin altitude not to rise above the altitude where these can sustain useful consciousness in a resistant individual.

    As to time, I have not seen any research done on the cabin altitude range that would sustain consciousness in a pilot of high resistance, conscious but irrational for a couple of hours or more before becoming inactive. Possibly an initial descent, discontinued, would extend that time.

    The two (major) impediments to the likelihood of such a scene in MH370 are the evidence of the initial handling of the aircraft and its switches and the altitude alarm.

    It seems possible though that cabin crew members could survive a deliberate incomplete or slow decompression for some time.

  141. Andrew says:

    The FAA Continued Airworthiness Notification to the International Community (CANIC) re the 737 MAX:

  142. Don Thompson says:

    The FAA has issued a “Continued Airworthiness Notification to the International Community“.

    This CANIC details Boeing’s ongoing work…

    “Boeing’s completion of the flight control system enhancements, which provide reduced reliance on procedures associated with required pilot memory items. The FAA anticipates mandating these design changes [to the MCAS] by AD no later than April 2019.”

  143. Andrew says:


    I agree that a crew member might remain conscious for a long period by using portable bottles, etc if the cabin altitude did not go too high. Unfortunately there is no evidence to confirm how, or even if a depressurisation occurred, even though we might suspect such a scenario.

    I’d be really surprised if Z was ‘highly resistant’ to hypoxia, given his smoking habit. Various studies have shown the performance of chronic smokers degrades far earlier than non-smokers.

  144. DennisW says:


    OK, my best estimates at this point are:

    1> closest approach of 8.8nm

    2> CoS path of 19.1 nm

    3> Average CoS speed 520 knots

    4> CoS altitude of 43,000′

    I do not like conclusions 3 & 4, but they are what they are without making some adjustments to the raw data.

  145. David says:

    @Andrew. You said, “I’d be really surprised if Z was ‘highly resistant’ to hypoxia, given his smoking habit…”

    Yes, I was thinking that his co-pilot might remain active the longer.

  146. Victor Iannello says:

    Boeing confirmed late on Monday (Mar 11) it will deploy a software upgrade to the 737 MAX 8, a few hours after the Federal Aviation Administration said it would mandate “design changes” in the aircraft by April.

  147. airlandseaman says:

    Dennis: You are in the range of 40,000 – 43,000 feet, considered now within the best guess range. And the speed is within 10 kts of the consensus value=530 kts. I think your analysis is further confirmation, by a different method.

    The key question is: When did MH370 reach its max speed and level off? Was it ~17:36, or more like 17:38 or 39? It’s down to that question. If you think it got there early, then the altitude was around 40,000, and if you think it got there later, then more like 43,000 feet.

    I was in the later camp until the new data started relieving speed vs. time detail from 17:30 to 17:45 we have need seen before. This new detail leads me to believe the time was sooner. I think the 20 kt spike in speed at 17:36 is real, and there were at least 3 descents between 17:30 and 17:36 to feel for the best combination of altitude and Ground Speed. I think the altitude at 17:30 was higher, and the altitude was traded for speed in steps you can see in the data.

  148. Victor Iannello says:

    Later, Boeing issued a statement confirming that “in the aftermath of Lion Air Flight 610, Boeing has been developing a flight control software enhancement for the 737 MAX.”

    Boeing said the software changes to MCAS will address the input from the AOA sensor and limit how much MCAS can move the horizontal tail in response to an erroneous AOA signal.

    Updated story here:

  149. Andrew says:


    RE: “Yes, I was thinking that his co-pilot might remain active the longer.”

    He was a smoker too!

  150. DennisW says:


    I think we agree that the KB passage was high and fast. Not sure how or if it contributes to a terminal location.

    A colleague and good friend of mine has written a PhD thesis on the meta-analysis of MH370. In this case that refers to the combining the analytics of various contributors (many of you are featured prominently). I was unaware of the math techniques used to combine the results of different studies (used prominently for medical studies).

    It is fascinating and traces the evolution of analytics (both the physical, and human variations) over time. When it is published, I will post a link.

  151. TBill says:

    I realize CVR content is considered confidential, but has Indonesia at least summarized what the Lion Air CVR indicates? For example, NTSB has already given some summary of Atlas Air CVR, without disclosing actual content.

  152. Andrew says:


    I’m not aware of the Indonesians having released any further information since the preliminary report. In the past they have included significant excerpts from the CVR as part of the final report. I suspect that will be the case here too.

  153. Julia Farrington says:

    Singapore has banned all 737 Max 8 from flying over Singapore airspace.

  154. Julia Farrington says:

    Boeing and the FAA must be 100% confident there will not be any further fatal 737 Max 8 crashes to issue these statements (and allowing them to continue to be flown) particularly before April when the new systems adjustments on these aircraft are made. I know this blog is for MH370 but right now, I am feeling as horrified about this recent crash as I was/still am about MH370.
    Boeing will be gathering their top international lawyers and throwing money at them to exonerate them from mass murder or whatever the legal terminology will be. Of course we don’t yet know the official cause of the et302 fatal crash but Boeing and the FAA’s intervention tells me that they do.

  155. oddball says:

    @Andrew (& all others who contributed),

    Again thanks all for the information regarding emergency oxygen supplies.

    It’s an interesting puzzle, given that whatever air pressure existed (or didn’t) in the cabin was replicated in the cockpit, so trying to disable people in the cabin without also taking out oneself in the cockpit would be tricky. Although the Pressure Demand mask would be a great advantage, it would still be somewhat risky groping in the dark.

    It’s also interesting that smokers may have an advantage over non-smokers when it comes to remaining conscious at low atmospheric pressures, but as has been shown with people who move to high altitude environments (on the ground), the body is very adaptable. I know from my one (thankfully failed) attempt at taking up smoking as a teenager, that a large draw on a cigarette leaves one severely oxygen deprived. In my case, my vision turned white and my sense of balance collapsed so that I had to grab the door frame to stay upright. That was enough to put me off for life, but as is apparent, smokers obviously adjust to these things. Perhaps smoking is an advantage in this scenario.

    I believe that another likely factor is the presence of nicotine in the bloodstream / tissues – this is the highly addictive stimulant and stamina extender that smokers most crave. Being fully tanked up with nicotine before the flight might give greatly improved high altitude survivability and endurance, and maybe ZAS was aware of this.

    From what I have gleaned about his personality, he had a great thirst for technical knowledge, and liked to be smart, so I can imagine that he would have planned this very thoroughly.

    Of course I have now idea whether he might have known about this case, but given that it involved an airline pilot and an air hostess, he might have. This case provides various salutary lessons in how not to murder your wife.
    Perhaps Crafts’ biggest mistake was his failure to control his environment, running a woodchipper beside a lake, visible from the road, in the middle of a severe snowstorm – not at all a suspicious thing to be observed doing… It’s clearly a novel method, but despite disposing of the body, it left a mess of evidence.

    On the other hand, an aeroplane at high altitude, at night, out of cellphone coverage and far away from almost all observers provides a pretty good environment for nefarious activities – especially for the suicidal pilot.

  156. Andrew says:

    @Julia Farrington

    I can’t help but feel that some of these groundings are more for show than safety. In Australia’s case, there are currently no Australian airlines that operate the 737 MAX and only two foreign airlines that operate the aircraft on routes to Australia. One of those, SilkAir, has already had its 737 MAX aircraft grounded by the Singaporean CAAS, while the other, Fiji Airways, only has two 737 MAX aircraft that operate a limited number of services to Australia. The Indonesian Transportation Ministry, on the other hand, is reported to have announced that 737 MAX operations will be suspended “temporarily for inspection”. I’m curious to know what they’ll be inspecting and how they will decide if the aircraft is airworthy (or not), given that the cause of this latest accident is not yet known.

  157. Andrew says:


    RE: “It’s also interesting that smokers may have an advantage over non-smokers when it comes to remaining conscious at low atmospheric pressures…”

    You may have misunderstood; smokers do not have an advantage over non-smokers when it comes to hypoxia. In an oxygen deprived environment, smokers will become hypoxic earlier than non-smokers.

  158. Victor Iannello says:

    @DennisW: Does your friend as part of his doctoral work estimate the location of the crash site?

  159. Julia Farrington says:

    @andrew. Thanks. Interesting information

  160. Victor Iannello says:

    Here is the NTSB’s Investigative Update for 5Y3591:

  161. TBill says:

    What do you make of it (Atlas Air)?
    The thing I noticed was that the 270 heading was requested by ATC, so that was fine. Why would the pilots go full throttle and point the nose down intentionally?

  162. DennisW says:


    Does your friend as part of his doctoral work estimate the location of the crash site?


  163. Victor Iannello says:

    @TBill: I’m not a pilot, but here are two possibilities that I offer for discussion purposes only. Others more knowledgeable should feel free to weigh in:

    1) Applying elevator down and increasing thrust is consistent with an attempt to escape from a stall. Perhaps the pilot falsely perceived a stall.

    2) Deliberate act to crash at high speed.

    Expanding on (1), near the end of the flight, fuel reserves are low, and applying full thrust would raise the nose (as was seen in the crash after the go-around attempt at Rostov-on-Don). The somatogravic illusion caused by the acceleration will amplify the pitch up sensation, which might have led to the over-correction. That said, it seems odd to me that a descent from 6,000 ft to the ground was not arrested at some point.

    If the plane entered turbulence while at minimum maneuvering speed (MMS), it would be advantageous to increase speed to increase the stall margin. The turbulence might have been the precipitating event that led to the thrust increase.

    Based on what we know, I think (1) and (2) are on the table. The contents of the CVR should provide the clues we need.

  164. Andrew says:

    RE: Atlas 3591

    ”…the engines increased to maximum thrust, and the airplane pitch increased to about 4° nose up and then rapidly pitched nose down to about 49° in response to column input.”

    Hmmm, there’s an obvious conclusion. The CVR transcript will be very interesting.

  165. Andrew says:


    I think (1) is unlikely. Never say never, but a well trained pilot would normally trust his instruments and not be fooled by the somotogravic illusion. In any case, a pitch down to 49° is extreme by any measure. I really hope it doesn’t turn out to be (2), but I have a bad feeling it will.

  166. Andrew says:

    Further to the above, the initial pitch up following the application of thrust is a normal aircraft response in non-FBW aircraft with underslung engines like the B767.

  167. TBill says:

    @Victor @Andrew

    I was speculating
    3) One and/or both pilots and/or jump seat passenger did not have a seat belt on, and severe turbulence threw them around violently potentially taking out the pilots and bumping in to the flight controls.

    4) Less likely but trying to adjust a load shift due to the turbulence

    but SatCom Guru wasn’t buying my turbulence theory…he is at (2)

  168. Victor Iannello says:

    @Andrew: Yes, you are probably right. If true, I wonder how this could occur with two pilots in the cockpit.

  169. DennisW says:


    I really hope it doesn’t turn out to be (2), but I have a bad feeling it will.

    Wow, under the circumstances I certainly did not expect that response.

    I would add:

    3> Modern state control theory allows exploring for “hidden states” in a control system – MCAS is suffering from one or more of them.

    A couple were actually discovered in the NASA re-entry vehicle which fortunately never occurred in practice. My best guess would be that the MCAS system on the 737 Max is still not completely understood. An airplane is a very complex system with dozens of states. Some of which are likely to be irrecoverable.

    The CVR data will be interesting.

  170. Andrew says:

    Interestingily, the wording in the NTSB’s update has been amended to remove any mention of control column input. It now reads:

    ”…the engines increased to maximum thrust, and the airplane pitch increased to about 4° nose up.The airplane then pitched nose down over the next 18 seconds to about 49° in response to nose-down elevator deflection.”

    The previous wording was:

    ”…the engines increased to maximum thrust, and the airplane pitch increased to about 4° nose up and then rapidly pitched nose down to about 49° in response to column input.”

  171. Victor Iannello says:

    The NTSB has corrected its report to read:

    The airplane then pitched nose down over the next 18 seconds to about 49° in response to nose-down elevator deflection.

    Prior, the reference was to “column input”, not “elevator deflection”. It’s interesting that the NTSB is making this distinction.

  172. Victor Iannello says:

    @Andrew: Our comments crossed. What do you make of this distinction?

  173. Andrew says:

    RE: Ethiopian 302

    EASA suspends all Boeing 737 Max operations in Europe

    How long before the FAA follows suit?

  174. Andrew says:


    I’m not sure what to make of the amended wording. The original version was either an error, or it might have been correct and the wording has been amended to avoid speculation about a deliberate act.

  175. Niels says:

    Re MH370
    Using my WGS84 path generation tool I’ve made a small study. Based on the best-fit third order polynomial interpolation functions D(t) and r(t) (representing the partially compensated Doppler and BTO derived range) I have generated a number of near TT paths for a range of bias frequencies (FFB = 149 – 153 H). For each of the FFBs I “straightened” the paths (20:41 – 00:19) as much as possible by varying the starting latitude of the path. To select the “best” of the straightened paths I looked at the generated speed profile for each of the paths. For all paths I transformed the GS into TAS for 21:11, 22:41 and 00:11 using @DrB’s weather data file (based on 21:00 and 00:00 wind data). At the moment I only considered LRC speed setting. Using @DrB’s fuel flow analysis sheets, I selected the best fitting FL in accordance with the 00:11 TAS and assuming zero fuel weight at 00:11. Then I calculated back the weight and expected TAS at 22:41 and 21:11. Finally, I calculated the RMS error for the TAS at 21:11, 22:41 and 00:11 comparing path derived TAS and LRC/FF derived TAS. I find the lowest RMS error for a FFB = 150.5 Hz, LRC FL 360, resulting in a 00:19 latitude of S34.5. While there are some assumptions in this approach (the main being a TT path and a constant flight level at LRC setting), it is interesting to see that even with the FFB and FL as free/fitting parameters we end up close to the initial value (150 Hz), close to standard cruise altitude and close to a track of 180 degrees. It will take some time to write this out in a document; it needs to be checked by others.

    If there are no mistakes in the models used and the calculations made, and as a follow up on this I would like to add possible errors to the D and r values before fitting the third order polynomial. This will result in a range of D(t) and r(t) functions, described by a range of coefficients. This range of input functions combined with the procedure above will result in a range of 00:19 latitudes. Also important would be to consider other thrust plans. All together this would probably require a higher level of automation than I have realized so far, as it possibly implies several hundred path calculations and analyses.

  176. Mick Gilbert says:

    @Victor Iannello

    Well, I know that if the ATSB made an unreferenced change like that to an update some sections of the media (well, one person in the media) would be all over them for secretly amending their report.

  177. DennisW says:


    How long before the FAA follows suit?

    It would be the right thing to do, IMO. Regarding the change in wording of the NTSB report, I also believe it is appropriate under the circumstances.

  178. Andrew says:


    I agree, however, judging by the following statement, it seems the FAA is holding off for now:

    I suspect the economic impact of grounding the aircraft in the US was a factor in the FAA’s decision to allow the aircraft to keep flying. That could all change if a systemic problem is uncovered during the Ethiopian accident investigation.

  179. Victor Iannello says:

    @Niels: Is there a reason for omitting the data at 19:41? The 180° path along the 93.7° latitude looks fine at that time, too.

  180. Paul Smithson says:

    In a comment yesterday @ALSM makes reference to a higher resolution analysis of the speed profile. I have attached my write-up here [] and the final speed profile chart here [].

    Long story short. I concur with ALSM’s more recent estimates of altitude. By my reckoning it is a smidgin below 40,000ft geometric. More importantly, I think it reveals a very strange speed profile. There are strong indications of speed instability throughout this period, including a very rapid (+40knots in ~60 seconds) acceleration approaching KB.

  181. airlandseaman says:

    Paul: You say: “There are strong indications of speed instability throughout this period, including a very rapid (+40knots in ~60 seconds) acceleration approaching KB.”

