On Sunday, MH370 family member V.P.R. Nathan presented several slides highlighting Ocean Infinity’s willingness to conduct another subsea search for the debris field of the aircraft. The search area, which may be refined later, extends along the 7th arc from about 33°S to 36°S, and out to about 45 NM (83 km) on either side of the arc. Ocean Infinity also plans to “fill-in” areas that were previously searched that had low quality or missing data due to equipment failures or challenging terrain.
The proposal is broadly consistent with the flight reconstruction and recommended search area that we presented in UGIB 2020 and the subsequent drift study. In fact, the “Last Estimated Position” (LEP) presented in those studies is designated the “IG Hotspot” in the slide. The proposed search area will also incorporate our high priority area due south of the LEP that was missed by the previous searches by the ATSB and Ocean Infinity due to the steeply sloping terrain in that area.
So, will Malaysia accept Ocean Infinity’s offer to conduct another search with the proposed terms of “no-find, no-fee”? At Sunday’s event, Transport Minister Anthony Loke expressed an interest in discussing the new proposal with Ocean Infinity. However, both he and Malaysian Prime Minister Anwar Ibrahim cautioned that a new search would be conducted only if there was “credible new evidence”. So, although we may be seeing progress towards a new search, it is unclear whether or not Malaysia will accept Ocean Infinity’s proposal.
@All
The distribution at 23:14 is extremely skew. It is roughly the upper half of a complete Gaussian distribution without any further truncation, and has an average value close to its smallest value (bizarre), while the distribution at 18:40 is an approximately symmetrically truncated Gaussian as expected during regular flight. The truncation on both sides comes from the temperature-stabilization, which suppresses large excursions with periods above 0.3 seconds (round off, turbulence and navigation effects excluded). A slight top truncation is due to insufficient compensation against small, fast excursions. In previously published plots of integrated BFO error distributions from 20 flights, there is also a long but very low probability tail towards negative deviations. I presume these predominantly come from cold starts in airports (e.g. a cold winter morning in Beijing) or from fast ascents during takeoff.
@DrB
Please look in the previous comment block for my reply.
@Victor
Interesting to watch search decision process. Loke seemed very well spoken and most encouraging. Strong, decisive, not wishy washy in the least.
@Viking
There are two issues with phone call BFO, slight amount of scatter in the 23:14 results, and in both cases it is a different channel so we do not know if the average value should have a different correction factor.
It is interesting to suggest two phone calls at 1840. The 1840 values are so smooth it tends to suggest steady flight (not descent in progress) but that is very subjective, unproven. The 2314 call is likely in the high winds below 22s so that is one thing, possibly descent in progress if it was a slow descent ~400ft/min (if it was active pilot).
@TBill
My analysis is insensitive to different channels unless one of them is hardware damaged, because it only relies on relative values. On the other hand, you are perfectly right about problems with comparing the absolute values.
@All
How is such a strange and unusual distribution possible at 23:14? The only realistic possibility is that the SDU operated under conditions outside its specifications. If it were due to events before 18:40 those would have affected data from the first phone call as well, so the unusual conditions occurred due to events after 18:40.
@Viking,
1. The truncated trig functions produce sawtooth errors, with peak-to-peak amplitude as much as 5-6 Hz.
2. Figure F-4 in UGIB (2020) shows a typical waveform.
3. Sawtooth errors have a flat-top probability density function with a full width equal to the peak-to-peak amplitude of the sawtooth.
4. It is not possible to reliably determine the shape of the probability density function of this error in one or a few phone calls. There are simply too few samples.
5. The oven temperature controller has a low bandwidth, and it does not have any frequency feedback, only temperature.
6. After initial temperature stabilization upon power-up, the contributions of OCXO temperature fluctuations to BFORs in a normal flight environment are negligible. That’s not to say the OCXO frequency does not drift, just that the drift is not due to crystal temperature changes.
6. The only remarkable thing about the first 2.5 seconds of the 23:14 phone call for MH370 is the very small fluctuation of 1 Hz peak-to-peak. That proves the random “read noise” is quite low.
7. The observed 6 Hz peak-to-peak range in BFORs at 23:14 is due to other error sources besides read noise, and it is probably dominated by the trig quantization, which in this case has a period of about 20 seconds.
You said: “It is also important to notice, that it would give more noise, not less as observed @23:14.”
I would not characterize the 23:14 phone call BFORs as having “less noise” since they have a 6 Hz peak-to-peak variation. The fluctuations in the first 2.5 seconds are observed to be quite small, but over this short period of time the change in the quantization error is generally expected to be less than 1 Hz anyway.
The DSTG’s empirical probability density function for BFORs is generally gaussian, as one should expect because the convolution of a number of different PDFs always tends to become gaussian. Observing a gaussian shape does not prove that one contributing PDF is not flat-topped.
@DrB
For 2314 are you able to throw out some points as being math noise? and thus arrive at a stable value estimate?
Hi Mr. Iannello,
Did you hear about Patrick Blelly and Jean-Luc Marchand study ?
What are your thoughts about this work ?
@Barry Carlson: This is true. You are right that for OI it’s certainly better to have a contract and rather be safe than sorry. But do you really think there is a chance that if OI finds the plane and announces this breakthrough to the media, that somebody could really come and “snatch” the plane (or parts of it) away ?
@Victor Iannello
@Barry Carlson: I have made the same remarks throughout the years, but I still get angry when I hear that Malaysia does not accept a zero(!) risk deal despite claiming to be “happy to reopen … whatever needs to be done must be done”. This is such an outrage. Victor, I know you have said in the past that Malaysia could have good arguments for their stance, but I can’t remember what you specifically mentioned and I still can’t see any really good ones. Why does Malaysia exact “credible new evidence” ON TOP OF a zero-risk-deal ?? In my mind either one is enough. Why both? Why demand anything when Malaysia bears no risk at all? That really beats me.
It’s a win-win-situation for Malaysia:
• Either the plane is found … which Malaysia claims to desire, so that would be a win for Malaysia.
• Or the plane is not found … then Malaysia has received the benefit of a free search and can tell the NOK and the whole world that they have done something, which is certainly a PR win (in addition to the free search).
So why does Malaysia not accept this zero-risk-deal ?
They are like “you can work for me doing an incredibly demanding, exhausting, difficult, very costly and maybe even somewhat dangerous job in the Roaring Forties (in case it’s not entirely remotely operated) but you have to pay for that AND on top of that we ask you to present to us evidence that you will find something BEFORE you find it”. This is just nuts.
Whenever I hear that Malaysia only accepts this zero-risk-deal if “new credible evidence” is presented to them, I can’t help but think that the Malaysian government doesn’t want the plane to be found. I find it hard to think of any other “credible” reason, because
a) there is absolutely no reason to demand for anything in return for a free search!
and
b) it’s so obvious that this hurdle of “new credible evidence” is almost impossible to overcome. First of all: What “new” evidence should all of a sudden emerge 10 years after MH370’s disappearance? The chances of that a very slim given the time that has passed and Malaysia obviously knows this.
And second of all: Malaysia’s wording is very clever. The word “credible” is very subjective and who is the arbiter of whether this standard is met? Right, Malaysia. So whatever scientific paper is presented to them, they can always argue that it is “not enough” to be sure. After all: You can only be sure if you have located the wreckage site.
So in fact the Malaysian government demands something almost impossible and traps offers such as OI’s current one into this circular reasoning: “Show me new hard evidence, such as the black boxes, and we allow a search.” But this evidence is what the seabed search tries to find in the first place, therefore this is a demand almost impossible to meet.
What if no new evidence will ever emergence ? (which is the most likely case given the elapsed time)
Then Malaysia will never agree to a new search – despite claiming they want to find the plane.
Sorry for getting emotional, but this stubborn demand for “new credible evidence” really appears like a shabby excuse.
@TBill,
There is no available method for selecting certain BFOs circa 23:14 as having greater “reliability”.
I can imagine two methods for predicting the “true” value:
(a) simply average all the phone call BFO values, equally weighting all data points, or
(b) fit a sawtooth by least squares to the BFOs, and use the average value of the sawtooth over the duration of the transmissions as the best representation of the “true” value.
There is not a large difference between the two methods, but Method (b) might be the best one can do with the information we have in hand.
In general, one can only apply Method (b) to the phone calls, because those are the only MH370 events with a sufficient number of BFOs in rapid succession to discern the sawtooth amplitude and phase.
@Peter Norton wrote:-
“I have made the same remarks throughout the years, but I still get angry when I hear that Malaysia does not accept a zero(!) risk deal despite claiming to be ‘happy to reopen … whatever needs to be done must be done’.”
I believe (from my own experiences) that there is a cultural adversion to even accepting that ‘they’ in some way or another may be culpable in the loss of the flight and the souls onboard. In addition, when things do go wrong, there is a distinct tendancy to ‘raise the drawbridge’ and walk away.
Finally, “we will” needs to be treated with caution, i.e. interpreted as “we might”, and also remember there is no such thing as a “free lunch”!
@Jérôme Chabot: They make some assumptions about pilot actions that I think are guesses, and their level of certainty is optimistic. However, their crossing of the 7th arc is not too far from the UGIB 2020 prediction, and their point of impact is within the area that we believe should be searched, so there is little reason to dispute their results.
@Peter Norton, @Barry Carlson
I basically agree with your sentiment’s gentlemen.
It was clear to many after only a few weeks that Malaysia did not want it found. I think it is clear that Malaysia still does not want it found, recent chest beating notwithstanding.
The early dumping of AMSA for Dolan’s ATSB, the ‘driving of the search north’, and the subsequent creation of the JACC (with Houston for PR ‘street cred’ cover), all had a faintly unpleasant odor about it, (to some of us local natives at the time) which seemed to be driven, more by the perceived need not to contradict, annoy or (God forbid) embarrass Malaysia, diplomatically, than with actually getting on with the job of finding the plane. It became a political / media circus.
The Chinese weren’t impressed either, especially when their IL-76’s were prevented from searching where their own satellites showed target of interest (deep south west end of the search area, plus a bit further). They packed up and went home pretty smartly after that little episode, and their official attitude has been one of ‘studied indifference’ ever since.
If you go back through contemporary ‘Senate Estimates’ hearings, (some of which are still available on YouTube) you will find that most of the time the dear Senators were asking more about ‘how much is this costing’ than anything else. In fact, they even questioned the ATSB’s choice of Fugro, (since they had ‘representations’ from some ‘aggrieved parties’ who strongly suggested that the ATSB had not selected the best offer that was tendered for the search contract), either on technical grounds, or on financial (best value for money for the Commonwealth) grounds.
It was in this political climate, that the phrase ‘credible new evidence’ came into being.
In all probability, it was undoubtedly created by a sly Australian public servant within the Australian Treasury, as a barrier, an impossible hurdle to hurdle, that would give the Government an excuse to stop spending any further money on any further searches.
The Malaysian Government loved it, and adopted it, as their ‘get out of (further searching) jail card’.
Money is still the public Malaysian issue.
In the multiple press conferences and short interviews the other day, the Malaysian Transport Minister kept repeating that any contract proposal had to ‘go to Cabinet’ because of the financial implications.
Most people mistakenly interpret that to mean the cost of the new search (if it occurs). Such costs would be a mere drop in the bucket, a few tens of millions, nothing more than loose change to a nation.
But that is only a ruse for the media and the NoK, it is not the real issue at all.
The real issue is that if the aircraft were to be found, and if the investigation proved that either the Airline or the Captain was culpable, the financial implications would be huge, and the political and diplomatic damage would be substantial.
I think it is clear that Malaysia never wanted 9M-MRO found, still don’t, and never will, for those reasons.
Now, that leads into their current problem, how to handle the building calls for ‘another search’.
At this point, I have to declare my hand clearly, (for those who may not be aware).
I am embedded in the deliberately piloted flight to the ‘deep south’ camp, specifically, into the GFZ.
I have never accepted that drift studies of the Flaperon confirm 35 South (give or take two or three).
I do not have a gut acceptance of WSPR technology either.
Now, since the Malaysia Government can not afford for MH370 to be found, (for the reasons above), handling calls or a new search becomes an exercise in political damage control for Malaysia, pure and simple, nothing else matters to them.
In my opinion, the Malaysian Government will only ever ‘reluctantly’ agree to a new search, (for PR brownie points only) in an area that their own information tells them it can’t possibly be, because they must be absolutely certain, in their own minds, that the search will certainly fail, and that OI will end up with egg on their face, yet again, before they sign any new contract.
They will ‘reluctantly’ sign contracts in the north, but only if they have to, for PR reasons alone.
Now, hypothetically, consider what would happen if the IG suddenly came out with some miracle reinterpretation of the ISAT data, that re-confirmed the absolute certainty of the initial positions near 38/39/40 South !!
Would all ‘the pilots’ suddenly get a fair hearing ? All of them are down there, Hardy, Keane, Bailey, and others.
(dropped off last post some how) !
Would OI consider searching down there ?
Would Malaysia allow it-or orbid it ?
It’s never too late for new evidence to be presented.
In the past I’ve been supportive of all scientific approaches to help narrow the search zone, and I fully acknowledge all the effort that has been put in over the last decade. I’ve made my points where it seemed necessary, but avoided dismissal of serious theories.
Allow me again to be the contrarian.
I’m now releasing a report on barnacle growth that excludes a cold water crash site. The paper was started nearly a year ago based on barnacle growth, but I’ve been distracted. It now incorporates new photo evidence that barnacle growth on the flaperon started AFTER the debris ran aground. Juvenile and adult barnacles are clearly observed to have grown for weeks atop worn off screw heads and abrasions that could not have occurred at sea.
The report is now on my website:
Barnacle Growth on MH370 Debris is Consistent with a 7th Arc Crash Site in Tropical Waters
I don’t mean to thwart search efforts based on old evidence and past assumptions. The search needs to resume at any and all viable candidate sites.
I’m simply pointing out that the barnacle evidence
which has previously been dismissed as inconclusive and inconsistent with a cold water crash site is fully explained by a crash at the Java site. The matching acoustic evidence came after the search was suspended, and can also be considered new evidence for a specific location.
Malaysia has just called for “compelling” new evidence. That seems like a political rather than evidentiary new threshold.
I can only hope that the Java candidate site with a relatively tiny area of a few square km can help in compelling new search.
@TBill and @DrB
WOW, that was a difficult challenge – particularly to answer in a pedagogical way. The short (not pedagogical) answer is that my first reply to DrB is enough. Sorry, it took me a couple of hours to come up with a pedagogical answer. I came up with one when I accidentally looked at my Globus.
Try to place yourself at roughly 6 times the radius of the earth, over the equator, slightly south of India (analogous to the relevant Inmarsat satellite). Then remember that the classical Doppler shift is only dependent on the projected part of the airplane speed along a straight line of contact. This goes roughly proportional to the third power of the distance along the surface of the earth, and the average (as well as the typical) distance on the route from Beijing to KL is roughly 1.5 times longer than for any solution to the 7th arc in the Indian ocean. Since the speed with which the signal moves along the saw-tooth curve is also roughly proportional to the Doppler effect this means that the BFO value changes at least 3 times more slowly there. In other words, the MH371 data is not representative for what happened during the two attempted phone calls.
There is only small effect during 1 minute, probably 1-2 Hz. This may explain why the last few points lie slightly higher than my analysis shows, but it is not changing any of my conclusions.
On the other hand, it does have significant impact for longer time intervals.
@370Location
I am looking forward to reading your new paper. It is in good agreement with my own conclusions based on scientific results from DTU Aqua (from research on global warming effects on barnacle survival and growth).
@All
For handling the statistical distribution of BFO values, it is necessary to understand the underlying reasons for unusual behavior. In order to obtain a good and stable technical performance at ambient conditions one uses a general layout for control boxes (including the SDU) placing the stabilized oscillator as far as possible from the power supply and the amplifier/microwave-source, producing large and variable amounts of heat, which is a dominant error source for periods below 1 minute. The distance reduces thermal crosstalk, which is essential for good and robust performance. It also introduces a delay in the reference frequency drift (affecting the BFO) compared to the power output (heat-release-time).
@Victor
Assuming it was not a fish-that-got-away story, they should probably try to spend a few days finding that wing…it’s a wing and a prayer but it is “new” evidence. Chances are low but if MH370 that could help solve.
@370Location
I have now read your paper. I agree with most of what you write, but have a few comments:
First, I find part of the introduction a bit confusing. I am not saying it is wrong, but some people may take advantage of the way you phrase things, and use this to ignore your work. I say this only to help you.
Concerning the pure technical content I would like to add that from the scientific publications by DTU Aqua it is clear that all types of barnacles they investigated showed an Arrhenius temperature behavior with the same exponential factor but with 4 (individual for each type) threshold temperatures.
T1: Minimum survival temperature
T2: Minimum normal life temperature
T3: Maximum normal life temperature
T4: Maximum survival temperature
Between T2 and T3 they grew according to Arrhenius (of course with individual front factor depending on species and food available)
Between T1 and T2, and between T3 and T4 they needed time to adapt to uncomfortable conditions (typically 2-4 weeks with no growth). After adaptation, they grew according to Arrhenius again.
From your paper T2=18degrees, T3=25degrees and T4=34degrees
This may change some of your conclusions.