    I disagree with this characterization. It implies that something was out of control. On the contrary, the speed profile appears to be deliberate and under control. It was not a steady increase as first believed. The new data provides the resolution to see the 3 (or more) deliberate steps in speed associated with descents to trade altitude for speed between 17:30 and 17:37. That is not a sign of “instability”. It is a sign of manual control to seek in trial steps the highest altitude and Ground Speed combination possible after the turn back.

  182. Mick Gilbert says:

    @Paul Smithson

    Looking at that data I’m struck by Mike’s earlier ‘The key question is: When did MH370 reach its max speed and level off?

    Does the data suggest that the aircraft ‘levelled off’ at any point?! That fluctuating speed profile suggests that the aircraft was not stabilised vertically. The military radar returns as detailed in the Safety Information Report have the aircraft between 24,450 ft and 47,500 ft coming across the Peninsula with speed variations that support the notion that the aircraft was indeed climbing and descending, that is, not stabilised vertically.


    … , the speed profile appears to be deliberate and under control.or not.

    This is yet another case of MH370 Rorschach syndrome. If it is ‘deliberate and under control‘ what’s the narrative for the earlier portion of the flight immediately after the turnback? If a fast and ostensibly high transit was the aim why was the aircraft dawdling around near its last known position at such a relatively low ground speed for so long? Had the HCM ATCC have managed the transfer by the book, they should have started chasing MH370 down at 1.25am, within three minutes of it not having radioed in. At the time when efforts should have commenced to locate the aircraft it now looks like it was less 20 nautical miles from where it went ‘missing’. That doesn’t strike me as being consistent with a ‘get out of Dodge’ strategy.

    As to ‘It is a sign of manual control to seek in trial steps the highest altitude and Ground Speed combination possible‘ that is, to say the least, a somewhat fanciful interpretation. It’s not like we’re talking about an experimental aeroplane on its maiden flight into unknown conditions. The B777’s performance was a known quantity and the aircraft was ostensibly in the hands of a very experienced B777 pilot. Under those circumstances why would you be playing the trial and error game?

  183. Paul Smithson says:

    I am told that my new analysis of speed from the Kota Bharu primary radar doesn’t open so I’m reposting again.

    1. The paper with a short description of the method and a series of speed profile charts for different altitudes in increments of 5,000ft, 1,000ft, 500ft.

    2. Best fit speed profile chart (at 39,500ft geometric). Two versions, one with 10-step speeds (approximately 40 second time periods). The second with 4-step speeds (approximately 15 second time periods) overlaid.

  184. Paul Smithson says:

    I’m with Mick on the above.

    The data we have is not compatible with the “hang a turn at IGARI and get out of dodge” narrative. From the turn exit to the start of the KB radar trace requires an average speed of ~456 over 6 whole minutes. A similar constant speed (mean 458kts) was maintained over the first minute (to 17:31:30) of the KB radar trace. That makes a total of 7 minutes with average speed a shade below 460. Then, over the next 5 minutes, speed climbs erratically to a peak of ~540kts approaching KB. Then speed comes back down again – varying up and down 10kts per minute between ~530 and ~520 until the end of the KB trace. If somebody thinks any of the above is materially incorrect, kindly demonstrate.

    1. Why spend 7 minutes at 460kt, then speed up to 540kt over the following 5 minutes, then slow down again?

    2. How do we explain the erratic acceleration and erratic speed beyond KB?

    3. What must the aircraft have been doing to gain +40kts in ground speed in one minute (173530 to 173630) at geometric altitude of ~40,000ft ?

  185. oddball says:

    @Paul Smithson,

    How accurate are the data from which you calculate the erratic speeds mentioned above? Is this simply from radar returns, and could air conditions vs radar returns account for some of this apparent variation in speed?

    I have a hypothesis for the slowdown immediately after the turn, but not for the later slow downs after a speed up, unless he really is playing with altitude vs speed for range. My argument would be that running with all unnecessary loads (including bleed air) turned off would (might) not be documented in the manual, so he might need to do some of his own ‘real life’ testing.

  186. oddball says:

    @Victor, all,

    1) Applying elevator down and increasing thrust is consistent with an attempt to escape from a stall. Perhaps the pilot falsely perceived a stall.

    2) Deliberate act to crash at high speed.

    Why can’t it be:

    (3) Elevator, actuator or linkage failure, similar to:


    (4) Blocked pitot tube al la AF447? (I know, different pitot tube manufacturers, not a know problem with 767, but also not impossible, I suppose.)

    It was approaching or in a turbulent storm cell, and one of the videos shows the plane flying in clear air, so I would doubt loss of horizon, especially when they would have been flying on instruments anyway.

    Also, 49 degrees down while maintaining GS into the wind doesn’t sound like stall conditions to me, but I’m not a pilot.

  187. Paul Smithson says:

    @oddball. The quantisation on the slant range is 0.016NM. That equates to 15kts for single step (3.814 seconds), 3.8kts for 4 steps and 1.5 kts for 10 steps. Apart from the resolution, there seems to be a bit of noise as well. A statistician would be able to give you a more definitive answer, by I think speed accuracy is about +/- 3 knots using the fitted azimuth smoothing method that I have described, particularly farther out from the radar.

    Speed accuracy / confidence interval poorer close to radar (by which I mean the last ~40 seconds incoming, the outgoing less of an issue). I haven’t calculated how much poorer. But even if the trend shown is “noisier” it is undeniably there.

  188. Niels says:

    Re 19:41 data

    That’s an interesting aspect. First of all, practically, all paths I’ve recently generated start at 18:28 (based on backward extending the r(t) and D(t) curves). For FFB values close to 150 Hz I observe with my path generation tool a dip in the speed around 19:41 and a slight curvature in the path. Probably my 19:41 position is therefore slightly more south then you would expect for a path ending S34.5. The question is if the dip is real or if it is an artefact related to the 19:41 (and perhaps 20:41) BFO values. Anyway, because of this I decided to apply the procedure to the 21:11 – 00:11 interval. Note that if you assume the dip is real it becomes rather easy to “backward” connect the path to near IGOGU at the right time for a direct turn south.

    The dip disappears for FFB values in the 156 – 158 Hz range, resulting in more straight paths ending in the S28 – S30 range, however for these FFB values I get rather high speeds in the 19:41 – 21:11 interval, while the 00:11 speed does not increase a lot. The rather big delta_speed from 21:11 to 00:11 then is hard to explain with the more standard speed settings.

    The relative insensitivity of the 00:11 speed to changes in the FFB (as compared to the 19:41 – 21:11 interval) is an aspect which is interesting on its own and a bit unexpected; I’m trying to understand it better. Note that in my procedure with the shift in FFB the 00:11 position is shifting because of trying to keep the path straight.

    I’m also comparing with your October 2017 results (great circle paths). Trying to bridge the different approaches, in your figure 2 it would be interesting to add a graph for the mean BTO error in addition to the RMS error (for LRC), and to see where this would be zero.

  189. Victor Iannello says:

    @Niels: When I generated the great circle paths, I should have recorded the average BTO (and BFO) errors in addition to the RMS error. It would take me a small amount of time to re-do those calculations and record the average errors.

    I am curious to see details of the entire continuous path you are generating.

  190. Victor Iannello says:

    @oddball said: Also, 49 degrees down while maintaining GS into the wind doesn’t sound like stall conditions to me, but I’m not a pilot.

    Of course not. That scenario was based on recovering from a stall by over-correcting.

    Elevator, actuator or linkage failure…

    That would be unlikely if the column input and stabilizer actuator position were both nose down.

  191. oddball says:


    if the column input and stabilizer actuator position were both nose down.

    Is that what the FDR shows?

    To my mind, the NTSB’s change of wording suggests they (i.e. column input vs elevator deflection) were not necessarily in agreement. To me, it strongly suggests the elevator deflection said nose down, irrespective of what the column input was. That could suggest linkage failure or actuator error, but I don’t know how the NTSB would know what the elevator deflection was, independent of the actuator position. Is there a separate, independent elevator position sensor? Or would it be from the wreckage itself?

  192. oddball says:

    @Paul Smithson,

    Many thanks.

    I’d have to go with the “performance testing scenario” for the later speed reductions then.

    In my hypothesis, the speed reduction immediately after the turn back would have been for “passenger confidence”, (as part of “controlling the environment” I mentioned earlier) followed by a gradual return to cruise speed, but I haven’t quite finished writing it all up / proof reading it just yet.

  193. TBill says:

    I am not following exactly, but I think you are finding 180 South path to fit. In case you are interested, please see my essay below “On the straightness of the BFO Trend.” I tend to feel 19:41 point is off a little or maneuver is in progress. I actually feel like the 38 South paths fits best until Arc5, at which point I envision a heading change to the southeast.

  194. Paul Smithson says:

    @oddball. Are you different to the rest of us an have an inbuilt GPS? How do passengers have any idea how fast an aircraft is travelling?!! Why should 460 “settle” them compared to a higher speed.

  195. paul smithson says:

    @Victor, Andrew, others.

    Can you comment on the feasibility of level acceleration at rate of +40kts from about 480 to about 520 TAS (~20kts tailwind subtracted from observed groundspeed) in ~60 seconds at this weight and geometric altitude of 39,500ft?

    Could an you also comment on feasibility of maintaining ~440TAS at the same altitude?

    I wonder whether we might be looking at a continuous thrust acceleration from 460 to 540 groundspeed, with a couple of step climbs along the way that create the pauses in the acceleration profile?

  196. TBill says:

    I would also be curious if there is enough time to dip down to say FL200 consistent with the eyewitness report of low flying aircraft. On the surface in my simulations, I needed all the time to speed up. But if MH370 actually managed to complete the U-turn at IGARI in a little over 2 minutes, which would be an extreme sharp turn, then there is couple extra minutes I suppose.

    The problem with all this is we need to know when the turn was started and when/where it ended.

  197. DennisW says:


    Interesting plots. I also took a pass at deriving velocity over time on the orginal data set. It was pretty trashy to say the least. Given some time (and enthusiasm), I may try again on the new data using spline interpolation on range, time, and azimuth. Last time I used simple boxcar filtering.

    I don’t have an opinion relative to the reason for the variations or even if they are completely velocity related. Assuming a fixed altitude is not unreasonable, but it does constrain the source of the variations.

  198. paul smithson says:

    @TBill. My hypothesis on that is:

    1. Complete the turn at IGARI onto the IGARI-BITOD leg at the last prevailing speed 471-473kts
    2. Commence a 180 degree turn at 15 degree bank angle at 17:21:59 and TAS ~445kts
    3. Continue with same TAS (now GS of about 458) all the way back to the start of the KB trace

    I think you will find that this provides an excellent match to the maximum NE positions in the military radar (which I take to represent the radius of the turn when plane is at near constant slant range from radar – and gets “captured again” having been lost by radar when the turn started). It also gets you on the right trajectory for what we know of the return trace shape/direction.

    It gets you back to the start of the KB trace right on time.

    How did I arrive at this solution? With a “trombone” model, where the radius of the curve is a function of speed, the position of the trombone’s curve must provide a decent positional match to the military radar, starting from IGARI at last known speed and getting back to start of KB trace on time.

    Each speed simulation gives you a different turn radius, required turn position and time back at start of KB trace. If you assume faster speed, the trombone must be further extended and has a wider radius – which produces increasingly poor fit to the observables from military trace and brings you back on the wrong trajectory.

    But rather than get sidetracked on that discussion, I’m hoping for some more commentary on what the updated KB speed profile might tell us.

  199. Victor Iannello says:

    @oddball: We don’t have access to the FDR data. My possibility (2) was based on the version of the report which referred to nose-down column input. If the elevator deflection does not correspond to the column input, then we have to consider a failure. Another possibility for the change in wording of the report is the column inputs for the captain and the FO might not have been the same direction.

  200. paul smithson says:

    Also note that:
    1. Model above predicted perfectly what we are now told was the position of turn exit/completion
    2. The start of the return trajectory is near perfect reciprocal of the IGARI-BITOD bearing. Which is why I think that turn onto the IGARI-BITOD leg was completed and only then was a 180 turn made onto the reciprocal course.

  201. TBill says:

    I am a little rusty but 15 deg bank is lazy wide turn? In my past simulations, I assumed the FI timing was correct, and they say the turn started almost immediately after 17:21:13…so I give it 10 seconds and start the turn. Then I would set A/P at 25 deg max radius, which I realize A/P may have been off and turn may have been sharper. That was before the Safety Information Report which gives some IGARI U-turn data points that investigators used to simulate, which gives a little different timing and location, and that I have not studied as much.

  202. paul smithson says:

    Sorry, Tbill. My bad – yes, 25 AOB not 15.

  203. paul smithson says:

    @Tbill. Interesting little discussion on bank angles here. Including the observation that if you want a tighter turn the operating procedure is to control speed rather than excessive AOB

  204. Victor Iannello says:

    @Paul Smithson: Here is a quick and dirty estimate of acceleration capability at a geometric altitude of 39500 ft, or about FL374 at the conditions near Kota Bharu. I make the following simplifying assumptions:

    1) The weight is 215 MT, and the (L/D) around cruise speeds is 17.
    2) The thrust capability of the engines scales as the ambient pressure.

    Using the maximum altitude table for LRC speeds, at 215 MT, a climb rate of 300 fpm (2.962 kn) can be sustained at FL404. The ambient pressure is 184.0 hPa. At M0.84 and ISA temperature, TAS = 481.8 kn. The maximum thrust at this altitude is therefore 215,000 kgf x (1/17 + 2.962/481.8) = 13,970 kgf.

    At FL374, the ambient pressure is 212.5 hPa. The maximum thrust is therefore 13,970 kgf x (212.5/184.0) = 16,134 kgf. The maximum acceleration is due to the thrust in excess of drag, and is equal to (16,134 – 215,000/17)/215,000 = 0.0162g.

    If I’ve done my math correctly, an acceleration of 0.0162g corresponds to a change in speed of about 18.5 kn/min. I’d use that as a first-order estimate of the maximum acceleration at FL374. At speeds lower than LRC, greater accelerations are possible. At speeds higher than LRC, the maximum acceleration is less.

  205. Victor Iannello says:

    @Paul Smithson: I suspect I am in the minority, but my first choice in selecting a configuration and route after IGARI is to consider that the autopilot and autothrottle are engaged. In a diversion, the workload for the pilot would be high at the turnback, and that is exacerbated by hand-flying at high altitudes, high bank angles, and high (and low) speeds.

    This contradicts what the safety investigators discovered in their simulations, as reported in the SIR. However, that work was strongly dependent on exact coordinates of the entrance and exit waypoints, and when the turn was begun relative to the entrance waypoint.

    Frankly, I don’t trust that these coordinates are accurate because they are from the military radar data. That same military radar data indicates an impossibly sharp turn to the left to start the turnback. My guess is that military data is inaccurate at the turnback, and also has numerous “missing hits” that were interpolated either by the radar system or by the analysts that interpreted the data.

  206. DrB says:

    President Trump says FAA will ground all B737-Max 8 and Max 9 aircraft in U.S.

  207. Victor Iannello says:

    The FAA has grounded the B737 MAX pending further review of data.

  208. Victor Iannello says:

    This excerpt is from the FAA’s Emergency Order of Prohibition for the 737 MAX.

    On March 13, 2019, the investigation of the ET302 crash developed new information from the wreckage concerning the aircraft’s configuration just after takeoff that, taken together with newly refined data from satellite-based tracking of the aircraft’s flight path, indicates some similarities between the ET302 and JT610 accidents that warrant further investigation of the possibility of a shared cause for the two incidents that needs to be better understood and addressed. Accordingly, the Acting Administrator is ordering all Boeing 737 MAX airplanes to be grounded pending further investigation.