@All
In addition to the half-Gaussian data distribution, there are three other differences between the data sets. As mentioned above, the 18:40 data consists of two attempted phone calls shortly after each other, while the 23:14 data is from one attempt only. This becomes clear by plotting them in the same figure, showing that the 18:40 set has additional interleaving points shortly after the beginning. Later the two sets have the same time-structure, and both span approximately 1 minute.
The next difference is that the thermal delay (for heat-transfer from the ‘heat-release end’ to the oscillator) is much longer at 23:14 than at 18:40. This is particularly clear in the beginning, where the BFO value remains practically unchanged for a long period after the amplifier starts sending signals at 23:14 while the BFO values start going up shortly after the amplifier starts at 18:40.
@Viking wrote:-
“… the thermal delay (for heat-transfer from the ‘heat-release end’ to the oscillator) is much longer at 23:14 than at 18:40 etc..”
The unredacted Inmarsat IOR sat data was analysed by Mike Exner in a post he made on 2017-06-12. His referenced portable document file is to be found at:-
https://tinyurl.com/6dypr2zz
In short he found that the Carrier to Noise levels (C/N0) were commonly low at the initial logon by the aircraft on both the IOR and POR satellites.
When you refer to “amplifiier”, I suspect you mean “transmitter”. The OCXO is unlikely to react to the TX turning on as you suggest. The OCXO is maintained at about +75°C and its power-supply is well regulated. However, if you read onward from the 2017-06-12 post the possibility of the SDU being off for a period until shortly before the 1840 logon is raised, and this will more likely relate directly with your initial BFO errors.
@All
Last, the temperature stabilization for the oscillator does not attempt to compensate the heat-transfer at 23:14, giving a dramatically longer settling time (for effective heat dissipation from the oscillator), and a long upper tail for the statistical distribution with absolutely no truncation from the stabilization. This is very unusual, and indicates a serious malfunction of the SDU.
@Barry Carlson said: The OCXO is unlikely to react to the TX turning on as you suggest.
It should also be noted that the SDU and HPA are separate LRUs, and the heat generated by the HPA should have little effect on the stability of the SDU’s oscillator.
@Barry Carlson
The analysis by Mike Exner predominantly discussed the authenticity of the data and the C/NO. I agree that there is nothing strange about these things. Everything looks normal.
When I say amplifier, I include all power consuming (and heat producing) parts, so it would be better to say Transmitter. However later in my analysis it is the amplifier-part which is most important. Sorry for the initial confusion.
My approach is different from Mike’s. I am looking into the distribution of BFO and power levels as a function of time during the two attemted phone calls. There I found no indication that the fact that the SDU was off for some time prior to 18:25 had measurable effects on the data collected at 18:40. This indicates that the cabin was not particularly cold during the ‘SDU off’ time. The pressure may well have been low prior to 18:25, but that was also back to normal at 18:40.
During both phone calls the power was falling sharply (roughly 10dB). I am pretty sure this was predominantly due to heating inside the amplifier. I will discuss the details later.
@Victor Iannello
It looks like our contributions ‘crossed’, so I saw yours much later.
It is most likely that the amplification happened in two stages. As far as I remember the most powerfull single-stage amplifier I had in 1990 had 40dB gain and 10W output. With those specifications it had great trouble with overheating causing automatic shut-down and 10 minutes waiting time for cooling. That would be totally unacceptable for the SDU. The correct design would therefore have a preamplifier inside the SDU box (responsible for the thermal crosstalk) followed by the HPA in a separate unit.
Sorry I forgot that important detail.
Therefore, please read preamplifier where I initially wrote amplifier or transmitter.
@Viking
Follow your motto “The devil is in the detail” a question:
Is there / will there be a paper (incl. references where possible, and with sufficient technical details) substantiating your findings? I think it is much easier to properly give feedback and have a discussion if there is such a detailed paper instead of a series of slightly descriptive posting.
@ALL: Regarding the phone call BFO values…As Victor noted above, the OCXO frequency is not affected by the heat from the HPA. The OCXO thermal control system maintains a very constant temperature (whole circuit including the xtal), regardless of the ambient temperature over a wide range (-40C to +55C). For the 1 minute period related to a phone call, the change is absolutely zero.
@DrB
I was checking your App. F.3 “BFOR noise measurements” from the “UGIB” paper.
It states:
“All BFORs in this plot use the same BFO Bias of 150.0 Hz for prediction, and we have subtracted 16Hz from the MH371 BFOs to compensate for the 16 Hz step reduction applied in the SDU at 16:00:25 (as a result of an automatic frequency calibration procedure) when at the gate prior to MH370 push-back.”
How did you know this “16 Hz step reduction”, was it explicitly logged or was it somehow extracted?
@Niels: We talked about this at length in the 2017-2020 timeframe. The log-on at 16:00 showed a BFO jump of around 16 Hz, which is consistent with information from the ATSB about the bias adjustment from the frequency calibration algorithm.
@Victor
Thanks Victor. I remembered the 16 Hz jumps and auto-calibration being discussed but didn’t remember the specific details -> I’ll have a look at the unredacted logs again.
@Niels: When I was in London Dec 2017, I had a discussion with the guy from Thales about how the AES maintains the OCXO calibration. He was a little vague about the details, but basically, the AES recalibrates the frequency offset from nominal frequency by periodically measuring the L band downlink offset when the plane is on the ground (as it did at 16:00Z). It recalibrates each time at least 26 hrs have passed since the previous ground logon calibration. So it’s not done on an exact schedule.
Rockwell holds a patent on a method that uses the L band downlink offset to calculate the required uplink offset continuously. That’s why Honeywell cooked up the scheme they use. Note: 16 Hz is ~ 10^-8 * carrier frequency.
@Viking,
1. The BFO fluctuations during the MH370 phone calls are consistent with and fully explainable by the quantized trig computations in the SDU of the pre-compensation Doppler shift.
2. There is no need to hypothesize that rapid BFO changes up to 4-5 Hz occurred over a period of several tens of seconds or less caused by OCXO temperature fluctuations. Nor is this even possible after the power-up transient has settled.
3. All the BFO-sequence examples I have seen from MH370 and MH371 show the same peak-to-peak amplitude near 5 Hz. This is to be expected when the aircraft is in typical flight, as I have explained previously.
4. Only the “frequency” of the saw-tooth varies among the calls, depending on the time rate of change of the cosine of the angle between the aircraft velocity vector and the line of sight from the aircraft to the satellite.
5. In fact, uniquely determining the value of this angle can be done solely from the slope of the measured BFOs between the “steps” occurring when the pre-compensation Doppler changes to a new value. Note that this calculation is independent of any assumed BFO equation or model. In fact, this BFO slope is the time derivative of the ACTUAL Doppler shift.
6.This change in the ACTUAL Doppler shift (i.e., the 5 Hz sawtooth step amplitude) occurs in the same time as the cosine of the LOS-aircraft angle changes by a fixed increment (i.e., the SDU computational increment). This allows the LOS-aircraft angle to be uniquely determined. However, there is a “mirror symmetry” of routes with the same Doppler shift and therefore the same LOS-aircraft angle, so in general there is a pair of route solutions for each phone call.
7. This slope technique is a new method of checking the viability of any proposed MH370 route, at least at the times of the phone calls. It won’t tell you BOTH the location and the course (track bearing), but it relates them in a numerical way such that if you assume one, you can find useful limits on the other. So, the combination of a predicted location and a predicted course can be checked for consistency, and this check is significantly more discriminating for the sequence of phone call BFOs than what is possible at those times when only an isolated BFO measurement is made.
@DrB: That is VERY clever! During the times when the Doppler pre-compensation is constant due to the discrete steps associated with the trig function, the slope of the BFO is essentially the rate of change of the true Doppler shift.
I really doubt anybody has considered this before.
So with that in mind, it would become harder to spoof the BFO with fake satellite inclination parameters so that a northern path has the BFO signature of a southern path, as the slope of the sawtooth would be different for a northern and southern path, even though the BFOs might be very similar.
@DrB, Victor
Very interesting idea indeed!
I just looked at the logs and if the aim is to determine the d(Doppler)/dt around 23:15 can’t you just remove the steps / staircase (and thereby effectively the compensation component). This would allow to fit on all available points for the slope.
Now the step height would be a (weak) function of the aircraft position -> you get a measured d(Doppler)/dt which strictly speaking slightly depends on aircaft position (through the removed step height). For each path you can calculate the the d(Doppler)/dt, which for given moments in time (and hence known satellite velocities) would depend on aircraft velocities, which is known for each path tested. For each path tested the aircraft position is also known and hence the measured d(Doppler)/dt more accurately. The two can then be compared. Does that make sense?
@alsm
Thanks Mike. So the conditions for the calibration to occur are time elapsed > 26 hours and of course being on the ground. In theory I guess it could be multiples of 16 Hz.