  209. Victor Iannello says:

    The satellite-based tracking refers the ADS-B data collected by Aireon’s satellite network. Aireon provided the data to the NTSB.

    (I assume this data is no better than the data from the FDR. I read somewhere that the Germans were asked to read out the FDR, but they did not have the proper capabilities.)

  210. DennisW says:

    Well…Boeing stock has not moved in response. Difficult to understand since the Max accounts for 1/3 of their revenue. Also a significant blemish on the brand. SW airlines is also unaffected. ??

    Personally, I don’t see this as a problem with a quick fix.

  211. oddball says:

    Hi Paul,

    @oddball. Are you different to the rest of us

    Yeah, indeed! 😉 Clearly, I’m from another planet. Didn’t you realise?

    How do passengers have any idea how fast an aircraft is travelling?!! Why should 460 “settle” them compared to a higher speed.

    Ahh-ha! You miss the point, rather widely, I’m afraid.

    I didn’t say “settle”, but I guess it’s relevant. The object is to control (as fas as possible), so maybe I should have used ‘reassurance’ instead of ‘confidence’ (which I like better), but the whole idea (in my #1 hypothesis) would be to keep the pax completely compliant, quietly in their seats with their oxygen masks activated, with nothing more untoward to think about than what they are already facing. FWIW, the primary objective of all that is targeted very directly at the young co-pilot…

    So maybe it’s best if I let you answer your questions yourself. (Clue: You need to think outside the box. It’s not really about the speed. Did you ever see the movie “The Prestige”?)

    I don’t (and didn’t) mean to offend you, but it seems maybe I have nevertheless. May I apologise? Or am I just being over-sensitive again?

  212. paul smithson says:

    @oddball. didn’t mean to be rude. just wondering how you reckon the aircraft speed would impinge on the passengers in any way. i reckon the vast majority of passengers are completely oblivious to aircraft speed

  213. oddball says:

    @paul smithson,

    Ok, thanks, no problem!

    I agree with you 110%. It’s extremely difficult to get any idea of how fast an aircraft is travelling from inside it. Oblivious is an excellent choice of word.

    What they are not oblivious to is what matters – what the ears detect:

    1. Sound
    2. Balance/tilt and acceleration/deceleration.
    3. (Ears &/or viscera?) G-forces.
    4. (Possibly) Air pressure changes.

    I think most can detect the difference between climb, cruise and descent, and more importantly, the transitions. After an alarming mask drop, many would (might) know that it must be followed by a rapid descent. So, a sudden deceleration and nose-down should trigger some degree of somatogravic illusion.

    Plus, I would expect an audible sound difference, as the engine power is quickly reduced. All this would give the pax confidence that, despite an unexpected mask drop followed by an unintelligible pilot announcement in English (to a Chinese-only speaker) and a high rate turn (back), an (apparent) descent is all part of what you’d expect. The point of the magic trick example is that “magic” is about having the audience all believe a certain thing, when something else entirely is happening before their eyes.

    And the point of that?

    1. To maintain control, and prevent the outbreak of chaos in the cabin.

    2. To apply intense social pressure to the FO now locked out of the cockpit in the cabin, to prevent him from considering looking to round up a posse and attempting to bust down the cockpit door. Imagine: If the entire contingent of passengers and cabin crew firmly (confidently) believes that the senior pilot Captain has it all under control, who in his right mind is going to allow an over-excited if not insane junior pilot to attack the cockpit door?

    Then, after a few minutes “descent,” a slow return to LRC speed (and perhaps even, a slow climb) would be much less noticeable, if done over several minutes. Add a few “hops and bumps” (as seem to have occurred) and it would probably add to the impression of an on-going high-rate descent. Especially if they (those who could understand English) were told to expect that… And presumably they have English & Mandarin speaking cabin crew who would repeat it for the Chinese anyway.

    Does that explain it (or anything)?

  214. DennisW says:


    Posts are getting weird. Do something.

  215. TBill says:

    Atlas Air discussion…interesting point that if NTSB thought it was deliberate act, then FBI would take over, which has not happened.

  216. oddball says:

    Posts are getting weird. Do something.

    Now THAT’s hilarious!

  217. Andrew says:

    Finally, Boeing recommends grounding the entire global fleet of 737 MAX aircraft:

    In Consultation with the FAA, NTSB and its Customers, Boeing Supports Action to Temporarily Ground 737 MAX Operations

  218. Victor Iannello says:

    @TBill: The NTSB’s change in wording is curious, as it changed the implications tremendously.

    When I began reading the article you cited, there was mention of “engine surge”, which I thought was new information. I think they are using this term for the rapid increase in thrust rather than the aerodynamic connotation.

    I don’t know what to make of the claim about the FBI. The FBI was on the scene from day one, whether or not they are leading the investigation.

  219. TBill says:

    “…change in wording is curious, as it changed the implications tremendously”

    Yes this has been a mentally exhausting couple of days. That change of wording was discombobulating. Also we have a couple trips coming up on American/Southwest so we were trying to look up aircraft to see if was the MAX.

  220. Andrew says:

    @paul smithson

    RE: “Can you comment on the feasibility of level acceleration at rate of +40kts from about 480 to about 520 TAS (~20kts tailwind subtracted from observed groundspeed) in ~60 seconds at this weight and geometric altitude of 39,500ft?

    Could an you also comment on feasibility of maintaining ~440TAS at the same altitude?”

    Sorry, I missed this earlier. I note that Victor has already calculated the available acceleration at LRC and I agree with his math (‘maths’ in my part of the world!). The LRC TAS at FL374/215T is 482 kn, so the drag would have been increasing as the aircraft accelerated to 520 kn and the acceleration would have been decreasing. A level acceleration in 60 seconds does not seem possible.

    I wonder if the aircraft initially climbed high to somewhere around FL400 but was then too slow (440 KTAS) and did not have sufficient thrust to accelerate in level flight. The pilot would have needed to descend to increase speed. The final groundspeed of 530 kn yields a TAS of 510 kn, assuming a 20 kn tailwind. A TAS of 510 kn gives M0.87 at FL380/ISA+10, which is the aircraft’s MMO.

  221. Andrew says:


    According to the NTSB Aviation Investigation Manual, the FBI does not ‘take over’ an investigation in the event of suspected criminal activity. The NTSB would continue to investigate and assist the FBI with its own investigation where necessary.

    “Aircraft accidents in which criminal activity is suspected may involve the participation of law enforcement officials in the investigation. In such cases, law enforcement agencies are not parties to the investigation; however, the Board should provide the law enforcement agencies with necessary access to evidence. If evidence indicates that a criminal activity was involved, the Board investigation will be conducted to the extent necessary to determine probable cause, address any safety concerns, and, where appropriate, to furnish requested assistance to law enforcement agencies. In the absence of any such evidence, the Board investigation will continue in its normal scope.”

  222. Victor Iannello says:

    @Andrew: We have an airline pilot trained by an airline that is flying an airliner. If time was of the essence and he wanted to quickly turnback, do you think it was more likely that he would climb to or beyond the maximum altitude capability of the plane and aerobatically turn, or simply slow down to reduce travel in the wrong direction and reduce the radius of the turn?

    Those seem to be the two limiting cases.

  223. paul smithson says:

    Fair question @victor and I look forward to hearing Andrew’s answer. But even if the latter, the next question you might ask is: having completed the turn (at around 17:24:40) he spends the next 7 minutes at average GS of 460, TAS of <450kts? How can that possibly be compatible with "time was of the essence"?

  224. Victor Iannello says:

    @Paul Smithson: In either scenario, the time at low speed seems too long. We are trying to guess “why” the plane was flown in a certain way, and that is not easy.

    I’ll add that using the most recent set of radar data, I estimate that before the CoS, the speed levels off to 529 kn at a geometric altitude of 39,500 ft (FL374). However, the estimate of altitude assumes the time, range, and azimuth are reasonably accurate close to the CoS when the azimuth is rapidly changing. If the altitude is lower, there is more thrust available, and the increases in speed are less likely to be from descents.

    I am trying to logically and scientifically assess possibilities without jumping to premature conclusions. That doesn’t mean that I believe that some of the scenarios that have been proposed are wrong. I just don’t know.

  225. Andrew says:


    As I said to Mike, I’m not a fan of high altitude aerobatic manoeuvres in wide-bodied airliners, even ones that might be classified as ‘gentle’. If time was of the essence, a level turn at a reduced speed might be preferable. However, if my understanding is correct, the radar data analysis suggests a speed increase at a rate that was not possible in level flight at FL380-ish. It seems to me the acceleration could only have been achieved if the aircraft was descending, possibly from an altitude close to the maximum. If that’s how it happened, then perhaps the pilot traded some speed for altitude in a bid to reach the maximum altitude as quickly as possible. In the process he might have ended up well below the min drag speed, with insufficient thrust to accelerate. That said, the motivation for climbing so high in the first place is unclear.

  226. Victor Iannello says:

    @Andrew: I am uncertain about my best estimate of FL374 across the CoS, and I am uncertain about the finer structure of the speed profile. Any small reduction considerably increases the thrust available for acceleration, and any “linearization” of the speed profile reduces the acceleration.

  227. DennisW says:


    My opinion for the low speed is based on a negotiation scenario. ZS was not in a hurry. He wanted to remain near landing fields for a few hours. Why would he possibly be in a hurry in any scenario?


    My opinion for climbing to a high altitude is the motivation to avoid other aircraft in the airspace beyond KB.

  228. TBill says:

    Isn’t is possible the aircraft slowed down well below 460 during a climb? If you look at @sk999’s digitized flight path, there is quite a speed drop during the turn at IGARI.

  229. paul smithson says:

    @Victor. What do you make the mean GS from end of Kota Bharu trace to the “phone registration” position (or adjacent primary radar positions)? Does mean GS 529kt fit all the way through to Penang?

  230. Niels says:


    Re 19:41 BFO

    “I tend to feel 19:41 point is off a little or maneuver is in progress”
    I agree but I’m far from sure which of the two it is; you can make a case for both.

  231. Niels says:

    @Victor, others

    Re “near TT” paths

    Please find a quick summary of my recent preliminary results through the link:

    There are several steps included which are new for me (the WGS84 related coordinate transformations, the use of wind data, the use of FF model and LRC speed plan) so critical remarks are welcomed. Let me know if you see something unusual or questionable / if I should explain the procedures followed more in detail.

  232. Victor Iannello says:

    A new story reports that Blaine Gibson continues to keep a low profile amidst threats and false accusations of doctoring of evidence.

    The poor treatment that Blaine has received from some claiming to be honest investigators is despicable.

  233. airlandseaman says:

    Andrew, Victor, Paul: It is frustrating to keep hearing mis-characterizations of what I have proposed. Could we please stop exaggerating and start seriously considering what happened.

    The evidence we have (radar, ads-b, etc.) and the statements in the Final Report (manual banked turn at 35 degrees) are consistent with a *relatively* steep turn to the left after IGARI. Relatively, meaning not a normal turn. I estimate roughly 5000 feet was gained in the turn (37200 to ~42000?) before the PF started heading back down. The low speed at 17:30 is consistent with a climbing turn. It does not have to be an “aerobatic turn”. It could all be in the 1±.2g range. Nothing all that dramatic.

    As Andrew notes, you cannot explain the speed steps assuming level flight and thrust increase. It just makes far more sense to recognize that the PF was probing for the best GS and highest altitude possible.

  234. paul smithson says:

    @ALSM. 17:30:xx at the start of the KB primary trace is 6 or so minutes after the turnback, no matter what our theory of how that turn was made. Average groundspeed over this period appears to have been a shade under 460kt, as it was for the first minute of the KB data series. If low speed resulted from the nature of the turnback, why should that low speed persist for the following ~7 minutes?

  235. Victor Iannello says:

    @airlandseaman: What you propose is entirely possible. I am not nearly as confident as you for many reasons. That doesn’t mean that I am claiming you to be wrong. Some reasons are:

    1) The 38° turn is based on simulations performed by Malaysian investigators that were designed to match two waypoints for the start and end of the turn as determined from the military radar data. That is the same data that produced the impossibly sharp turn after IGARI, and the impossibly low speeds reported in the DSTG report. I suspect the military radar data was at the limits of the detection range of the radar installation at Western Hill, and there was considerable measurement error as well as missing data. I suspect that the inaccurate, sparse radar data was supplemented by interpolation either by the radar system or by investigators interpreting the data. That same interpolation caused the rendering of the impossibly sharp turn.

    2) A 38° turn is beyond the envelope protection limits of NORMAL flight control mode. When 35° is exceeded, the envelope protection attempts to return the bank angle to 30°. A pilot can overcome this by turning the wheel hard. That combined with a climb to an altitude equal to or exceeding the maximum LRC altitude might not qualify the turn as “aerobatic”, but it is far from normal.

    3) The accuracy of our altitude estimations are limited by the accuracy of the time, range, and azimuth of the radar data near the CoS. We know that the azimuth near the CoS was rapidly changing. It might be that there was processing at the radar head that reduced the accuracy of these measurements, as we saw for the previous data set where the radar system introduced errors downstream of the radar head. For this reason, I advise that we use the current data set with caution until we are sure of what we have.

    4) The available thrust for acceleration is highly dependent on the assumed altitude, as most of the available thrust balances the drag. As a result, relatively small changes in altitude have a large effect on the achievable acceleration. In a previous comment, I estimated the acceleration available at LRC speed at FL374 and 215 MT to be 18.5 kn/min. Over the course of the 6.16-minute inbound segment to Kota Bharu, that represents a change in velocity of 114 kn. Now, of course the acceleration would be greater than this at speeds lower than LRC, and higher at speeds greater than LRC, but over the course of 6 minutes, it is evident that substantial speed can be added, even if the maximum thrust was not applied for the entire interval. Because of uncertainty I believe exists in the estimation of the local slope of the speed profile and the estimated altitude near the CoS, I believe it may be possible to explain the speed profile with level flight.

    I think assembling a scenario that involves a sharp turn and a steep climb followed by a descent is a useful exercise. I’m not trying to discourage you from doing this. I’m simply less certain than you that it occurred.

  236. Victor Iannello says:

    @Paul Smithson: As I said earlier, I think the duration of the low speed for any scenario remains unexplained. It’s not particular to Mike’s theory, is it?

  237. airlandseaman says:


    We do not really know (or believe) the path between 17:21 and 17:30 and won’t unless and until the Malaysians release the military radar data, and even then, we may not know. Therefore, it is impossible to say what the average speed was during that time period. All we know is the last ADS-B ground speed at 17:20:34 (471.3 kts/37200 ft) and the first civil PSR derived radar speed at 17:30 (~457 kts/43200 ft). Given the wind at different altitudes, the TAS was ~490-495 at 17:21 and 430 kts at 17:30.

  238. paul smithson says:

    What I’m saying is that I’m puzzled that he seems to be saying the slow speed at 1730xx is explained by events at 1722-1724. I’m not following the reasoning.

  239. paul smithson says:

    @Victor. BTW, also allow me to answer my own question regarding groundspeed between end of Kota Bharu track and Penang. Measuring from last 5 points of KB to the first five points of the segment crossing the west coast of the peninsular, I get mean groundspeed of 529.6kts. This based on the new KB data and the old Butterworth data as previously shared by @ALSM. It remains to be seen how the new Butterworth dataset changes this mean GS but I suspect nothing much over a segment of this length/duration (65nm/7.4 mins). I am looking forward to seeing how the whole primary radar trace speed profile pans out once we have spliced it together and applied some noise suppression.

  240. airlandseaman says:

    Paul: The ground speed after KB remained ~530kts all the way to Penang. After the turn at Penang, the ground speed slowed because of the difference in tailwind after the turn, but the TAS remained about the same until 18:22.

  241. paul smithson says:

    I can see that I am going to have to write up properly and share my “trombone” model that joins IGARI transit to first KB radar position. It’s essentially very simple with regard to the assumptions employed.