@Niels
I am considering to make a complete new publication, but as the many comments show it is a complicated matter, and therefore I prefer taking the discussion first.
@DrB
I considered using the method you proposed on BFO derivatives (or a similar method using BTO derivatives) already in 2016. At first glance it is an excellent idea. However, after some thinking I realized that the resulting uncertainty is too large. There are a couple of reasons for that. First of all the intrinsic 1Hz resolution limits the precision. Basically you take two large numbers (each with 1Hz resolution) and subtract these. This gives a large relative uncertaity. Secondly, the round-off error (discussed yesterday) severely complicates things, and roughly doubles the uncertainty. Thirdly, the method is potentially very sensitive to height changes.
@All
I will reply to your comments very soon (mostly today). I am going to merge some of the comments/replies to avoid repeating too many things.
@All
Concerning heat release from a pre-amplifier:
I agree with your comments on the heat release from the HPA. This one is almost certainly well temperature stabilised (probably military grade) and isolated form the SDU box. However, heat from a preamplifier (e.g. 33dB gain with a few W output) is sufficient to causing thermal crosstalk.
Just listen to this personal observation: Someone with limited knowledge of HF amplifiers borrowed my 10W device in 1990. He placed it on a plastic chair and pushed it hard to get maximum output. After a few minutes the chair melted (just from the heat coming out of the ‘exhaust’ from the fan), and only the emergency thermal shut-down prevented it from falling on the floor or starting a fire. After this we had to throw the plastic chair in the bin. Fortunately it was a cheap one.
@Victor
I still claim B777 BTO/BFO spoofing is impossible. JJW had an expert on a couple weeks ago, who talked about the conditions necessary, including wheels down, changing out memory chips, etc. This spoofing rumor is just conceptual with no proof anyone could do this on a real B777. Let alone have it not be obvious from the sat data logs. Hitchen’s Rule applies that which can be claimed without evidence can be dismissed without evidence.
@TBill: Yes, the penetration vector proposed in that episode was ridiculous.
@Niels,
You said: “. . . can’t you just remove the steps / staircase (and thereby effectively the compensation component). This would allow to fit on all available points for the slope.”
Sure. For MH370 at 18:40 there are about 19 steps of about -5.6 Hz and 3.1 seconds each, so its slope is roughly about -106 Hz / 59 seconds = -1.80 Hz/second. Fitting the actual data with the compensation steps removed to see the actual Doppler shift (plus a constant) allows a reliable determination of the slope (and its error) by least squares fitting. Note the slope at 18:40 is negative because the positive Doppler shift (caused by the aircraft moving toward the satellite) is getting smaller with time.
I will also note that one can “fit” the pre-compensation steps to each BFO data set by adjusting the amplitude, period, and phase so as to minimize the least-squares residuals of the straight-line fit. What this process does is minimize the discontinuities when the steps occur.
It’s also necessary when finding the Doppler Angle ” (i.e., what I call the angle between the aircraft velocity vector and the line of sight from the aircraft to the actual satellite location) from its sine value that the proper quadrant is chosen. For instance, at 18:40 the Doppler Angle is in the range 0 to 90 degrees, but at 23:14 it is in the range 90 to 180 degrees. The quadrant is chosen based on whether the aircraft track is reducing (0-90 degrees) or increasing the satellite range (90-180 degrees).
At 23:14 we have 3 steps of roughly -5.6 Hz each about every 21 seconds, so its actual Doppler slope is about -16.8 Hz / 63 seconds = -0.27 Hz/second (7X smaller than at 18:40). At 23:14 the Doppler shift is negative, because the aircraft is moving away from the satellite) but the slope is again negative because the Doppler shift is becoming more negative with time.
The actual BFO slope is proportional to the time derivative of the cosine (which is therefore minus the sine) of the Doppler Angle, That means at 23:14 the slope is small, so the aircraft is flying less obliquely to the tangent of the “arcs” than it was at 18:40, when it was flying more or less directly toward the sub-satellite location and the magnitude of the BFO slope was much higher.
You also said: “. . . Does that make sense?”
Yes. It’s also worth noting that one does not need to know the satellite velocity vector in this slope method, so long as you assume it does not change appreciably during the 1 minute phone call.
@Viking:
You said: “. . . the method is potentially very sensitive to height changes.”
I think you mean the vertical speed. Sure. The vertical speed must either be known or assumed to be constant (and not necessarily zero).
You also said: “First of all the intrinsic 1Hz resolution limits the precision. Basically you take two large numbers (each with 1Hz resolution) and subtract these. This gives a large relative uncertaity .”
Over the duration of a call, about a minute, the change in BFO is -106 Hz at 18:40 and -17 Hz at 23:15. It is true that the data are quantized in 1 Hz increments. However, the impact of that quantization is a best-fit line deviation of 1.0 / SQRT(12) = 0.3 Hz per point. So, the quantization error is actually not a limit in the uncertainty of the slope, because it is considerably smaller than the RMS read noise per data point.
As Niels described, we can simply add the pre-compensation steps to the data and then fit a straight line to the “uncompensated” actual Doppler shift. That will give the slope, and we can then find the uncertainty in the slope from the fitted residuals. That slope uncertainty can then be converted to the uncertainty in the sine value, and that can be converted to the Doppler Angle uncertainty. So, we can solve for the Doppler Angle and have a reliable estimate of its uncertainty for each phone call by “removing” the pre-compensation steps and fitting a straight line by least squares.
@TBill (and Victor)
I agree completely. Brute-force BTO/BFO spoofing as described by JW is impossible in the real world, in particular after 9/11 security measures.
However, a special type of BFO spoofing may in principle be possible using an analogy to methods developed to compromise cryptography. I will discuss it in more detail later.
I do not believe it was used for MH370.
@DrB
I do not see how you arrive at those values for the change in BFO within 1 minute.
@DrB
I think you forget that the large change at 18:25 was due to a re-booting effect.
– and
I do mean sensitive to vertical speed (not absolute height).
@DrB: I said During the times when the Doppler pre-compensation is constant due to the discrete steps associated with the trig function, the slope of the BFO is essentially the rate of change of the true Doppler shift.
What I said is not exactly true. When the AES pre-compensation is held constant, there will indeed be contributions to the slope of the BFO from changes in the uplink Doppler (L-band), but also changes in the downlink Doppler (C-band) and changes in the EAFC correction. Nonetheless, all of these are quantifiable for a given time and position/velocity vector of the aircraft, so some discrimination of paths should still be possible.
It won’t surprise many here if I reiterate that I believe the aircraft will ultimately be discovered at +/ – 39.6 south, on the 7th arc because that is where BTO error optimisation tells us it ought to be. And at the end of the game, I think that most here agree that BTO is the best among an otherwise fuzzy heap of evidence to go on.
@Dr B. I’m also intrigued by your sawtooth derivative as a novel analytic method, independent of assumed BFO bias drift or absence thereof. I’ll be happy to submit my path-predicted parameters at phone call times for you to run those numbers.
@Victor,
During the duration of the 18:40 and the 23:14 phone calls, the predicted BFOs without the actual Doppler shift change by 0.4 Hz or less when neglecting the SDU quantization effect. Therefore, assuming the sum of all the other terms, except for the actual Doppler shift, is constant is a very good approximation and introduces no material error.
@Viking,
You said: “I think you forget that the large change at 18:25 was due to a re-booting effect.”
No, I didn’t forget about the OCXO transient on warm-up. Six years ago in 2018 I demonstrated the warm-up transient settles out in 3 minutes. So, after 18:28, the transient is gone.
The 2018 transient analysis is available here.
Therefore, the warm-up transient is long gone by 18:40 when the first phone call came in.
You also said: “I do not see how you arrive at those values for the change in BFO within 1 minute.”
After I get the code running and tested, I will prepare and post some figures to demonstrate the application of the BFO slope method to three phone call events. These will make it clear that the BFOs actually follow a sawtooth waveform within the normal read noise.
@Paul Smithson,
You said: “I’ll be happy to submit my path-predicted parameters at phone call times for you to run those numbers.”
I can do that once I have the bugs out. It would be convenient if your inputs were in the format in my BTO/BFO model available here.
@DrB
@Victor Iannello
Gents, I am more than happy to ‘fess up to a bit of a “dog watching TV” moment here regards the recent discussion regarding (and I’ll use Paul’s words here) using the saw tooth derivative as an analytical tool for the phone call BFOs. I’m picking up the palpable excitement and there’s a smattering of words I recognise but that’s followed by that thumping whoosh of most of it passing overhead.
Any chance of CliffsNotes or “Saw Tooth Derivative as an Analytical Tool for the Phone Call BFOs for Dummies” please?