    Start at IGARI at known time and TAS per last ACARS
    Complete the turn on to the IGARI-BITOD leg
    Allow the time and position of the start of turnback to vary (that is your trombone sliding in or out, creating a longer path overall via the length of the “arms”)
    Allow any speed from start of turn to the 1730xx as long as it is constant
    Assume a turnback at max bank angle of 25, noting that the diameter of the turn is affected accordingly (this widens the width across the tubes of the trombone)
    Make a 180 turn on to reciprocal course, then a single subsequent course adjustment towards KB (which is also suggested by the depictions of the military trace).

    There are, of course, many speed scenarios that can get you back to 1730xx on time, but with this set of simplifying assumptions they alter both the position and the diameter of the turn (length and width of trombone). Only one solution matches the timing AND puts the turnback in the correct place AND matches the return trajectory (which is determined by turn diameter).

    This model correctly predicted:

    1) the turn exit position as subsequently published (to within 0.5NM)
    2) the time of that position (to within 1 second)
    3) the predicted speed at 1730xx, assuming a constant TAS from the start of the turn onwards.

    I cannot prove that this is how the turnback was made, but it seems to me to be a pretty reasonable theory insofar as it matches the observable facts while relying on some parsimonious and reasonable assumptions.

  242. airlandseaman says:


    1. “Allow any speed from start of turn to the 1730xx as long as it is constant” Why constant? Very unlikely IMO. You already know there was a huge (~60 kts) speed reduction between 17:21 and 17:30.

    2. “Assume a turnback at max bank angle of 25” Why 25? It may not be accurate for the reasons Victor cited, but 38 degrees is the only “data” we have. What is the basis for ignoring this?

    3. “…it matches the observable facts …”. Well, we know the TAS was NOT constant.

  243. Mick Gilbert says:


    All we know is … the first civil PSR derived radar speed at 17:30 (~457 kts/43200 ft).

    Mike, you cannot ‘know‘ that the target’s altitude was 43200 ft at 17:30 based on the PSR data.

  244. airlandseaman says:

    Mick: What is your estimate? How do you know mine is wrong?

  245. Victor Iannello says:

    @Paul Smithson: I don’t know how you are accounting for wind, but I would superimpose the (constant) wind field on top of the turn at TAS.

  246. DennisW says:


    I’ve read your detailed “analytics” of the radar data. I have the very same feeling I had 4 years ago (or so) when people were sticking pins in the map at 38S. Just stop. You are extracting (trying to extract) more from the observables than the observables are capable of telling us. Having seen this movie a few times, I can tell you it does not have a happy ending.

  247. paul smithson says:

    @victor. yes, i used GDAS winds at FL340 1800z throughout

  248. paul smithson says:

    @mick @alsm the groundspeed at 1730xx is essentially insensitive to altitude assumption. it could have been at any altitude, within reason.

  249. paul smithson says:

    @All. Attached is the .kmz path that results from the azimuth-smoothed version of the KB primary radar data, assumed altitude 39,500 geometric. There are no doubt some further refinements that could be made to the azimuth fitting model. In addition, note that the azimuth fitting process will have the effect of obscuring (dampening & spreading out) actual course corrections that may have occurred. Changes in altitude would alter the course over ground shown here, particularly closer to radar.

  250. oddball says:


    Further to your recent comments, am I correct in thinking that you are vehemently opposed to the pure murder-suicide-by-pilot scenario?

    Is that what you meant when you said: Posts are getting weird? Would you elaborate, please?

    Personally, I now think straight murder-suicide is by far the most likely option, and negotiation is a pure and deliberate smoke screen, designed to throw us off the scent. (I think that’s a good analogy, BTW!) 😉

    Anyway, to answer your question:
    Why would he possibly be in a hurry in any scenario?

    Let’s start with: what he was doing was highly illegal. That’s why criminals usually arrange their getaway cars in advance.

    Second, while he was over Malaysia, he would be well within range of Malaysian fighter jets, which could come alongside and attempt to force him to land. If nothing else, this would very quickly demonstrate the situation to all the other 238 passengers and crew. If his plan was anything other than recovering from some unknown in-flight calamity, this would be very contrary to his intentions. It would also explain why he turned off the transponder, ACARS and SatCom – because he wanted to “go dark” and be very hard to find, right from the start, when he was most at risk.

    As I see it, the only scenario in which he would not be in a big hurry would be if he was suffering a genuine in-flight problem. Then, he’d want all the time available.

    Going to max altitude would have perhaps three purposes.

    1. Again, to make himself harder to find
    2. To enable his getaway to be quicker
    3. To hasten the effect of hypoxia.
    4. As you say, potentially avoid other aircraft.

    Regarding your Just stop etc, what is your proposed strategy going forward? What is the purpose (of this blog), who is allowed to contribute, and what may they contribute?

    Surely, this is group brainstorming, and to cripple that is to reduce its possible effectiveness, no?

  251. oddball says:

    Oops, sorry, faulty close tag after than…

  252. Peter Norton says:

    vertical speed profile comparison between Lion and Ethiopian

    I previously thought about how confusing the situation could have been on LNI610. (penultimate paragraph, Seattle Times quote)
    Similar thoughtshere

    PS: Some alleged insights into Boeing’s project management.

  253. oddball says:

    vertical speed profile comparison between Lion and Ethiopian

    FWIW (probably not very much)…

    These graphs remind me of the first Hybrid Toyota I ever drove (A Camry). It was a rental, but still had the manual in it, so I read it up to try to find the problem (solution), but never got to the bottom of it and just returned it after a couple of weeks.

    The problem with it was that it was impossible to achieve “moderate braking”. It would brake ok for a gentle slowdown, but anything more and it would start to brake hard. Dangerously hard, for anyone following. So you’d ease off and suddenly end up with no braking, and run the risk of rear ending someone at the other end. Quite often found myself yo-yo braking. Infuriating to drive!

    I know the Toyota Hybrid system has a very complex brake actuator system, because for most braking above minimum Regen Braking Speed, it tries to use only regen braking. Brake harder and it then uses all it’s got, including those antiquated steel disk friction thingies. Then, when the speed drops too low for adequate regen, it switches over to all mechanical braking. As you can probably imagine, this is no easy task, and I don’t know whether it was because of a fault there, or somewhere else.

    According to the manual, it also had a feature called something like “Emergency Brake Assist” which basically slams the brakes on for you if it senses “heavy” braking. It seemed to me that this might be what was happening, but it was adjusted badly, and thought that normal braking was an emergency. No, I’m not that crazy a driver, that I habitually stand on the brakes at the last second. In fact, I see idiots do that all the time, behind the last car in a queue at red lights, and it never ceases to amaze me. Especially the ones who overtake me to get there!

    But it strikes me as similar, and reminds me of another report of a pilot and a plane (can’t remember which) where they were engaged in a similarly dangerous 10-minute battle for control before the pilot finally figured it out – he had the Autopilot in the wrong mode or something similar.

    The fact that the prior Lion Air flight had the same problem but then fixed it suggested this new feature is doing something that the pilots just don’t grasp correctly – something contrary to what they are used to. Must be easy to work around, once you know! Kind of suggested by the earlier pilot not making a big fuss about it – probably cos it’s in the book.

  254. DennisW says:


    There is absolutely nothing to suggest a murder/suicide scenario was likely. Nothing!!! Nada!!!

    ZS was a patriot. In love with his country and frustrated by the political situation and the graft. No way was he suicidal and a murderer.

  255. Andrew says:

    @Peter Norton

    RE: “vertical speed profile comparison between Lion and Ethiopian”

    The thing that I find interesting is the time frame. The Lion Air aircraft’s vertical speed did not start fluctuating until after the flaps were retracted, about two minutes after take-off. The Ethiopian aircraft’s vertical speed, on the other hand, seems to have begun fluctuating almost immediately after take-off. MCAS should not have become active until after the flaps were retracted, so it seems that something else was happening during that period immediately after take-off.

    RE: “PS: Some alleged insights into Boeing’s project management.”

    Some people feel that the B787 project was the beginning of the rot that was responsible for the issues that aircraft had on introduction to service. There was a good article in The Seattle Times, but I can’t find it now.

  256. Donald says:


    First, and most importantly, I hope you are recovering well and feeling better!

    Correct me If I’m wrong (and I very well may be), but in your own negotiation (Nothing!!! Nada!!!) scenario and by your own words and account(s), one of your tenets has been that Z HAD TO FOLLOW THROUGH with his threat, period. That he was NOT bluffing.

    That sure looks a lot like suicide/mass murder, does it not?

    I hesitate to mention (but apparently some need to be reminded) that the outcome here is not in dispute. It’s EITHER an accident (mechanical/non-nefarious), which any reasonable person has completely ruled out (sorry Paul and Mick) OR a murder/suicide. 2 pilots and 237 souls have perished.

    The ONLY realistic wildcard scenario is a miscalculation by Z (fuel, hypoxia, cockpit breach) which prevented him from just what exactly (in your scenario Dennis)?

    You can’t have it both ways. Landing on Christmas Island is a long ways off from disappearing into the southern ocean and murdering a plane load of human beings. But surely the threat put forward in your scenario by Z and/or others wasn’t to fly to Christmas Island?

    Help me understand because I’m genuinely confused by your logic? Thanks in advance though I won’t expect a fleshed out and detailed response. You’ll just go to the “whacko” schtick as you’ve in truth got nothing that makes even a scintilla of sense.

  257. Victor Iannello says:

    @Andrew: If one of the AoA sensors was exhibiting very flaky behavior, I wonder if static pressure error (caused by an incorrect correction for AoA) caused altitude, vertical speed, and airspeed errors before the flaps were up and before the same bad sensor caused the MCAS issue.

  258. Andrew says:


    Perhaps, but my understanding is the static pressure correction provided by the AoA is relatively small, so a malfunctioning sensor might not cause such big fluctuations in the indications. The malfunctioning AoA sensor in the Lion Air aircraft did not seem to cause big changes in the indications either, until the MCAS activated after flap retraction. I guess we’ll just have to be patient until more information is released.

  259. Mick Gilbert says:


    Re: ‘Mick: What is your estimate? How do you know mine is wrong?

    Mike, I don’t have one as you can’t possibly generate an estimate for altitude based on range/azimuth data alone.

    Just to be clear I’m not taking issue with issue with anyone estimating, guesstimating or guessing anything. What I have a problem with is the use of the word ‘know‘ when refering to said estimates, guesstimates or guesses. While some words have different meanings in the US and Australia I’m pretty sure that ‘know’ is not one of them.

  260. Ventus45 says:


    Addis Ababa Airport HAAB is 7,625 ft amsl.
    What flap setting(s) would be used, compared to JAKARTA WIHH which is only 64 ft amsl ?
    I am thinking a lesser flap setting, so faster VR/V2, and thus flaps might have been retracted earlier (time wise) than on LionAir.

  261. Andrew says:


    I’m not familiar with the B737 operation, but these days it’s common to use reduced flap settings whenever possible. Flap retraction does not commence until after the aircraft reaches the acceleration altitude and accelerates past the relevant speed. The acceleration altitude varies between airlines/aircraft types, etc, but is normally between 1,000 – 1,500 ft AAL. I wouldn’t expect to see flight path fluctuations below that altitude if MCAS was the problem, even if a reduced flap setting was used for take-off.

  262. oddball says:

    Atlas Air & others, credible or not?


    There was a good article in The Seattle Times, but I can’t find it now.

    Is this the gist of it?

    This gives much more depth…

  263. Peter Norton says:

    @Andrew: I share your thoughts. Maybe this is the article you search? It’s about problems induced by outsourcing.

  264. Andrew says:

    @Peter Norton

    Yes, that was the gist of it. I also remember an article about the close ties between the FAA and Boeing and a decline in independent regulatory oversight in some areas related to certification.

  265. Mick Gilbert says:

    @Peter Norton

    Was it this one? It’s an oldie.

  266. Andrew says:

    @Mick Gilbert

    That’s not the article but, again, the gist was very similar. The concerns expressed in the 1995 article are just as relevant today as they were over 20 years ago.

  267. airlandseaman says:

    Mick: I guess you missed the memo. The 17:30 estimated ltitude is not based on “range and altitude “. It is based on conservation of total energy. Try to catch up.

  268. DennisW says:


    Yes, I am feeling good. Thx for asking.


    you can’t possibly generate an estimate for altitude based on range/azimuth data alone.


  269. TBill says:

    Greg Feith’s comments on Atlas Air are interesting I have much respect. We should really ask Greg Feith to comment on MH370 at this stage, but I know he was thinking human factor in the early days.

    @Victor @Paul @ALSM
    Let me try to summarize the IGARI turn discussion, and correct me if I am wrong:

    If we say MH370 flew a level U-turn at apporx. constant speed (per Paul and maybe Victor), then there is an uexplained mini-loiter where MH370 is going slow before a speed-up to KB. Historically Victor’s position has been constant speed/altitude for MH370 from IGARI to VAMPI, unless more radar data can be provided. Now we have more radar data, but we still do not know enough.

    If we say MH370 flew a zoom climb to higher altitude during the climb (ASLM, TBill) there is a big slow down by the end of the turn, which explains the apparent “mini-loiter”.

  270. Julia Farrington says:

    Interesting for the satellite info. Aireon

    Just an aside, had Boeing and the FAA not grounded Max 8, someone should have asked Boeing CEO’s etc to fly on a MAX with their families. I bet none of them would have taken up the offer.

  271. Victor Iannello says:

    @TBill said: If we say MH370 flew a level U-turn at approx. constant speed (per Paul and maybe Victor)

    I am firmly in the undecided camp. Mike makes a persuasive case for a turn, climb, and accelerating descents towards Kota Bharu. But I also appreciate the work that Paul is doing to explore solutions that are more in line with how one might expect an airline pilot to fly a wide-body airliner. As we all know, if we had access to the military radar data, we’d be in a much better place.

  272. formula says:

    @ Victor “As we all know, if we had access to the military radar data, we’d be in a much better place.”

    Have those with access to the military radar data disclosed that they have performed like calculations to those being discussed or are willing now to do so? I presume not, which of course points to the moral duty of the Malaysian Government to make the data available.

  273. DennisW says:


    The calculations being made here are not generally recognized as appropriate for the observables obtained. I still have absolutely no idea how ALSM and Paul are deriving altitude from the Kota Baru radar data. Further it is not at all clear to me how the radar data has in any way helped to refine a terminal location.

  274. formula says:

    @ DennisW – given where the search has reached, failing to turn stones might look negligent perhaps?

    There must be a general welcome for seeing whatever present endeavours produce (preferably assisted by data now available only to Malaysia) and then let experts opine as to what may have been learned to further the search.

  275. oddball says:

    @airlandseaman, @TBill,

    FWIW, I really like the idea of a highly-banked, high-rate climbing U-turn which ultimately trades speed for altitude. It fits my intentions hypothesis very well, regardless of whether this is the normal way to fly a wide-bodied airliner. From the moment the transponder stopped, there’s nothing normal about the flight, so it fits all the better, and to my warped mind, possibly illustrates just how clever ZAS’s plan really was.

    One of the benefits of such a turn is the virtual “blink-and-you-miss-it” factor, which is what all good magicians strive for! (Sorry DennisW, but having acquired a copy of Ewan Wilson’s book yesterday, I’m surprised by how closely his assessment of ZAS’s personality matches mine – if you can follow the deliberate ambiguity there?)

    That said, I do agree with you (DW) that we are not necessarily getting any closer to locating the end point at this stage. In hindsight, we humans are not that smart, and 5 years down the track, a lot of valuable evidence (and opportunities to find it) will have been lost. Please allow me to throw in another of my oddball comments:

    Lots of undersea wildlife would have known within hours where the plane went down. If we could have followed the right ones, we would have found it too. That said, it is probably still possible, if only we could detect the scent of corroding plane parts in the water (which are probably not of much interest to wildlife).