@TBill:
In your comment, what “new evidence” are you referring to ?
@370Location:
« Malaysia has just called for “compelling” new evidence. That seems like a political rather than evidentiary new threshold. »
My feelings exactly.
@ventus45:
« The real issue is that if the aircraft were to be found, and if the investigation proved that either the Airline or the Captain was culpable, the financial implications would be huge, and the political and diplomatic damage would be substantial. »
That might very well be the reason.
« The Chinese […] when their IL-76’s were prevented from searching where their own satellites showed target of interest (deep south west end of the search area, plus a bit further) »
I must have missed this or forgotten. I couldn’t find any source for that. Do you have one?
Out of curiosity, I took the BFOs from during the two phone calls, fitted a quadratic, and made the following figure:
https://drive.google.com/file/d/1bQd2mXgAOJSTvNOttvfY_vcg59-BUtM7/view?usp=sharing
(BFOs from 23:14 have had 140 hz subtracted to fit on the figure, and Delta-UT is UT in hours relative to the average of all UTs during each phone call.)
A quadratic fit works well. The rms scatter is actually lower for the 23:14 call (0.8 v. 1.1 hz).
I take no position on the interpretation of these results or what relevance they have to sawtooth quantization.
All. Some selected recent media reports:
– The ‘IG’ opinion finally gets some media attention:
https://www.9news.com.au/world/mh370-where-is-missing-plane-on-7th-arc-and-will-we-ever-find-malaysian-airlines-boeing-777/e6afa618-dd20-4265-81fd-dc8a62054aaf
– At least as announced, the Australian support for a new search is limited to “technical information”, neither funding or support to OI being mentioned:
https://www.foreignminister.gov.au/minister/penny-wong/media-release/mh370-10-years
– In Oz we haven’t seen the BBC show yet, which might include the gist of the below, but I see in at its 3:45 on that Liverpool Uni will not have concluded its wide study into the merits of WSPR claims until October/November.
OI might be awaiting that before evincing any enthusiasm for it:
https://www.youtube.com/watch?v=9SEMSQDO-pg
@DrB: Here’s a test case. At 18:40, we have for the UGIB path (assuming my quick calculations are accurate): LLA = (7.5180 deg, 94.5830 deg, 16,300 ft), GS = 337 kn, VS = -1840 fpm. According to my calculations, the rates are: Uplink Doppler = -1.52 Hz/min, Downlink Doppler = 0.48 Hz/min, AES pre-compensation = 1.48 Hz/min, EAFC = -0.17 Hz/min. Therefore, the BFO rate is 0.26 Hz/min with the AES pre-compensation varying. If the AES pre-compensation is held constant, the BFO rate is -1.21 Hz/min, which is less in magnitude than the Uplink Doppler rate (-1.52 Hz/min).
[These numbers were updated since the original comment was submitted.]
@David: I really haven’t tried to promote the UGIB 2020 path and the associated recommended search areas in the media the way many have promoted their paths. In fact, I shun most media contacts because few have the capability and willingness to accurately cover our work. However, I assure you that anybody that is seriously considering re-starting a search is well aware of our work. This blog has always been geared towards technical discussions among informed investigators. The target audiences of the media and the other blogs are much different. That’s part of the reason that extraordinary technical claims go unchallenged.
@Mick Gilbert: We know that the BFO varies with time, position vector, and velocity vector. But now we have potentially a new parameter (BFO rate without AES compensation) discovered by Bobby that also varies with time, position vector, and velocity vector. We have this new parameter because the AES compensation is theorized to change in discrete steps, and the BFO rate between discrete steps can be measured. This new parameter is available from the burst of BFO data we have at the time of the phone calls. So, we are exploring whether we can use this data to further constrain the set of hypothetical routes. I’d say the jury is out until we study it further, but it may indeed prove useful.
The link below is a feasibility study for estimating a kinetic energy budget and its potential for generating seismic energy for a (hypothetical) forward fuselage section impacting the seafloor.
https://www.dropbox.com/scl/fi/n77um4ph01ltwxgll1bn0/MH370-Feasibility-Study-of-Radiated-Seismic-Energy-Generated-from-a-Forward-Fuselage-Section-Rev-0.pdf?rlkey=grgxagcmiel5nuvel67nxjsgm&dl=0
@Victor Iannello
Thank you for that, Victor.
https://www.linkedin.com/posts/ocean-infinity-llc_dave-field-managing-director-for-australia-activity-7170712536013619201-qJv6/?utm_source=share&utm_medium=member_desktop
@DrB
First of all:many thanks for your feedback on my posting yesterday.
Earlier you wrote:
“This change in the ACTUAL Doppler shift (i.e., the 5 Hz sawtooth step amplitude) occurs in the same time as the cosine of the LOS-aircraft angle changes by a fixed increment (i.e., the SDU computational increment). This allows the LOS-aircraft angle to be uniquely determined”
How do you actually know that the SDU computes the relevant inner product of vectors this way (using the cos of the LOS-aircraft angle)?
Based on available TT, GS, Lat, Lon there could be different ways to implement the calculation of the f_comp in a basic processor. For example one could just use add, subtract, multiply and look up tables for sin and cos and calculate using those and avoiding the need to use a divide routine, square roots, or to first solve for an angle.
Also, thinking about it a bit more, I would just use a spherical earth approximation, and indeed use a stationary satellite position, as they do. But then there could be many different cos and sin’s involved with their discretized look-up tables.
@DrB
The “many different cos and sin’s” would typically be sin(lat), cos(lat), sin(lon), cos(lon), sin(track), cos(track) in one possible implementation I have in mind.
@Niels
Thanks for making the computational details clear. In light of incomplete information from Inmarsat on their detailed software I just took the statistical approach.
@Victor Iannello
Concerning your numbers published on March 8 at 9:19pm.
Your numbers agree nicely with mine. In my statistical approach this is abbreviated as 1-2 Hz (as I wrote two days ago).
@DrB
As a further follow-up on your comments and proposed method, I actually spent some time finding an optimum combination of such methods (including both BFO and BTO) before publishing my first paper. The results are summarized graphically in figure 2 of my paper.
From this I could conclude one important thing. The final major turn (FMT) happened shorty after 19:41, and it was close to a U-turn (because of the sign change).
@Victor Iannello: « it would become harder to spoof the BFO with fake satellite inclination parameters so that a northern path has the BFO signature of a southern path, as the slope of the sawtooth would be different for a northern and southern path, even though the BFOs might be very similar. »
Harder or impossible ?
Does that mean a northern path can definitely be excluded now based on this ?
Have you determined if the sawtooth does indeed differ for a northern and southern path ?
@TBill says: « spoofing […] is just conceptual with no proof anyone could do this on a real B777. Let alone have it not be obvious from the sat data logs. »
How would spoofing manifest itself obviously in the logs ?
@Paul Smithson
You claim using only BTO values the optimum solution is near -39S. This is not correct. First of all, BTO alone does not give a good basis for finding an optimum solution. It basically just defines the 7th arc.
If one insists using BTO only, there are two ‘solutions’. One is slightly north of your number (near -37S as far as I remember – it takes a lot of mathematical skills to calculate). It is an extremely shallow and broad solution on the limit of mathematical stability. Actually, I think it is not stable.
The other BTO optimum is near my solution. It is narrow and deeper, but I would not call it good.
As DennisW once wrote: The problem is underdetermined. More information (than just BTO) is needed. I think DennisW actually claimed the entire problem was underdetermined, but I disagree on that.
@Peter Norton
A perfect spoof would be invisible in the log, but it would lead the investigators to wrong conclusions.
The only ways to identify a spoof are:
1) In the log, if the spoofer made one or more small mistakes.
2) Using other information (biofouling, beaching pattern, seismic data, contrails, eyewhitness reports, etc.)
@All
I think it is now time to proceed according to my original schedule:
For the 23:14 data, it looks like the thermal crosstalk-delay increased 3-4 times compared to the 18:40 data. This is particularly clear in the beginning of the data sets. Everything looks normal (in terms of induced noise and delay) at 18:40, but at 23:14 there is practically no noise until much later.
In addition, the overall re-cooling time (seen later in the data sets) increased roughly 10-15 fold. How is that possible? There are three possibilities according to my experience with similar boxes for controlling semiconductor-lasers or optical detectors:
1) The SDU was so cold and/or moist that significant amounts of frost or water had condensed on the surface of the frequency-oscillator
2) The cabin pressure was reduced 3-4 times (to below 0.25 bar)
3) Somebody had hardware-modified the box.
We choose to discard option 3, since nobody on board had the necessary technical knowledge, no realistic motive, and certainly not sufficient time (and no adequate equipment because it would not get through security after 9/11) to make such modifications. For a corresponding laser box, it would have taken me around 1 hour. Somebody without daily experience (only general knowledge of electronics) would need one day.