    When we compare this with the crash of AF447, there are some worrying difficulties. As I now understand it, AF447 reported its GPS location via SatCom every 10 minutes. It hit the water on its belly, so the energy/momentum was dissipated over a large area. I think means the shattering of the plane would have been substantially less than in a high-speed nose-first impact (a la Atlas Air), where the energy & momentum of most of the rest of the plane is being delivered to the same impact point, at which the first arriving wreckage provides a more aggressive destruction zone for the later to arrive fuselage to destroy itself upon. Whew! Sorry about the long sentence.

    What I’m suggesting is that if MH370 entered the water at a similar speed, more-or-less nose first, the disintegration would have been far greater than that experienced by AF447. If it went in faster under pilot control, then even more so. (From that, I think it safe to assume that was ZAS’ intention, whether achieved or not.) Meaning the pieces would be much smaller, would sink more slowly, and would therefore disperse over a greater area, especially since the ocean depth is probably greater. If small pieces are indeed making their way to the other side of the SIO, then it seems apparent that the wreckage would be very widely dispersed on the seabed, so no doubt is harder to detect (again, as intended). And we really don’t even know where to continue looking yet.

    In summary, it took ~2 years to locate AF447, despite the location being known to within a few hundred square km. IMHO, it is highly possible that MH370 is within the areas that have already been searched, as per the following excerpt:

    … had to take into account the failure of four different searches after the plane went down. The first was the failure to find debris or bodies for six days after the plane went missing in June 2009; then there was the failure of acoustic searches in July 2009 to detect the pings from underwater locator beacons on the flight data recorder and cockpit voice recorder; next, another search in August 2009 failed to find anything using side-scanning sonar; and finally, there was another unsuccessful search using side-scanning sonar in April and May 2010.

    The searches all took place in different, sometimes overlapping areas, within 40 nautical miles of the last known location of the plane.

    The full story is here:

    Anyone know if anyone has successfully developed any “(under)water sniffing” sensors? We have two cats which are “not great” at hiding their scats in our garden, and as I have noted, even stupid dung flies can locate their dust and dirt covered target down to the centimetre, flying from God knows how far away.

    Similarly, our fruit bowl happens to be located next to a wall, and again, simply by extremely sensitive scent detection, dumb fruit flies can find their way into the house to land on the wall just above the bowl, conveniently signalling that it’s time to check it for the piece of overripe fruit.

    Maybe this is the kind of thing we need to start thinking about in our quest to find 9M-MRO?

  276. DennisW says:


    Your Wilson comment:

    I’m surprised by how closely his assessment of ZAS’s personality matches mine

    Wilson has two commercial failures in his recent history (the last as the CEO of a one plane aviation company). He is nothing more than an opportunist relative to the MH370 narrative. Wilson has yet to find something he is good at. Maybe he should try pyschology next.

  277. oddball says:


    Can I extrapolate that to “no other criticisms” then?

    For example, WRT the dispersal field comparison? The possibility it has already been missed?

  278. oddball says:

    Thanks TBill,

    (paywall but you get to see a little)

    If you hold your mouth right, you can get to read it all clearly.

    The deliberate blur seems to timeout after a while. Maybe it’s because I went to the author then to his latest article, then closed the sign up offer popup…

  279. DennisW says:


    Yes. I have no strong opinions on the debris field or why it was missed. The delay in the aerial search South was an avoidable mistake.

  280. Mick Gilbert says:


    Re: ‘I guess you missed the memo. The 17:30 estimated ltitude is not based on “range and altitude “. It is based on conservation of total energy. Try to catch up.

    Wow! ‘… conservation of total energy …‘. Maybe you missed the memo, Mike. A B777 is not a simple object in a Newtonian thought experiment. The good people at Boeing were thoughtful enough to fit their aircraft with a couple of features that allows it to alter its energy state on command. Those two fossil-fuel converters under the wing and a row of sheet metal on top of the wing allow it to add and dissipate energy.

    Moreover, when it comes to the notion of conservation of total energy your assumption of a climbing turn requires energy to be added to the initial 17:20:34 state. You can’t turn a 215-odd tonne airplane tightly through 180° and climb 8,000 feet to its maximum ceiling just by trading 40-odd knots of airspeed. You’re assuming that thrust was added. So we’re hardly talking about a conserved energy state.

    In any event, let’s not forget that this discussion started with your statement, ‘All we know is the last ADS-B ground speed at 17:20:34 (471.3 kts/37200 ft) and the first civil PSR derived radar speed at 17:30 (~457 kts/43200 ft).‘ As I thought I had made clear previously, my problem is with your use of the word ‘know‘.

    You can dress it up anyway you like but when it comes to your stated ‘~457 kts/43200 ft‘, the ‘43200 ft‘ part is at best an estimate based on some assumptions that may or may not be correct. You cannot possibly know that the aircraft was flown at that altitude at that time.

    And to the extent that I’m sounding like a pedant, well, good. If we are posting stuff that Joe Public can read then there is an implicit responsibility on us to be accurate. At least half of the nonsense surrounding MH370 started out with someone of standing stating their opinion, estimate, assumption or guess as though it were a fact. Next thing it has entered MH370 Lore. Mathematicians don’t use ≅ and = interchangeably. Surely it is not too much to ask to use words like ‘estimate’, ‘hypothesise’ and ‘assume’ where they more accurately convey the level of certainty. Let’s keep ‘know’ for facts.

    And … yes, there you go. All caught up now, thanks.

  281. DennisW says:


    Good post. Long overdue, IMO. There is a lot of unpoliced arm waving going on. Frankly, I am tired of it. I will be very hard on it from now on, until Victor (who does not seem to mind) bans me.

  282. oddball says:

    @The Experts

    Could one of you please point me to the latest accepted satellite handshake dataset, please, preferably with the currently accepted calculated values for speed wrt and distance from the satellite also included? I’ll accept whatever are the accepted BTO/BFO calculations and conversions so don’t require all those details for now, but happy to have if included. Likewise, satellite wobble compensation data/calcs.


  283. Dennisw says:


    you said:

    I’ll add that using the most recent set of radar data, I estimate that before the CoS, the speed levels off to 529 kn at a geometric altitude of 39,500 ft (FL374).

    Could you please post a link that explains how you came to the conclusion?

  284. oddball says:


    I will be very hard on it from now on, until Victor (who does not seem to mind) bans me.

    There’s nothing like autocratic opportunism, I say! 😉

    So long as Sauce for the Gander applies!?

  285. Victor Iannello says:

    @Andrew: If the WSJ article is correct, it was disorientation that caused the crash of 5Y3591. (This was option [1] above.) Perhaps that the thrust increase led to the pitch-up moment and caused a somatogravic illusion and an overcorrection with nose-down inputs. Even though pilots are trained to recognize this, in this case it appears they did not. It’s hard to believe that 6,000 ft was not sufficient to recognize their predicament and recover.

    By Andy Pasztor
    March 15, 2019 4:06 p.m. ET

    Federal air-crash investigators suspect that pilot errors, rather than aircraft malfunctions, led to an Atlas Air cargo plane’s nosedive near Houston in February that killed all three people on board, according to people familiar with the details.

    National Transportation Safety Board experts, these people said, are focusing on a likely sequence of events that started with the crew of the Boeing Co. 767 approaching Houston’s George Bush Intercontinental Airport on Feb. 23 inadvertently commanding dramatically increased engine thrust. Turbulent air could have jostled the arm of one of the pilots, causing the engines to rev up to takeoff power, one of these people said.

    The sudden surge in thrust, which the safety board disclosed in an earlier factual update, forced the nose of the plane to pitch upward and startled the cockpit crew, according to these people. Almost immediately, according to the preliminary data released by the safety board, the crew responded by sharply pushing down the nose of the aircraft.

    The board previously said the nose was pointed downward at a 49-degree angle with the plane still about 30 miles from the airport, creating a much steeper descent than a normal landing approach.

    The seemingly disoriented crew failed to regain control—despite commands to pull up from the jet’s high-speed dive—and the wide-body plane plowed into a marshy area.

    The safety board said the crew had the required training and medical certificates.

    The safety board hasn’t issued any final conclusions, and the leading theory currently pursued by investigators could change as more information is developed. A spokesman for the board said it had no comment beyond the factual update released earlier. Boeing had no immediate comment.

    Atlas Air Worldwide Holdings Inc. and the union representing its pilots both declined to comment, citing the ongoing investigation.

    But at this point, such cockpit slipups are considered the most likely cause of the crash. The plane was skirting around some storm cells before the plunge, but people familiar with the details said the turbulence didn’t cause any structural damage or lead any system to malfunction.

    In its update, the safety board said the “engines increased to maximum thrust” as the plane was flying at roughly 6,000 feet. After a brief nose-up movement, according to the update, the Boeing 767 entered a steep descent in a “generally wings-level attitude until impact with the swamp.”

    Many airline and aerospace industry officials have watched the probe closely because the 767 model is widely used as a passenger jet around the globe.
    The Atlas Air flight, en route to Houston from Miami, was flying cargo for Inc.

    The last fatal U.S. airliner crash also was a cargo flight. In 2013, a United Parcel Service Inc. Airbus A300 slammed into hill while approaching to land in Birmingham, Ala., killing both pilots. The safety board determined that a series of pilot errors and violations of safety procedures caused that accident. The aircraft descended too quickly toward a runway shrouded by clouds, and the pilots waited until the last seconds to try to initiate a go-around.

  286. Mick Gilbert says:


    Thank you, Dennis.

  287. Victor Iannello says:

    @DennisW said: There is a lot of unpoliced arm waving going on.

    What are you talking about? Did you see my previous comment where I said that Mike’s scenario was possible but not certain? Over the course of the 6-minute inbound segment to Kota Bharu, the energy added by the thrust might have come close to balancing the energy removed by the drag. However, the case of offsetting effects is a special case, and is in general not true, which is why I said the analysis was possible but not certain.

    I think his analysis would be improved if he included the time-dependent thrust, drag, and flight path angle, and I told him so in private. That doesn’t mean that there are no insights we can learn by what he presented. Mike is a pilot with 50 years of experience, and even if I don’t always agree with him, I listen to what he has to say.

  288. DennisW says:


    What I am talking about it is very very simple. Tell me how Paul or ALSM have derived altitude at Kota Bharu. You claim to endorse Paul’s analytics. Back it up or shut up.

  289. Victor Iannello says:

    @DennisW: I believe the statement you attributed to Paul Smithson was mine. Here is the basis:

    I looked at a very simple way of averaging the data by subdividing the inbound segment into 6 subsegments of about a minute duration, using a single segment across the CoS of about 2 minutes duration, and using 6 subsegments for the outbound segment of about 0.9 min duration, and calculating the average speed for each subsegment using the time, azimuth, and range at the endpoints of each subsegment. I estimated the geometric altitude by finding the altitude which minimized the variation in groundspeed for the subsegments starting with the one just before the CoS. The assumption is the plane accelerated towards the CoS, and then the speed was nearly flat. This assumption was then tested by calculating the variation about the average speed.

    The best estimate of geometric altitude using this method is 39,542 ft, which corresponds to a standard (pressure) altitude of around 37,387 ft (FL374). For the eight subsegments starting at 17:35:42, the average speed is 529.3 kn and the standard deviation is 2.4 kn.

    One might believe that my methodology of using the endpoints of the CoS to estimate the average speed across the CoS is flawed because of the distortion of range/azimuth data near the CoS. I don’t ascribe to this belief. The points near the CoS are the most sensitive to altitude, so are best used to discriminate the altitude. If we believe there is distortion near the CoS, then we should not try to discriminate altitude using any of these methods. Either the data near the CoS is valid, or it is not.

    At the prevailing meteorological conditions over Kota Bharu, at a groundspeed of 529 kn, TAS = 509.7 kn, and M=0.877. That Mach number is above Mmo but within the thrust capabilities at FL374. It may be that the autothrottle was disconnected, and the acceleration and speed were manually controlled.

    The calculated speed profile is shown here for the geometric altitude of 39,452 ft.

  290. Victor Iannello says:

    @DennisW: If you want me to ban you, I will. There is no need to stroke-out again because you don’t understand something.

  291. DennisW says:


    Look, my approach to the KB situation is the only approach that merits analytic cohesion. The others are simple arm waving. Good f’ing grief.

  292. Victor Iannello says:

    @DennisW: You last comment crossed with mine. You say You claim to endorse Paul’s analytics. Back it up or shut up. First, I don’t endorse his analytics. I am interested in learning more and testing it, and I tell him here and in private when I disagree with aspects of it.

    As for shutting up, I’ll say here whatever I damn well please. This is my last warning. The next time you act like this, you are out of here. I try to be patient with your social ineptness, and I have warned you in private, but I have my limits. You act like a child having a temper tantrum.

  293. DennisW says:

    @DennisW: If you want me to ban you, I will.

    No need. I am done here.

  294. Victor Iannello says:

    @DennisW: Your analytic approach is a gross approximation that hides the altitude discrimination near the CoS that we are trying to extract.

    Take a break Dennis or I’ll make the decision for you.

  295. oddball says:


    It’s hard to believe that 6,000 ft was not sufficient to recognize their predicament and recover.

    I reckon!

    It’s hard to believe that none of the 3 pairs of eyes apparently in the cockpit would be able to take in a rapidly unwinding Altimeter. Or the disappeared Artificial Horizon, or the advancing ASI. Or the associated sounds and, I presume, Audible Warnings.

    The similarities with AF447 are uncanny – except [maybe] that there, one pilot was countering the PF’s appropriate control inputs. Could this be the same but opposite? Maybe they thought the opposite to the initiating 447 pilot, that they were in a stall?

  296. David says:

    @Andrew. It will be interesting to learn what the defect write up and maintenance history of Ethiopian 302 was, noting that of JT610.

  297. Victor Iannello says:

    @oddball: If you are going to look for a similar crash, I would look at FZ981, where the pilots over-corrected for the pitch up and forward acceleration after go-around thrust was applied during an aborted landing. The plane was flown right into the ground, albeit from a much lower altitude than 5Y3591.

  298. TBill says:

    Re: Atlas Air
    Thank you for posting WSJ article which otherwise had a paywall.
    NTSB gets high marks from me for letting everyone know on a timely basis what they are thinking about the crash on a timely basis, although WSJ is making the claim. If you are correct on (1) OK but I said (3) maybe the turbulence caused the pilot(s) to knock the controls. I was thinking a little like the Kalitta incident where the hypoxic co-pilot was hitting the controls. But in any case sounds like NTSB has a reasonable initial hypothesis.

  299. Julia Farrington says:

    I fly regularly on Ryanair and assume all 737 Max aircraft will be grounded including the one Ryanair has ordered;Ryanair max 200 which Boeing calls part of the MAX family. (I think Boeing will have to change that unfortunate nomenclature)

  300. oddball says:

    @Victor: I would look at FZ981

    Thanks, done. Still like AF447, but agree, 981 also sounds like windshear induced, with the same result.

    Regarding the Sat Data and speed/distance analysis, is there an accepted latest version somewhere, or do I have to search for it and hope I get lucky?

  301. Andrew says:


    RE: “It’s hard to believe that 6,000 ft was not sufficient to recognize their predicament and recover.”

    Absolutely, and with two other pilots in the cockpit it’s hard to believe their predicament wasn’t recognised very quickly.

    The pitch-up response to the application of TOGA thrust is very strong in non-FBW aircraft, to the point where, during a go-around, the pilot needs to push forward on the control column to prevent the aircraft over-pitching. In the FZ981 accident, the pilot pushed forward on the control column and simultaneously trimmed nose down continuously for 12 seconds, causing the aircraft to pitch down to an extremely nose-low attitude that was not recoverable. If the Atlas pilot pushed and trimmed forward at the same time, then I guess it’s possible that something similar occurred.