Peter Norton asked: Harder or impossible ?
Work in progress. I’d say the merits of this new approach are far from proven.
@Niels said: The “many different cos and sin’s” would typically be sin(lat), cos(lat), sin(lon), cos(lon), sin(track), cos(track) in one possible implementation I have in mind.
I think that if the waveform observed is truly due to changes in AES pre-compensation in discrete steps, there are other possibilities, such as the finite resolution of the output of the pre-compensation algorithm or the frequency output of the SDU, and not necessarily discrete approximations of the trig functions.
@Viking: There is no “thermal crosstalk”. We need to ditch that idea. The OCXO frequency is not measurably affected by the heat generated in the HPA (or any other AES component). In fact, the HPA and SDU (which contains the OCXO) are in separate LRUs separated by several inches. The thermal control system used in the OCXO is capable of maintaining a constant circuit temperature over a wide temperature range (-40 to +55C). Over a 1 minute period (or an hour), there is absolutely no possibility that HPA heat could significantly change the OCXO frequency. See a small sample of my OCXO and related files here: https://bit.ly/3It0OJx
@Viking I agree that the problem is underconstrained for bto optimisation if you start on arc 2. Because to do that removes all practical latitude discrimination. If you include a requirement to join up with the last known path and no loiter then you do indeed have a BTO optimized solution space. It stretches from roughly 38.5 to 39.8 south. Dstg and IG mark-1 were biased a little north of there (circa 38S) because they sought to optimise BTO & BFO combined.
@Victor Iannello says: « Work in progress. I’d say the merits of this new approach are far from proven. »
In my humble opinion it would be very exciting if you could show that the sawtooth differs for a northern and southern path and thereby proving with mathematics alone that only 1 of these 2 paths is possible – completely independently of the found debris which some claim was planted.
By eliminating these other theories (presented in the Netflix documentary or even on CNN this week highlighting the WSPR theory) it would help focusing the search efforts.
@Victor:
I think it would be advisable to redact the private parts of the link posted by @ventus45 in this comment because it reveals his full name and email address.
@Peter Norton: OK.
@Victor
Besides questions about the exact origin of the saw tooth waveform in the BFO data, I have a bigger worry: By thinking about how a most simple implementation of the pre-compensation calculation based on Lat, Lon, Track and GS as input could look like (and considering that the pre-compensation does not need to be fully accurate for the system to function), I wonder if the eq. 5.9 in the DSTG book exactly represents what is implemented in the hardware. Spheroid vs. elliptical earth model is just one example of possible subtle “shortcuts” in the implementation. Another typical point is how the ac-sat distance (the denominator in eq. 5.9) is exactly taken care of. Remember the “geographic dependency” like in fig. 5.4.
@Niels,
You asked: “How do you actually know that the SDU computes the relevant inner product of vectors this way (using the cos of the LOS-aircraft angle)?”
I don’t know how the SDU computes the pre-compensation Doppler shift. What I do know is that the amplitudes of the saw teeth matching the BFOs in the three phone calls are close to the same value (about 5 Hz). Now, in these particular cases the aircraft speeds are also close to being equal (cruising at high altitude). If the pre-compensation frequency steps are equal in amplitude at about the same aircraft speeds, that means there is a similar quantization in the calculation of the Doppler shift. The Doppler shift is, in its simplest form, equal to (aircraft speed / c) times the carrier frequency times the cosine of the “Doppler Angle” (i.e., the angle between the aircraft velocity vector and the vector from the aircraft to the nominal satellite location). So, if the speeds are the same, the observation that the saw teeth are close to the same amplitude implies that the quantization is equivalent to equal steps in the cosine of the Doppler Angle. I note that, even though the BFO slope (which is proportional to minus the sine of the Doppler Angle) varies by a factor of seven, and thus the Doppler Angles would appear to be quite different in relative value, the step size still looks to be the same amplitude in frequency space. The simplest model of this observed behavior is that, equivalently, there may be steps in the cosine of the Doppler Angle.
I also note that at 480 knots the maximum uplink Doppler shift is 1,356 Hz. If the cosine of the Doppler Angle were quantized to 8 bits in magnitude, say in a look-up table, the frequency steps would be 1356 / 256 = 5.3 Hz, which is consistent with the visible steps in the BFO data during the phone calls.
I don’t yet have an explanation for how often the steps occur. This varies from about 3 seconds (at 18:40) to 20 seconds (at 23:14 for MH370 and also in the MH371 event). These step periods don’t seem to match the predicted Doppler slopes. That is, the time it takes for the computed Doppler to change by 5.3 Hz is about 100 seconds, which is much longer than the observed 3-20 seconds between steps in measured BFOs.
Thus, the calculated Doppler slopes and the observed BFO slopes in between steps are NOT equal. I think that means that the SDU is doing something presently unknown to us that modulates the pre-compensation frequency in 5 Hz steps much more frequently than would be the case with a simple look-up table.
So, the “BFO Slope Method” is still an unvalidated theory. I can reasonably predict the step amplitude, but not the step rate.
I wonder why the transmitted frequency has these steps so often? The only explanation I have for the existence of the BFO saw teeth is that the SDU is impressing the saw teeth in the calculated pre-compensation of Doppler shift toward the nominal satellite position.
Despite the current disagreement in the BFO step frequency, it may eventually be possible to discriminate between a course of 180 degrees at 18:40 with no descent or a westerly course with an ongoing descent. I think this may be possible because the SDU applies no pre-compensation for vertical speed, and this may assist in discriminating routes with a descent. In addition, at 18:40 the Doppler to the nominal satellite location (used to pre-compensate the transmission frequency) and the actual Doppler shift to the satellite are rather different. That circumstance may also help to discriminate between the west and south courses used in the two methods of matching the 18:40 BFOs. First, however, we need an explanation for why the time between BFO steps is so short. Once this is understand, we will then know if we can reliably determine the 18:40 course.
@Peter Norton:
@Victor:
Careless of me.
Thanks both.
@Peter Norton
When I said “new evidence” I was referring to the claim by skipper Christopher (Kit) Olver that he snagged what appeared to be a commercial aircraft wing in his trawl net off Robe, South Australia in October 2014—about 6 months after MH370 disappeared.
I rate it low chance, but worth a small effort if we have location. I am thinking OI has more shallow depth AUV’s that might be used.
@DrB
Any chance that lateral manoeuvring during a transmission causes the Doppler compensation to be run more frequently than would be the case for straight and level flight?
@DrB
Thanks for further clarifications. I think we are more or less on the same page. Yesterday I also found that the slopes observed in the BFO data don’t seem to match the expected d/dt(Doppler) for the 23:14 phone call, being a factor 7 or so off. Assuming the step size is indeed 5-6 Hz (at 23:14 they are at least a couple of Hz in magnitude) this actually means that the steps are not due to the pre-compensation mechanism trying to compensate for the (change in) Doppler shift and having some unknown discretization effect in the calculation. Unless (and @Mick Gilbert asked this) the slope in the BFO data is not what we expect (based on the overall, hours long development of the BFO data). But it would be quite coincidental that a temporal disturbance in the overall BFO trend would (with the BTO and BFO data suggesting something close to straight and level flight) overlap with the phone call attempt. And we also have the discrepancy in the MH371 event.
So, for the moment I tend to conclude that we are not looking at discretization events originating from the Doppler pre-compensation calculation.
You mention: “The only explanation I have for the existence of the BFO saw teeth is that the SDU is impressing the saw teeth in the calculated pre-compensation of Doppler shift toward the nominal satellite position.”
Do you mean with a purpose, or as a “parasitic” side-effect of some process?
@Viking:
Thanks for taking the time to read and comment on my MH370 barnacle growth report.
I’ve removed a non-technical paragraph of the intro. I suspect that was not the portion you see at issue, and I’m open to other refinement.
Comparing Arrhenius growth curves related to global warming is interesting. I couldn’t find the particular paper you’re referencing, but I get the gist. It is relating the rate of growth change with temperature. I note in the paper that several sources show faster growth at the warmer end of the range, but that doesn’t change the conclusions. I compare the flaperon drift into tropical winter temperatures with the towing of the oil platform from the equator to NZ in 30 days, with lepas anatifera reaching 52mm. With a 39mm barnacle on the flaperon, that is still conservative evidence that it grew to that size within a month. The reproductive range of 18-25C is important for attachment and initial growth. Variations in that growth rate are less important, and Arrhenius effects on the two paths should be comparable.
I see a approximate match between the local Reunion waters temp curve starting early-June with the shell growth temperature curve.