    There are two things I’d interested to find out:

    – What caused the thrust increase? I find it hard to believe the thrust levers could have been inadvertently ‘knocked’ to a high thrust setting, even in turbulence. However, it might be possible that one of the TO/GA switches was accidentally pressed. That said, the TO/GA switches on other Boeing types aren’t normally active until the flaps are extended.

    – Was the autopilot engaged, or was the pilot flying manually?

  302. oddball says:


    Been thinking about this too.

    Would it be possible that they could have been slammed by a severe wind shear during the descent which suddenly tossed the plane from nose down to nose high. That might explain the leveling out just before the TOGA power and the dive began. I’m guessing something triggered severe somatogravic illusions in all three and was so intense that they thought they had stalled… Maybe they felt like the plane was suddenly “going backwards”, despite what the instruments said.

    Maybe they were suddenly all thinking Bagram Air Base 747, or AF447, or GXL888T and reacted accordingly?

  303. Victor Iannello says:

    @TBill said: Thank you for posting WSJ article which otherwise had a paywall.

    It’s the best American news source, and well worth the price.

  304. TBill says:

    DrB has posted a lot of tools in particular CBTO/CBFO spreadsheet.

    You can see the measured BTO/BFO and calculated for any time and position you want to enter. I mostly use DrB’s spreadsheet and ALSM has posted ping rings of the Arcs that can be displayed on Google Earth.

    So if I want to make up a new proposed route, I fly with flight sim connected to Google Earth, and I can see when I fly past a ping Arc, and then I use DrB’s tool to calc BFO for the time and position.

  305. Victor Iannello says:

    Here’s an interesting segment on Blaine Gibson that will appear on HBO’s Vice News.

  306. paul smithson says:


    I promised to get the “trombone model” of the turnback written up properly. To my mind, it provides a very robust and plausible case that the turnback was a simple 180 turn onto reciprocal course, executed at AOB25.

    This model accords nearly perfectly with:
    1. The maximum NE extent of travel according to military radar
    2. The bearing and positioning of the early part of the return trajectory
    3. The speed expected at the start of Kota Bharu radar trace
    4. The position/time abeam WPT2 (end of turn per Police Report) to within <2 seconds

    If this hypothesis is correct, MH370 was flying at groundspeed of close to 460kts for a full 10 minute period: from the start of the turnback (17:22:01) to more than one minute into the Kota Bharu trace (17:31:55).

    Depiction here

    Paper here

    Spreadsheet here

  307. paul smithson says:

    Victor has kindly reminded me that WPT2 should be attributed to the Safety Investigation report, not the police report.

    I should emphasise that the model doesn’t rely upon WPT1 or WPT2 at all. I am simply noting that WPT2 accords rather well with the model predictions. I believe, for reasons stated, that the discrepancy at WPT1 is due to an aberration of military radar coasting/signal processing.

    If you think about the military radar position at Western Hill, it’s understandable why military radar lost the target (it had just turned right at IGARI and radar failed to follow it) and why it regained target during the turn (halfway through the turnback the course is perpendicular to the radar, so slant range is constant).

  308. airlandseaman says:

    Paul: Comments on the turn model:
    1. There is no factual basis to support the assumptions that the turn was at a constant altitude and constant ground speed. It’s reasonable to test that and other “flat cases” cases, but it is certainly not the only case, and not the most likely case IMO.
    2. The simplified model assumes a constant ground speed (AKA no wind case?). As Victor noted, the wind is actually important here. At IGARI, there was a 20 kts headwind and at 17:30, a 20-30 kt tail wind, depending on altitude. Thus, the TAS changed from ~491 kts to ~430 kts between 17:21 and 17:30. This change in TAS over the time of the turn suggests a climbing turn, not a flat turn.

  309. paul smithson says:

    @Mike. I do *not* assume that the aircraft travelled at constant GS. The model works out the AVERAGE GS over the entire path that simultaneous matches the key parameters of x+r and turn diameter and gets you from LKP to KB_1 on time. Overlaying a wind vector on the model will not materially change this. It will simply reduce the average TAS (since the majority of the path has a tailwind).

    You say there’s not factual basis for level altitude turn. I’d agree that there’s no evidence pertaining to altitude at all. And I’d suggest that constant altitude is a sensible default assumption unless we have good reason to believe otherwise. In addition, I can see no reason by a speed reduction resulting from a climb should persist over a period of ~10 minutes.

    I you are so convinced that your posited climbing turn fits the observable facts, kindly model the resulting path and share the results.

  310. airlandseaman says:

    Paul: You are missing the point. How do you explain a 60 kts drop in TAS?

  311. paul smithson says:

    Um. Aircraft slowed down. And stayed slowed down for the next 10 minutes. What am I missing?

    I pointed out the other day in this pilots’ discussion on bank angle selection that the SOP would be to slow down and then make the turn, and the model results would be consistent with that.

    “At least over here, if there needs to be a control on the radius of a turn, it will be done by limiting the aircraft speed and not requiring excessive bank angle.”

  312. Andrew says:


    RE: “Been thinking about this too.”

    I doubt that it was anything quite as dramatic as you described. Something (or someone) obviously triggered a large thrust increase that seems to have caused the aircraft to momentarily pitch up. The subsequent pitch down might have been a gross overreaction by the pilot, but -49° is incredibly extreme. It’s hard to fathom how the nose got so low and why the pilots were unable to recover.

  313. Victor Iannello says:

    @Andrew: Some of the facts do line up with an intention to crash at high speed. What doesn’t make sense is how that was accomplished with two other pilots in the cockpit. Likely, the CVR will explain a lot.

    The crew-ATC exchange at around 12:36L is also strange. The FO talked over the ATCO three times, as if he did not understand the one-way nature of radio communication. That could indicate a high level of stress.

  314. Peter Norton says:

    Victor Iannello: “Some of the facts do line up with an intention to crash at high speed.”

    I don’t believe in an intentional crash here. And neither does the NTSB apparently. Actually, there are facts that all but exclude this hypothesis:

    • CVR suggests loss of control
    • nose was pulled up from -49° to -20° before impact

    « the cargo plane’s engines surged to maximum thrust and it briefly pointed its nose 4° up […]. The jet then rapidly swung to point 49° downward and began its drop toward the muddy bay. The NTSB […] says the plane’s stick shaker, which warns of an imminent engine stall, did not activate. That means it’s unlikely the pilots pointed the nose down to avoid stalling. The [NTSB] previously said cockpit audio suggests the pilots lost control […]. As the plane dropped, the [NTSB] says, it accelerated to 495 mph and gradually pulled up to a 20° descent. » (source: NBC)

  315. TBill says:

    Do we know the timestamps of WPT1 and WPT2? @sk999’s digitized flight path follows your basic curve, also hitting close to WPT1 and WPT2, but shows getting abeam WPT2 about 17:25:33 with quite a slow down from IGARI to completion of turn.

  316. Victor Iannello says:

    @Peter Norton: If the steep, high speed descent was intentional (and I present that only as a possibility), the pull up at the end could have been from an action of one of the two other pilots in the cockpit.

    I wouldn’t be so certain that the NTSB has ruled out deliberate actions. Their statements to date have been ambiguous, at best.

  317. oddball says:


    I doubt that it was anything quite as dramatic as you described.

    Hopefully, not, but I suggested it because of an experience I had about a week ago. And I was basically on terra firma. I was parking my car, swinging in from the RHS beside a dark blue car. (Important detail [blue] – not really! But memorable.) As I was about 1/2 way into the parking space, I turned my attention to the left to check my distance from the blue car. Due to the coincidence of actions, I did not observe that the blue car had already started reversing out of its space, and the blue car was my frame of reference at that moment. Suddenly, I perceived that I was heading into my parking space much too fast and pressed much harder on the brakes. Nothing changed, and for a moment, I could not figure out why my car was not stopping. I instantly began thinking ‘brake failure’ and started working on what to do next – which completely disrupted my attention to the parking job I had previously been doing.

    Just then, the front of the blue car passed the centre of my field of view and I could see the stationary ground beyond, shattering the illusion. I was already fully stopped, but did not know it until then. It was brief, but quite unnerving.

    Probably the most interesting aspect was that my mind instantly chose ‘brake failure’ and not ‘perception failure’, suggesting we(I…) trust our perceptions more than we trust our experience. Looking back, the brake pedal felt normal, braking had felt normal, my speed was already low (much lower than the speed of the other car going backwards), and I had no reason to suspect brake problems, but the sudden and unexpected movement of the other car overrode all the other information I had available to me in that moment.

    Maybe that’s like what is happening to the unfortunate pilots, in each of the several similar crashes?

    but -49° is incredibly extreme.

    Totally agree! I imagine this is the maximum designed descent angle for the 767, and only occurs with the elevator held hard ‘over’ (ok, under). I can’t imagine any normal pilot pushing the controls that far in, and surely, the resulting pitch down would just about lift you into your harness? Surely, they would perceive that. (No pilot would fly in bad weather with his harness undone, right?)

    It’s hard to fathom how the nose got so low and why the pilots were unable to recover.

    Elevator control failure, I’m still picking. Maybe the slow reduction to -20° was the result of a hurried manual elevator trim up adjustment, in the hope of being able to pull up despite the ‘hard under’ elevator. And maybe the Full Power move was also part of an attempt to “power out of it”? Maybe, despite the apparent stress in the cockpit, they did everything they could have?

  318. Andrew says:


    I agree, sensory illusions can be extremely disorienting, particularly when pilots are deprived of external visual references, as occurs while flying in cloud or sometimes at night. One of the most difficult things pilots must learn when they start flying on instruments is to trust the attitude indicator and to block out the sensations generated by their vestibular system. It becomes much easier with practice and experience because vision is such a powerful sense, but nobody is ever completely immune. Was it a factor in the Atlas accident? Maybe, but at this point we simply don’t know.

  319. Peter Norton says:

    Victor Iannello: “the pull up at the end could have been from an action of one of the two other pilots in the cockpit”

    Yes, but how would that square with “Crew communications consistent with a loss control of the aircraft began approximately 18 seconds prior to the end of the recording” and why wouldn’t the other 2 pilots have intervened in the first place (instead of only pulling up at the end)?

  320. oddball says:


    DrB has posted a lot of tools in particular CBTO/CBFO spreadsheet.

    Many thanks. Prodigious! My thanks to DrB.

    Should keep me quiet for a while.

    BTW, JFTR, I genuinely did manage to get through the WSJ paywall somehow. Maybe I should clarify, my wife and I did. I was complaining to her about it after fiddling about on her iPad for a while, and she said “You can.” Sure enough, by that stage, somehow the article was clear, and dismissing the popup allowed me full access. But I didn’t Copy & Paste it, as I didn’t want to get Victor in trouble for a copyright violation. Anyway, I don’t know exactly what I did, and have never been able to do it again, so maybe it was just at the moment they were updating stories, or something. I wonder how Victor got it.

  321. Victor Iannello says:

    @oddball: There is no need to wonder. I subscribe to the WSJ.

  322. oddball says:

    @paul smithson,

    I promised to get the “trombone model” of the turnback written up properly

    Thanks for your efforts and for posting that. It’s much clearer as an image! However, may I beg to differ, please?

    To my eye, the trombone pipe doesn’t follow the radar track after coming “around” IGARI, and even though the data is clearly a serious mess between ‘start’ and ‘end’ of the turn, couldn’t that be explained by a number of possibilities? The density of the “drawn” track is much less over that period, suggesting a lot of misses, so I suspect false returns and bad filtering. Not to mention the ~100° (“UFO”) turn near the ‘start’. Apart from that, I’d prefer to see a model that followed the consistent parts of the well-recorded track, rather than what seems to be trying to fit the data to a model.

    To my mind, that would result in a turn back of about 190~200°, which I guess “changes everything”? But where the data looks clear and consistent, I think it should be given greater credence.

    I’m wondering if any of the following points can explain the fade-out of the returns during the turn:

    * The plane was climbing
    * The plane was descending
    * The plane was changing speed (slowing)
    * The contrail boosted the return signals to and from (but not during) the turn
    * Appearing out of the “shadow” of the contrail near the start of the turn created the apparent 90°+ turn.

    (I’m not convinced of any of this myself [except maybe the 4th], but up for debate.)

  323. paul smithson says:

    @TBill sk999 digitised radar path is an integration of the DSTGs *KALMAN FILTERED* 10s military positions. It does not represent where the aircraft went, or even where the military radar thought it went. It is best regarded as a sort of “smoothed” version of the military radar. This should be fine during settled segments but not during the turnback.

    I had worked out the implicit timestamp for WPT2 ages ago but I can’t work out now how I’d done that and I don’t have time to go back and relocate it. You need to establish where the “time zero” was from which the simulator sessions were being measured.

  324. TBill says:

    We know they were expecting about 2 minutes from WPT1 to WPT2 which would be fast or tight. In the report I do not see any times mentioned for WPT1 and WPT2.

    Above you say the radar lost the target at IGARI, but the yellow line on your figure does not show that, so is that yellow line extrapolated?

    At Penang, do we have any idea why the Butterworth radar is showing the gaps? I noticed a really old post from you suggested maybe it was set for lower altitude aircraft making landing approach. I noticed on Jean-Luc’s video of air traffic, that some of the air traffic just before MH370 passed directly over Penang from the south.

  325. Victor Iannello says:

    @TBill: Cruising speed was maintained around Penang, and it is unlikely that MH370 did not remain at cruise altitude. I don’t know the reason for the gaps in the civilian radar coverage.

  326. paul smithson says:

    Thank you for prompting me to do this, TBill.

    The Safety report has initial conditions GS=475kt passing abeam IGARI
    We know that IGARI abeam = 17:20:31
    Distance from there to WPT1 is 15.26NM which takes 116 seconds @GS475kt
    So time at WPT1 should be 17:22:27
    On p278 of the report they say that the target time, per radar trace, to get from WPT1 to WPT2 is “two minutes and 10 seconds”
    Add 130 seconds to 17:22:27 gives you time at WPT2 of 17:24:37
    Now we can calculate speed required to get from WPT2 to first point in the IGARI trace (new, 2019 version) at 17:30:39, distance 45.35NM.

    And the answer is …drumroll… 451kts.

    Unless there is a mistaken assumption above, this demonstrates conclusively that the military radar time & position at WPT2 requires mean ground speed from end of turn to the start of KB trace of 451 knots. I think it’s fair to characterise that as independent corroboration of my model prediction of slow speed from the turnback all the way to the start of KB primary radar trace.

  327. Victor Iannello says:

    I’ve had some private discussions with some very knowledgeable individuals about the B737 MAX crashes. Here is an interesting hypothetical scenario based on some of those discussions.

    Suppose the crew of ET302 was fully aware of the circumstances of JT610, and after they entered a steep descent due to MCAS incorrectly trimming nose down, they opted to cut-out the electric stabilizer trim, as training advises. The question is, how difficult would it be to apply nose-up elevator and manually trim?

    One of the knowledgeable individuals provided the following:

    The relevant section [of the B737 FCTM] states:

    “If manual stabilizer trim is necessary, ensure both stabilizer trim cutout switches are in CUTOUT prior to extending the manual trim wheel handles.

    Excessive airloads on the stabilizer may require effort by both pilots to correct the mis-trim. In extreme cases it may be necessary to aerodynamically relieve the airloads to allow manual trimming.”

    The pilot would pull back on the control column to raise the nose, which obviously causes the elevators to move upward, however, the elevators are not powerful enough to overcome the nose down pitching moment of the stabilizer. The upward deflection of the elevators causes a turning moment on the stabilizer that holds the stabilizer against the nose down limit (ie leading edge up) and if the air loads are excessive, it may not be possible to reposition the stabilizer using the manual trim wheels. The pilot would first need to unload the elevators and then manually trim, which would take quite some time and is counterintuitive.

    Basically, pulling back on the control column would make it difficult to manually apply nose-up trim with the trim wheel. As the stabilizer has more pitch authority than the elevator, the only way to recover from the descent would be to push forward on the column (apply nose-down elevator) to unload the stabilizer so that it can be manually trimmed nose-up in order to arrest the descent.

    That could be very difficult for a crew to work together to do this in the time available if the descent begins at a relatively low height above the terrain.

  328. airlandseaman says:

    Victor: Interesting scenario and theory. Being so close to the ground when this started, it would be very difficult to unload the stabilizer enough without hitting the terrain. BTW…I have personally experienced a similar issue with the trim on Piper Super Cubs. The trim wheel tends to slip (does not move the stabilizer) unless the aerodynamic forces are close to neutral.

  329. Victor Iannello says:

    A new WSJ story on the B737 MAX crashes:

    Ethiopian Airlines Black Boxes Showed ‘Clear Similarities’ With Lion Air Crash

    Data from ET302 black boxes has been validated by Ethiopian, U.S. Investigators, transport minister says

    By Gabriele Steinhauser and Robert Wall
    Updated March 17, 2019 1:43 p.m. ET

    ADDIS ABABA, Ethiopia—Data retrieved from the black boxes of a crashed Ethiopian Airlines plane showed similarities to that from the Lion Air flight that plunged into the Java Sea in October, Ethiopia’s transport minister said Sunday, adding to the pressure on aircraft maker Boeing Co.

    “Clear similarities were noted between Ethiopian Air Flight 302 and Indonesian Lion Air Flight 610, which will be the subject of further study during the investigation,” Transport Minister Dagmawit Moges said. Both flights were on Boeing 737 MAX 8 aircraft.

    Ms. Moges declined to give details of the similarities that had been identified, including whether Boeing’s new anti-stalling software that has been associated with the Lion Air flight had been activated. She spoke after French air accident investigations bureau BEA had sent the data from both the cockpit voice recorder and the flight data recorder to Ethiopian authorities.

    Investigators from both Ethiopia and the U.S. National Transportation Safety Board have validated the data, she said, and a preliminary report on the accident will be published within 30 days. Representatives from the U.S. Federal Aviation Administration and its European counterparts were also present in the downloading of black box data.

    U.S. authorities last week grounded the 737 MAX after saying it had found enough similarities between the two crashes to take the rare step of keeping the aircraft out of the air. The move followed similar actions by a number of regulators around the world.

    American and Canadian officials cited satellite data showing a similar flight path between the two planes—both new jets that crashed after just a few minutes.

    Boeing has said it believes the plane is safe but agreed with the grounding.

    The Ethiopian findings of similarities between the two crashed planes come from an initial analysis of data retrieved from the plane’s black boxes. Voice recordings and flight data contained in these devices are typically the most important components in any crash probe. The similarities add a new level of scrutiny on the Boeing jet.

  330. Victor Iannello says:

    Boeing CEO Muilenburg Issues Statement on Ethiopian Airlines Flight 302 Accident Investigation

    CHICAGO, March 17, 2019 /PRNewswire/ — Boeing Chairman, President and CEO Dennis Muilenburg issued the following statement regarding the report from Ethiopian Transport Minister Dagmawit Moges today.

    First and foremost, our deepest sympathies are with the families and loved ones of those onboard Ethiopian Airlines Flight 302.

    Boeing continues to support the investigation, and is working with the authorities to evaluate new information as it becomes available. Safety is our highest priority as we design, build and support our airplanes. As part of our standard practice following any accident, we examine our aircraft design and operation, and when appropriate, institute product updates to further improve safety. While investigators continue to work to establish definitive conclusions, Boeing is finalizing its development of a previously-announced software update and pilot training revision that will address the MCAS flight control law’s behavior in response to erroneous sensor inputs. We also continue to provide technical assistance at the request of and under the direction of the National Transportation Safety Board, the U.S. Accredited Representative working with Ethiopian investigators.

    In accordance with international protocol, all inquiries about the ongoing accident investigation must be directed to the investigating authorities.

  331. Peter Norton says:

    oddball: “If you hold your mouth right, you can get to read it all clearly.”

    I’ll try. I have some doubts though … 😀

  332. oddball says:

    @Peter Norton:

    I’ll try.

    No need. Victor already copied it here:

    And as above a bit, I could never repeat the fluke…

  333. TBill says:

    Do the proposed software fixes cover that adequately? Presumably there will be less susceptibility to a single AoA reading, and the movement will be milder if triggered.

  334. Victor Iannello says:

    @TBill: My guess is there will be a software fix that will make it easier to recover from a configuration in which MCAS is incorrectly applied while still adequately correcting the relationship between pitch moment and elevator position at high angles of attack when it needs to.

    The crash of ET302 will be problematic for Boeing, especially if proper procedures were followed for runaway trim in the wake of the crash of JT610. An important question is whether the crew cut-out the stabilizer trim, in accordance with training. If they did, but could not manually trim to avert a crash, that would not bode well for Boeing. The decision to not ground after JT610 will certainly be questioned.

  335. Andrew says:


    Previous reports suggest the ‘fix’ will include a requirement for no disagreement between the two AoA sensors and a limit on.the total amount of stabiliser trim that can be applied by MCAS, such that recovery from a runaway would be possible using only the elevators. Yet to be confirmed.

  336. TBill says:

    Ok but you are estimating the WPT1 and WPT2 timestamps. We probably have to look at total time and distance from IGARI to the first KB points. Conceivably an ascent could have started about when the transponder was turned off, before WPT1, and so the timing may be different. Wonder if we asked if they would tell us the WPT1 and WPT2 times.

  337. Jethro says:

    [Derogatory comments about others deleted.]
    I would like to end this comment by saying what a pleasure it is to read this blog over the past number of years and I refresh it daily – as I’m sure many other “lurkers” do – for the insights given by the protagonists led by Victor, who post detailed and astounding research on a regular basis.

  338. Mick Gilbert says:

    @Victor Iannello

    Very interesting piece on the (self-)certification of the MAX here.

    Just two of the interesting bits:

    Current and former engineers directly involved with the evaluations or familiar with the document shared details of Boeing’s “System Safety Analysis” of MCAS, which The Seattle Times confirmed.
    The safety analysis:

    – Understated the power of the new flight control system, which was designed to swivel the horizontal tail to push the nose of the plane down to avert a stall. When the planes later entered service, MCAS was capable of moving the tail more than four times farther than was stated in the initial safety analysis document.

    – Failed to account for how the system could reset itself each time a pilot responded, thereby missing the potential impact of the system repeatedly pushing the airplane’s nose downward.

    – Assessed a failure of the system as one level below “catastrophic.” But even that “hazardous” danger level should have precluded activation of the system based on input from a single sensor — and yet that’s how it was designed.

    The people who spoke to The Seattle Times and shared details of the safety analysis all spoke on condition of anonymity to protect their jobs at the FAA and other aviation organizations.

    The original Boeing document provided to the FAA included a description specifying a limit to how much the system could move the horizontal tail — a limit of 0.6 degrees, out of a physical maximum of just less than 5 degrees of nose-down movement.

    That limit was later increased after flight tests showed that a more powerful movement of the tail was required to avert a high-speed stall, when the plane is in danger of losing lift and spiraling down.

    The behavior of a plane in a high angle-of-attack stall is difficult to model in advance purely by analysis and so, as test pilots work through stall-recovery routines during flight tests on a new airplane, it’s not uncommon to tweak the control software to refine the jet’s performance.

    After the Lion Air Flight 610 crash, Boeing for the first time provided to airlines details about MCAS. Boeing’s bulletin to the airlines stated that the limit of MCAS’s command was 2.5 degrees.

    That number was new to FAA engineers who had seen 0.6 degrees in the safety assessment.

    “The FAA believed the airplane was designed to the 0.6 limit, and that’s what the foreign regulatory authorities thought, too,” said an FAA engineer. “It makes a difference in your assessment of the hazard involved.”

    The higher limit meant that each time MCAS was triggered, it caused a much greater movement of the tail than was specified in that original safety analysis document.

  339. Mick Gilbert says:

    Apologies, a bit of a spacing issue there.

    The first extract ends with ‘… all spoke on condition of anonymity to protect their jobs at the FAA and other aviation organizations.

    and the second one starts with

    The original Boeing document provided to the FAA included a description …’

  340. Andrew says:

    @Mick Gilbert

    Wow, that’s incredibly damning. A major investigation into the FAA’s certification practices and the relationship between the FAA and Boeing is surely warranted.

  341. MH says:

    Just wondering what other Boeing aircraft were “(self)-certified” ? this is alarming !!

  342. Mick Gilbert says:


    Re: A major investigation into the FAA’s certification practices and the relationship between the FAA and Boeing is surely warranted.

    Ask and you shall receive.

    Feds probing regulators’ approval of Boeing 737 MAX jet following two fatal crashes

  343. TBill says:

    I am retired engineer from the corporate world, not airline industry, but I see a lot of possible similarities with Boeing management philosophy. The one difference I see (from following MH370) is Boeing was held in such high regard as far as US public opinion (compared to my industry), so they were quite untouchable as far as criticism and regulatory changes due to extraordionary safety record. Probably less so going forward.

  344. Victor Iannello says:

    The AoA data for ET302 is very similar to JT610, according to this Reuters article.

    PARIS (Reuters) – Investigators probing the crash of an Ethiopian Airlines Boeing 737 MAX jet eight days ago have found strong similarities in the ‘angle of attack’ data recorded by the doomed aircraft’s cockpit recorder and data from a Lion Air jet of the same model that crashed in October, a person familiar with the matter said.

    Graphs of the two sets of data are “very, very simliar,” the person said on Monday, asking not to be identified because the matter is still in the early stages of investigation.

    The AoA sensors are inherently reliable devices. I’m told that the same sensor type is used on the B737 MAX as the B767. We know that the AoA sensor was replaced prior to JT610 because of anomalous behavior. It is very unlikely that there were three bad AoA sensors.

    When the AoA data for JT610 was published, the error in the faulty AoA reading seemed to jump around, but the smaller changes in AoA readings for the two channels seemed to agree. This suggested to me that perhaps there were bad “bits” either in the digitizing of the signal or in the software.

    We now know that ET302 had similar AoA erroneous readings as JT610. This again points to a problem in the digital electronics or the software.

  345. Victor Iannello says:

    This figure shows some of the parameters from the FDR of JT610. Looking at the Angle of Attack signals, notice that once speed is established, the offset between the left and right channels remains nearly constant, and the higher frequency content is nearly identical. This suggests the problem was not mechanical.

    Also note that the bad AoA reading introduces errors in the altitude and airspeed readings. This is because the static pressure is incorrectly adjusted for AoA for the channel with the erroneous AoA reading.

  346. oddball says:


    That FDR trace image is incredible!

    It must have been an absolutely terrible few minutes.

    A couple of things I don’t understand – what does the trace for “Flap Handle Position” indicate? That they took off with flaps up? Then lowered, then the raised, then lowered them for the rest of the flight? Or is it inverted? I presume so, but even so, the flaps positions seem weird.

    I also notice that the Stick Shaker Left went off before they even left the ground, accordinging to the Air / Ground trace, and that the Master Caution went off following the first change in the “Flap Handle Position”.

    It seems to me that the crew tried to reverse virtually every move of “Trim Automatic” with a “Trim Manual” moments later, until close to the end when the Auto Trim won out. Also that, when the “Flap Handle Position” was up, there was less fighting over trim.

    My question is, was there a bad AoA signal? They appear to match to me, i.e, both went up and down at the same times, once flight was established. To me, the telling thing is that the speed traces diverge during the takeoff roll, and that the Left Stick Shaker went off at that time – exactly when the speed traces start to diverge, but while they are clearly still on the ground.

    As I see it, rightly or wrongly, it appears that the software set the “Zero Points” for both AOA sensors before they had even started to operate, and before there were any valid readings from either of them. That would expain why it “thought” there was an AOA error, because it asssumed the random, pre-flight resting positions where valid starting positions, which they clearly need not be.

    This is clear from the start of the AOA traces, at which time, the Left AOA sensor was sitting high and the Right was sitting Low. You can see that the traces both move towards each other during the roll, then simply follow each other for the rest of the flight, but the computed air speed traces show consistent differences until the final dive, when the air speed goes high and altitude goes low.

    To me, the fact that the Left Stick Shaker only stopped for a brief few seconds near the end of the first dive shows that the Stick Shaker Signal was erroneous for most of the flight, as the plane seems to have experienced just about every possible flight condition during those few minutes, but the software controlling the Stick Shaker signal was “almost never satisfied”.

    To me, it looks like the logic is utterly screwed, as if, in at least two places, the data is upside down.

    For example, the Stick Shaker signal could be inverted, On when it should be Off and vice versa, and the Flap Handle Position likewise.

  347. oddball says:

    Sorry to be long-winded – the consequence of being a land-lubber, I suppose.

    Thinking more about those AOA signals. I’m sure that is what is going on, and why the MAX ever got off the ground in the first place.

    As I understand AOA sensors, they typically consist of a rotating vane on an arm, on the side of the aircraft outside the cockpit. They simply align with the airflow, and send a signal indicating their position.

    During development and testing, no doubt the AOA sensors were all new and relatively stiff to move, due to new seals and bearings.

    Later, after a bit of use (and warm Indonesian & Ethiopian air), they would tend to loosen up.

    While still new, they would tend to stay in whatever position they were in at the end of the previous flight, probably both nearly horizontal. So, resetting the zero points while the plane was still stationary would probably work “okay”. Good enough to get by, anyway.

    However, after loosening up, this would not be reliable. For example, parked beside a gate, the tail is facing the taxiway. Another plane gets pushback, turns and heads out, sending a blast of jet exhaust over one side of the parked MAX 8. This acts backwards over the AOA sensor on the exposed side, and pushes it to its full extreme position, either up or down, depending on the eddies, the wind, locations of parked service vehicles and whatever.

    So, one sensor is now at full extreme, and the other could equally be anywhere, middle, down or up. If the software isn’t designed to deal with this…

    During “testing” this may never have happened so never showed up, and the awful thing passed and got airborne. Yuck!

  348. Victor Iannello says:

    @oddball: The AoA errors look like bit errors, which suggest something is not working correctly in the chain of digital processing, be it electronic or software. I doubt the problem is mechanical.

    I expect that the AoA sensor is designed with a slight torsional spring load which keeps the vane locked against a stop until aerodynamic forces align it with the airstream. I don’t know whether or not the low speed discrepancies are significant. However, once the vane aligns with the airstream, the offset between the two readings remains constant.

    If you accept that the readings should track at all speeds, it looks as though large motions are of opposite sign, but small motions are of the same sign, as if the higher order bits are complemented on the bad channel, but the lower order bits are correct. And if there is a problem like this, why isn’t it seen on all B737 MAXes?

    My wild guess is a software glitch in an A/D routine.

  349. oddball says:

    Sorry but I can’t resist!

    Addendum to Boeing 737 MAX 8 Operations Manual:

    Immediately before takeoff:

    Both pilots, open your side windows and stick your heads out.

    Make sure Left and Right AOA sensor vanes are horizontal.

    If not, use broom handle with wire hook to move vane to horizontal position. If it drops back down, wedge in horizontal position with a freshly chewed lump of chewing gum at the back end of the vane. This will shrink and drop off at high altitude. Refer to cabin crew for gum.

    Warning: Do not attempt to take off with both AOA vanes wedged with gum.

  350. oddball says:


    However, once the vane aligns with the airstream, the offset between the two readings remains constant.