A major finding in the paper is that the flaperon barnacles are growing atop the scouring from beaching. That drastically changes the timeline previously assumed for arrival and discovery of the flaperon, and the rate of growth on the barnacles. A crash site in tropical waters on the 7th Arc now completely explains the barnacle evidence. It also disrupts previous attempts at matching temperatures along a drift path back toward a cold water crash site, or later arrivals predicting a 34S latitude.
@Kenyon:
Thanks for exploring the energy vs seismic detectability for impact of a sinking fuselage fragment with the seabed.
You appear to compute the momentum using only the estimated weight of the fuselage material, along with 27 cubic meters of seawater. That might be correct for a hollow fuselage sinking without entrained water. As I wrote last year, it doesn’t account for the mass of water contained within a half-sealed fuselage section. The mass of the contained water, plus an additional amount of water dragged along with and behind the debris, could increase the momentum estimate as much as 10x your calculation.
Regardless, an impulsive sound was detected by dozens of seismometers and hydrophones. It appears to be at seabed depth, yet from a century of cataloged quakes in the immediate area, all were deep in the subduction zone below the Sundha Trench at a depth of 30-90 km.
Something anomalous happened at that site that doesn’t match geological events. The sound was very loud on hydrophones, but seismically weak. The timing was just as MH370 was sinking near the 7th Arc. The epicenter accuracy is within a few km, and well within the 7th Arc BTO accuracy.
Seismometers can be exquisitely sensitive. It is about whether local noise near a seismometer masks a more distant event. In some cases cattle or road traffic might dominate. Seismic magnitudes are about the volume of crustal movement, and usually at sub-1Hz frequencies. The detections of the Java event are in the anthropogenic range above 1-2 Hz, usually filtered out of quake analysis.
Lightning bolt strikes on land are easily detected on nearby seismometers, but the higher frequencies don’t propagate like quakes. The same is seen for the seabed Java event.
Like not being able to imagine what might create a loud enough implosion to be detectable on distant hydrophones, let’s be cautious about dismissing a seabed impact as detectable by seismometers.
Fortunately, most of the seismic data is public. Analysis by expert seismologists can validate the depth, polarization, and other parameters that further separate an isotropic event from crustal movement.
@Niels: Another possibility is the jumps in BFO are related to the temperature controller of the OCXO. If for instance the heater is ON/OFF with a hysteretic temperature range, there might be a jump when the heater turns on followed by a decay. Differences in the cycling time between 18:40 and 23:14 might be related to the slower temperature decay due to differences in heat transfer rates.
@airlandsman
Please read my contributions from the last few days again. I am sorry that I initially used the word amplifier, where I should have written pre-amplifier.
The point is that the HPA needs a pre-amplifier (inside the SDU box), since it is impossible to have very high gain for HF HPA’s. Particularly for older models.
The reason for my mistake is that during the last many years I did most work in optics. There it is more easy to have higher gain in HPA’s due to more favorable quantum effects (e.g. in Yb fiber-amplifiers). This also allows much higher wall-plug efficiencies.
@TBill says:
« skipper Christopher (Kit) Olver that he snagged what appeared to be a commercial aircraft wing in his trawl net off Robe, South Australia in October 2014—about 6 months after MH370 disappeared. »
No pictures of this wing ?
@370Location
If you wish, I can send you links for the two papers I refer to, but I fear they may be behind pay walls for you. I can send copies directly to you if you send me an email.
Concerning the details, it quickly becomes complicated. However, one simple thing is the complete absense of species breeding in temperate and/or sub-tropical waters in the biofouling. According to the DTU Aqua research some of this would almost certainly have survived if the airplane came down in the southern Indian Ocean.
The main message of the research papers is of course good news for marine life in a significantly warmer climate, but our investigation of MH370 can also benefit here.
@Mick Gilbert says:
« Any chance that lateral manoeuvring during a transmission causes the Doppler compensation to be run more frequently than would be the case for straight and level flight? »
Since Jeff Wise’s podcast co-host sometimes meets the Rockwell CEO and can ask him questions about their SDU, I was thinking: Can’t the IG likewise obtain the required information from Thales rather than the hard way through number crunching ?
@All
After eliminating option 3 (from yesterday), we can extract more information and decide which of the two initial possibilities is most likely. The magnitude of the crosstalk seems to be unchanged or increasing (remember to compensate for more heat due to two attempts at 18:40). This is ruling out frost or water as the cause, since both would have reduced the magnitude of the crosstalk (they insulate and increase the effective thermal mass at the same time). Therefore, the only possibility is that the cabin pressure was low!
Peter Norton said: Since Jeff Wise’s podcast co-host sometimes meets the Rockwell CEO and can ask him questions about their SDU…
They are confused. The “Rockwell” CEO he is referring to runs Rockwell Automation, based in Milwaukee, which includes the brand Allen-Bradley. The avionics group, designated Rockwell Collins, is based in Cedar Rapids, Iowa, and split off from Rockwell International some years ago (2001).
@Victor
Juan Brown informative update on the United max that ran off the runway on landing, looks like speed was too high for the turn at the end, and runway was wet. Juan says having trouble keeping up.
@All
The 10-15 fold increase in overall re-cooling time also practically rules out frost (because it would have thawed in the pre-amplifier heat after one minute). Instead it points to a low cabin temperature in addition to low pressure, since such a long re-cooling time means that the effective temperature difference (driving the cooling, since the stabilization-mechanism is unable to cool actively) must be much smaller than usual. Without detailed knowledge of the design of the SDU, we cannot calculate the exact cabin temperature. However, it must have been so low that the heating power from the temperature-stabilizer was permanently at maximum to explain the asymmetric data distribution. This also explains the absence of any signs of active stabilization. That only occurs if the heating power (from the temperature-stabilization) is completely off, or if it is pinned to maximum. It is extremely unlikely that it was completely off. I have never seen that error (except when the user forgot to switch on the whole box), while I have seen practically all other thinkable errors in such stabilizations. More importantly, if the heating were completely off the rest of the data would have been absent or looked very different. Therefore, we can be sure it was pinned to maximum.
@Peter Norton
No pics of the purported wing that was cut loose from the fishing net. This has been a “popular” news story out of Oz over couple months ago.
@Mick Gilbert,
You said: “Any chance that lateral manoeuvring during a transmission causes the Doppler compensation to be run more frequently than would be the case for straight and level flight?
I doubt the nominal Doppler frequency calculation is done at a variable rate, but the frequency steps certainly occur at a variable rate. Lateral maneuvering may be one of several possible explanations for the rapid steps observed at 18:40. There may also be a simpler cause.
So far, the only model we have to explain the saw teeth in the measured BFOs is that the uplink frequency is quantized. It may be that this is not simply a quantization in the calculation of the pre-compensation of the predicted Doppler w.r.t. the nominal satellite location. Suppose the AES uplink frequency controller is ALWAYS quantized with about 5.3 Hz “steps”.
In this model, the “least significant bit”, or LSB. of the 1646 MHz transmitter frequency is about 5.3 Hz. I also note that the 16 Hz adjustments of the frequency “offset” for 9M-MRO would be 3 bits at 5.3 Hz each.
It is clear that the AES uplink frequency is digitally controlled by the SDU. Therefore, it must have a LSB of a few Hz. My guess is that it is not 1 Hz, but about 5 Hz. The GES at Perth measures the downlink received frequency with a 1 Hz resolution. However, there is no need for the LSB of the uplink frequency to be as small as 1 Hz. Considering the uncompensated Doppler shifts and the drift of the OCXO, 5 Hz is certainly adequate for proper system operation. Manufacturers don’t usually make their system more capable and more complex and more expensive than what is required to work well. I don’t see any benefit of having the SDU control the uplink frequency with 1 Hz resolution. 5 Hz will do just as well.
Back to your question, the period of the steps tells us the slope of the SDU-calculated transmitter frequency (assuming we know the amplitude of the least significant frequency step). The more rapidly the steps occur, the higher is the slope of the predicted nominal Doppler that is being pre-compensated. That is the only variable term in the SDU calculation. Remember that the only frequency adjustments made in the SDU are the offset and the Doppler pre-compensation of horizontal speed. There are a number of other terms which affect the BFO, but they are either outside the knowledge in the SDU or purposefully ignored (such as aircraft vertical speed, which is a much smaller contributor to Doppler than is the horizontal speed).
At 18:40 a standard turn from west to south takes about a minute and reduces the nominal Doppler by 594 Hz. That’s an average slope of 10 Hz/second, which is much too fast to match the 5.3 Hz / 3.0 seconds = 1.8 Hz/s based on the observed saw teeth at 18:40. With no turn, the nominal Doppler slope for a constant-speed west course is only -0.05 Hz/s. So, it seems that something was happening at 18:40 in the horizontal course which was not a constant speed westbound, because the BFO slope in this case is far too small. Neither was it a standard turn to the left (from west to south), because then the nominal Doppler slope is much too large. These bounds indicate the aircraft could make a maneuver which matches the nominal Doppler slope, but I don’t yet know for sure what it might be. Possibly a change in speed? The estimated nominal Doppler slope of -106 Hz/minute (the equivalent of -5.3 Hz in 3.0 seconds) observed in the BFOs between steps can be matched by a smooth deceleration from 480 kts ground speed to 415 kts in one minute while on a westbound course at a constant flight level.