    Yes, clearly, they are both tracking perfectly in flight, but for some reason, the left one is greatly offset from the right. It could be for either reason – yours or mine – but what we need is to see the traces from the moment of Power On. The traces show relatively minor changes in value for what must have been significant changes in AOA, so I don’t know how that offset occurred. It is interesting that they very nearly meet at the end of the trace, but who knows what happened to them in that instant. One went fully up, the other fully down? Maybe this was how they started out?

    Judging by the Left Stick Shaker, the sofware thought the Left AOA was too high, which aligns with the trace, but equally clearly, the Left side was way out of calibration.

    If they had swapped control sides, would that have altered the way the Trim system responded?

  351. oddball says:

    Uhhh… (self-answer)

    but who knows what happened to them in that instant

    Power voltages -> 0, so signal voltages -> 0. (Are they analog or digital?)

    But FDR internal power holds up and continues recording a little longer due to internal storage capacitors?

    Enough, later…

  352. Barry Carlson says:


    Shortly after the release of the JT610 DFDR trace I managed to acquire a higher resolution version, and came to the conclusion that the RH AoA vane after the A/D conversion had a differential of plus 21.6°. Knowing that it was impossible to fit the vane incorrectly, the probable difference was a factor of 16 in A/D conversion. The small difference of < 1° from the expected 22.5° could be attributed to roll angle / relative airflow.

    On the other-hand, I don't know what coupling method is used in detecting the vane shaft angle; possibly 3 phase synchronization.

  353. Andrew says:


    RE: ‘A couple of things I don’t understand – what does the trace for “Flap Handle Position” indicate? That they took off with flaps up? Then lowered, then the raised, then lowered them for the rest of the flight? Or is it inverted?’

    The trace is ‘inverted’ for want of a better description, ie a positive signal indicates the flaps are extended. I suspect Flap 5 was selected for take-off. The flaps were then retracted in stages, to Flap 1 and then Flap 0. Shortly after the flaps were retracted, they were extended to Flap 1 and then Flap 5, before being retracted a second time.

    If you have a look at the ‘TRIM AUTOMATIC’ trace, you can see the MCAS nose down trim commands (ie the longer duration pulses) began as soon as the flaps were retracted the first time. They ceased for a short time when the flaps were re-extended and re-started when the flaps were retracted the second time. That is consistent with the MCAS, which is not active when the flaps are extended.

  354. Andrew says:


    It is not uncommon to see aircraft AOA vanes sitting at all sorts of odd angles when the aircraft is parked on the ground. The vanes then align with the airflow when the aircraft accelerates during take-off. These things have been around a long time and, as Victor mentioned, they are normally very reliable.

    The problem could very well be software related, as Victor also suggested. It would be interesting to find out if there are any differences in the ADIRU mod states between different batches of aircraft in the 737 MAX fleet.

  355. oddball says:

    Sorry guys! This thing has got me wound up.

    Never paid attention to it before, but does anyone know what the scale on the RHS of the AOA traces indicates? Is the central divider supposed to indicate an AOA of 0°, and wouldn’t you expect 0° on both during the takeoff roll? If nothing else, they should both read the same during the roll, and if they don’t, shouldn’t it be flagged and the takeoff aborted?

    If both points are true, then it looks to me as if the LH AOA trace is about 4 times more above 0 than the RH trace is below 0.

    In other words, they are both wrong, with the LH one much more so.

    It seems to me the AOA is now a very critical item of flight instrumentation. Failure to correctly calibrate them at the start could account for these crashes (and others), given they can blow about freely when stationary on the ground, and the whole problem (In the MAX case) could simply be caused by a programmer who doesn’t know about AOAs, and simply called the “Reset/Calibrate” routine while still at the airbridge, instead of at a certain speed during the roll.

    This strikes me as a clear design deficiency, because of the inability to calibrate before the roll. Maybe some do, I don’t know, but I think all AOAs should contain calibration actuators, that return them to zero position, then to specified points, maybe +/- 75%, at power on (while stationary), just like a pilot tests his flight controls. If the readings are not within spec, then a Faulty AOA is immediately indicated. Not much good finding out 6 months later, but this has happened before – TWA843 (1992). And so simple a fix.

  356. Mick Gilbert says:


    Bill, it will be very interesting to see how this plays out for both Boeing and the FAA. On the basis that you can put a line through deliberate malicious intent this whole MCAS mess will come down to yet another confluence of those two enemies of safety; a failure of imagination and the normalisation of deviance. Those two do not operate in a vacuum. Nor do they just get together one night and make Rumsfeldian ‘stuff’ happen.

    It is going to take a very high level inquiry or a criminal investigation or both to get to the bottom of the this. In terms of the former, I see that Congressman DeFazio, chairman of the House Transportation and Infrastructure Committee, said he would hold hearings. In terms of the latter, there’s talk of a grand jury. That will all take time though.

    At the heart of the certification issue is ‘Organisation Designation Authorisation’. Another of those ‘good idea poorly implemented’ things.

  357. Mick Gilbert says:


    It would be interesting to find out if there are any differences in the ADIRU mod states between different batches of aircraft in the 737 MAX fleet.

    A very high level comparison between the two accident aircraft;

    Lion Air 610. PK-LQP. LN:7058
    First flight date: 30/07/2018

    Ethiopian 302. ET-AVJ. LN:7243
    First flight date: 30/10/2018

    So, the accident aircraft were built within 3 months of each other. While LSN is a better indication of production run placement than MSN, it’s still not ‘perfect’. Based on Boeing’s production rates there were probably 150 or so airplanes built in that three month period.

  358. Peter Norton says:

    @Oddball: “they should both read the same during the roll”
    not working below60 knots

    @Victor: “This suggests the problem was not mechanical. […] The AoA errors look like bit errors […] if there is a problem like this, why isn’t it seen on all B737 MAXes?”

    I was about to ask you the same question. Wouldn’t bit errors affect more aircraft than just LNI610? Barry Carlson mentioned, that it would be impossible to fit the AoA vane incorrectly. But this would explain why only this flight was affected. The FDR graph looks as though one vane was mounted with an offset, although I read that if you tried fitting the sensor offset by 20°, the attachment screws wouldn’t pick up on the captive nuts. But didn’t the maintenance crew remove (and then refit or replace) the AoA sensor prior to the fateful flight? I think I vaguely remember something to that effect.

    Maintenance Lapse Identified as Initial Problem Leading to Lion Air Crash

    « Crash investigators have concluded preliminarily that improper calibration of an airspeed [AoA?] sensor during maintenance touched off the sequence of events that led to October’s fatal Lion Air jetliner crash in Indonesia […] The suspected maintenance lapse was the initial misstep that ended with the months-old Boeing Co. 737 MAX aircraft plunging into the Java Sea, killing all 189 people on board. Two days before […] mechanics installed but failed to properly calibrate a replacement airspeed sensor called an angle-of-attack indicator. Based on information downloaded from the flight-data recorder, last month’s interim report revealed a constant 20° difference between signals from the AoA sensor on the captain’s side—which had been replaced—and those from the co-pilot’s-side sensor. »

  359. Victor Iannello says:

    @Peter Norton: We’re all guessing based on the limited facts at hand. The AoA is a reliable sensor. For JT610, likely it was factory-installed before it was replaced by maintenance. For ET302, likely it was factory-installed. The Reuters story said the behavior of the bad AoA sensor was the same for ET302 as it was for JT610. These assumptions lead me to doubt the problems are related to how the AoA sensor was installed or maintained.

  360. Victor Iannello says:

    Peter Lemme also considered a “bad bit” as a possibility for the AoA sensor problems on the B737 MAX, similar to what I hypothesized.

  361. oddball says:

    @Peter Norton,

    not working below 60 knots

    Many thanks for your comment and link. Obviously, I’m not fully up with the play, so that was very helpful. Yes, I didn’t expect the AOA vanes to move reliably until they reached a decent airspeed. This is just one of the problems…

    … mechanics installed but failed to properly calibrate a replacement airspeed sensor called an angle-of-attack indicator.

    Seriously!? You’re not kidding me? Is this REALLY the state of the art? Surely not!?

    I can go down to any old Auto Parts store and by myself an ‘Any Brand’ after market engine tachometer (rev counter) for my 4, 6 or 8 cylinder car, and fit it to just about ‘Any Brand’ of car, and expect two things: 1. It will work, and 2: It will be accurate to within a few percent.

    Yet a brand new, spare aviation part, for a specific model of plane, needs to be “calibrated” when it is installed, and, not having been “calibrated” can produce an error of 20°? To put that in context, the total swing of an AOA sensor must be no more than about 60°, so this represents an error of 33%? Can you elaborate, please?

    Also, if one sensor was replaced, why? Was it also “reading” a 20° error, and the fault was not in the sensor at all?

    What I can say now, assuming everything you guys are telling me is true, is that I am horrified. I am in utter disbelief. I think I never want to fly again!

    To me, the idea that, in 2019, a new, modern, recently (re-)designed aircraft – no, a commercial airliner – capable of carrying more than 20 passengers, can be designed and built so utterly badly, and sold to be relied on by the flightless, unsuspecting, fare-paying public to transport them at high altitude through the air, is just appalling.

    The fact that this fault could come down to something as pathetic as a “bad bit” in a digital data path is unbelievable, to me at least.

    Now that I have had time to “process” some of what I’ve been told about how the AOA system apparently operates, I’m disgusted. From the DFDR traces, there was apparently a very plain and obvious, major fault with the AOA system, right from Power On. If I have this correct, there are only two AOA sensors, and one of them was completely “out of whack” from the word go, AND YET, this out of whack sensor was being relied upon by a major flight control system, to repeatedly and continually countermand the pilots’ control inputs. Sorry, just appalling.

    I’ve worked in electronics for most of my career. I have some idea about its strengths and weaknesses. I know what amazingly cunning things you can do with it, and I know just how extremely fragile it is. Silicon chips are delicate, for several reasons. First, the transistors and other components built on them are microscopic, and second, silicon melts at fairly low temperatures. It’s highly “fry-able” and easy to “blow up”. Yes, things have come a very long way since I started, and in part that’s what makes this so unbelievable.

    One of the most unbelievable aspects of the modern aviation industry is that, at the most critical moments, so much is left to human pilots to try to figure out, under extreme duress, with insufficient time. The crash of AF447 is just one example. That “should never have happened.” Lion Air “should never have happened.” Ethiopian almost certainly should never have happened either.

    Putting AF447 aside for now as a completely different issue, the Lion Air flight should never have even got off the ground. It should never have even been allowed to commence its takeoff roll.

    And, most importantly, it should not have been the pilot’s decision to make. Nor should it have been the maintenance crew’s decision. On the ground, the plane was perfectly safe. In the air, it was simply an accident about to happen (not even “waiting to happen”). It was almost inevitable.

    And the decision about whether or not the plane could take off should have been made by the designer and manufacturer – Boeing, in this case.

    How? Very simple. Electronic smarts. Designed in.

    Let’s digress a little. I don’t know how long front loading washing machines have been around, but it’s a long time. I once had dealings with a big commercial unit that was so old it had no electronics in it. [FWIW, timing was achieved by several “Crouzet” style, geared-motor switches.] Yet, despite being almost antique, it was capable of multiple different wash cycles, controlled by a motor-driven plastic “punch card” system. You just inserted a different card in the slot, pressed Start and off it went. When it was ready, that is. Even though it had no electronics, the entire machine would not start at all if the front door was not properly closed. It had a simple, electric interlock, based on some clever mechanical bits and a micro-switch. And what is the worst that could have happened? A flood over the floor, and the washing not getting done. Obviously, a flood was sufficiently unacceptable that the mere possibility of one was deemed sufficient to “withhold service”. The interlock switch was built-in, to eliminate that possibility, and had complete authority. All without any transistors whatsoever. Even more than that, when the bowl was full of water, another interlock preventing the door from being opened.

    Not so, it seems, with modern commercial airliners. Yep, modern airplanes have redundant systems… So what!? When one of two sensors making up a vital redundant system is plainly faulty, the plane is still able to take off – despite not really being able to fly.

    The definition of redundancy is “the inclusion of extra components which are not strictly necessary to functioning, in case of failure in other components.”

    However, when one of two has already failed, there is no redundancy, and for a critical system like this, the plane should not fly. Especially when one of the supposedly redundant components is not even recognised as faulty. (The next question should be, can the failed component actually be switched out? And can it be done, mid-flight? But that is not my point. The first is that there was no recognition of a fault. Nor safety system to report and diagnose or control it. Let alone a notion of how to deal with it.)

    However, the modern jet engine is FADEC-controlled. Those impressive-looking throttle levers are now really just big, electronic “volume controls”, connected to some electrical wires.

    So, while still on the ground, it would be trivial for the Aircraft Safety Monitoring & Control System to say, “UNFIT TO FLY”, and prevent a pilot from applying anything more than taxiing power, or achieving anything higher than slow taxi speed. This would require a bit of intelligent design, but if an ancient non-transistor washing machine can be designed to be safe, so can a jet plane.

    I welcome any comments.

  362. oddball says:

    Oops: “…store and buy myself…”

  363. Peter Norton says:

    good points, Victor

  364. paul smithson says:

    @TBill. regarding WPT2 timestamp and the mean speed from there to beginning of Kota Bharu trace. There’s a discrepancy between my response above and my previous estimate last year which I’ll need to resolve. The previous estimate had within 1s match and speed ~457 (if memory serves) back to start of KB. The one above has 7 second difference and 451 knots. Of course both are “slow” so the conclusion isn’t fundamentally different. However, I’d like to avoid unexplained discrepancies and will try and track that one down. As you say, there’s an issue with working out the timestamps of WPT1 and WPT2 in the SIR.

  365. Mick Gilbert says:

    An interesting new twist in the Lion Air 610 crash.

    As the Lion Air crew fought to control their diving Boeing 737 Max 8 jet, they got help from an unexpected source: an off-duty pilot who happened to be riding in the cockpit.

    That extra pilot, who was seated in the cockpit jumpseat, correctly diagnosed the problem and told the crew how to disable a malfunctioning flight-control system and save the plane, according to two people familiar with Indonesia’s investigation.

    The CVR for that incident flight, JT43, would make very interesting listening.

  366. TBill says:

    fyi Gysbreght’s recent analysis suggests MH370 was flying without A/P from the turnaround to beyond KB. He is somewhat agnostic on the U-turn itsef but it tentatively looks like a >30% bank.

    I am thinking, if A/P is OFF, that could be because the pilot was trying to turn off all normal L/R Gen and L/R Backup Gen and L/R XFER buses, whereas that would turn off SATCOM and voice and data recorders, not sure about O2 masks. In other words, loss of A/P may support the intentional depressure hypothesis.

  367. paul smithson says:

    Tbill. Can you remember how Gysbrecht arrived at that conclusion? I see nothing in my analysis to indicate conclusively one way or another that the aircraft was on autopilot. What I do see is evidence that it was under control and flying a pretty steady course, albeit with subtle course changes along the way.

  368. TBill says:

    Gysbreght’s analysis is posted on JW last couple days, he and I have been discussing on the side. If you need me to, I can copy and paste to my Google Docs and/or Twitter for your access.

    As for my comments on aircraft power settings, I am a new user of the more rigorous PMDG FSX flight sim model that Victor uses. My main current interest there is fiddling with the electrical configs, but I realize PMDG may not be totally correct either. I still also use the older PSS777 FS9 for some things.

  369. Peter Norton says:

    @Mick Gilbert:
    RE: Non-revenue pilot in the cockpit jumpseat prevented crash.
    This makes the perfect case for a 3-pilot-cockpit, like in the old days,
    while zeitgeist goes in the opposite direction (towards a 1-pilot-cockpit or even 0-pilot-cockpit) …

    Any chance the LNI43 CVR content will be released you think (provided the direct relevance for LNI610 proves true) ?

Leave a Reply