One flight model which can match both the 18:40 BFO values and the observed step period of 3.0 seconds is a 270 degree course, a linear deceleration from 480 kts GS to 415 kts GS and an ongoing descent with -1800 fpm rate. The predicted BFO, ignoring the quantized transmitter frequency, is steady at 87 Hz. Thus, it is possible to match the BFO data at 18:40, including the steps, by including the ground speed reduction with a straight course and a descent. Because the westerly course is nearly straight at the nominal sub-satellite point, the predicted BFO is insensitive to aircraft speed, and it is virtually the same at 480 kts or 415 kts. However, the slow-down is needed to also match the 3.0 second step period.
There may be other combinations of maneuvers which match the 18:40 BFOs and steps. In my example solution, we can see that knowing the step period provides additional route information which is especially useful in discerning aircraft speed changes.
@Niels,
You said: “So, for the moment I tend to conclude that we are not looking at discretization events originating from the Doppler pre-compensation calculation.”
Please propose an alternative theory.
You also said: “Do you mean with a purpose, or as a “parasitic” side-effect of some process?
As mentioned above, I think the quantization is in the 1646 MHz transmitter frequency. If this is the case, it is certainly purposeful and ever-present for all AES uplink transmissions.
@DrB said: Please propose an alternative theory.
As I stated above, cyclic heating of the OCXO from a hysteretic controller.
@DrB, Victor
We could check for a typical OCXO how stable the temperature is controlled and what, for a high quality quartz crystal, this means for frequency stability. I would expect they could do a bit better than 4 ppb for short time scales.
Perhaps @alsm could help here?
Also the saw tooth like pattern would be a bit difficult to understand for this case.
Regarding the uplink frequency: I would think there is a base frequency (the 1.6 something GHz) which is based on multiplying the OCXO clock frequency (through a PLL multiplier?)and which is kept as stable as possible. Relative small shifts for channel select and Doppler pre-compensation are then probably added to the base frequency by mixing (multiplying harmonic signals). I hope again @alsm or others could help and provide more details?
In general, to understand electronic system behavior and certain artefacts without knowing the circuit lay-out is tricky. There are so many options. Even knowing the circuit lay-out it can be sometimes be tricky to fully understand it..
@Niels: Just thinking out loud as what could cause the periodic shift in BFO. If a PID controller is used, the current controller for the heater would be inefficient unless it is PWM, which is of course possible. Maybe likely. But a PWM heater could introduce switching noise into the signal amplification and processing. An approximate saw tooth waveform would occur if the temperature ramp during heating was much higher than the temperature decrease rate when the heater was off.
@Victor
PWM is possible and perhaps indeed a saw tooth waveform could be the result. But again I would think the frequency could be stabilized just a bit better then we see.
I was thinking in the direction of the OCXO frequency multiplication: like a “slip” and relock of a multiplier.
@Victor Iannello:
Thanks for the correction concerning Rockwell. The question about Thales still stands though: Doesn’t Thales help with the search for MH370 by providing the IG with the required information about the SDU ?
Victor Iannello says: « it would become harder to spoof the BFO with fake satellite inclination parameters so that a northern path has the BFO signature of a southern path, as the slope of the sawtooth would be different for a northern and southern path, even though the BFOs might be very similar. »
@DrB:
Have you (or Victor) been able to verify this or are you planning to do so ?
It would be a big achievement and would help focusing the search if you could prove mathematically that a northern path is impossible because the sawtooth doesn’t match.
@DrB
Thank you for that, Bobby, much appreciated.
@ALL: I spent a lot of time many years ago trying to track down the exact OCXO data sheet, but never did score. Don helped look for it, but nether of us found the exact data sheet. That said, I’ve had a lot of experience with TCXOs and OCXOs from several vendors over the decades. And I have some FCC type certification test data on the Thales OCXO (freq varied monotonically only 30Hz over -30 to +50C ambient).
I am confident that the oven controller is an analog (probably PID; at least 2nd order) type controller (not an On-Off modulation or PWM). IOW, the heater is supplied with a continuous DC voltage, controlled by the loop. Therefore, there will be no jumps in the temperature or the frequency due to a “bang-bang” type controller. (Victor is right about PWM. I used bipolar PWM on Peltier thermal control systems for microwave radiometers. Huge noise generator.)
Also, the electronics are sealed in a hermetic can, so moisture is not a factor.
I’m not sure about this, but based on the measured L band frequency over the operating temperature range, the oven could be a double oven.
The photos I have of the OCXO show that it has a massive heatsink to provide a very long thermal TC. A burst of hot air from an adjacent LRU would have no impact on the OCXO temp or frequency.
I added a couple of files (see FCCID.io-1313881.pdf for more OCXO pics) in the Google Folder here: https://bit.ly/3It0OJx
Good luck finding documentation:
Honeywell Frequency Reference Oscillator (OCXO) Assembly 81771-MBE
Honeywell Frequency Reference Oscillator Schematic 81771-WDME
Racal part number 81771-MBE
Finally, I suggest looking at the Rockwell Patent in the Google Folder. It goes into a lot of detail on how the uplink frequency is controlled to cancel Doppler during dynamic conditions (turns, etc.). Honeywell uses a different method, but they have to account for the same dynamic conditions.
@All: Today, somebody unsuccessfully attempted to reset the password to this blog site to gain access. The IP address of the attacker is 59.42.120.122, which corresponds to Guangzhou, China, although if a VPN was used, we can’t know the original location.
@Victor Iannello,
The first 5 Hz step appears in the 18:40 data set in the interval between two BFO measurement only 0.2 seconds apart. That very narrow time window means the step can’t be produced by any real aircraft maneuver or thermal effect in the OCXO. It must be introduced by software, probably when a slowly varying calculated frequency value trips the next bit of transmitter frequency control, and there is a jump (step) of about 5 Hz.
@Peter Norton,
You said: “In my humble opinion it would be very exciting if you could show that the sawtooth differs for a northern and southern path and thereby proving with mathematics alone that only 1 of these 2 paths is possible – completely independently of the found debris which some claim was planted.”
Unfortunately, no distance-based and Doppler-based data can discriminate between two “mirror image” routes (N and S) with a line of symmetry through the satellite. The ranges, the range rates (i.e., the Doppler shifts), and the second derivatives of the range (i.e., the slopes of the Doppler shifts) are identical, so one cannot tell them apart. In effect, there are always two route solutions to match BTOs and BFOs, as I mentioned recently. When the satellite is over the equator, a North route and a South route are both possible. That does not mean you can’t tell a North and a South route apart, but doing that requires more information than range and range rate. We have this for MH370 from take-off to 18:22 by means of ACARS and radar position reports. So, we know that any route after 18:22 must connect with the pre-18:22 route and be flyable. In UGIB (2020) we showed a feasible route which ended in the SIO. I do not think it is possible to establish a northern route satisfying all the BTOs and BFOs and which seamlessly connects to the pre-18:22 positions, but it’s almost impossible to “prove” this can’t be done. We also know that MH370 crashed in the SIO. Claims of planted MH370 debris seem preposterous to me, as do claims of spoofed Inmarsat data.
@airlandsman
Thanks for an excellent description of the performance of the OCXO. Your temperature shift data agree with the official 0.3Hz/degree shifting which I used in my analysis. I do not mind a slightly higher number. That would go from good agreement to Bulls Eye.
I also played with push-pull systems using Peltier elements. The oldest models were horrible, since they easily went into oscillation (therefore your huge noise). The first commercial one of good quality I met was launched in 1993. Good home built ones were developed around 1989.
@Victor Iannello
Unusual with Guangzhou. Most of the official Chinese hackers are from Beijing or Shanghai. Probably it was through VPN or some amateurs. Guangzhou is a large town.
Good you managed to stop it. Hacking is a large problem these days.
@All
What would our best estimate for the temperature at 23:14 be? If we assume a specified operation range of 0-40 °C for the SDU, the temperature would have to be below minus 30 °C, since the stabilization also could not ‘lock’ intermittently after the pre-amplifier had supplied heat for close to a minute.
Please note the importance of these two conclusions. They rule out any solution in the southern Indian Ocean, and they totally rule out an active pilot during the last 2 hours of the flight, since he would have died as the special oxygen supply in the cockpit ran out. In addition, temperatures below minus 30°C are hard to survive for prolonged time without special clothes.
The security-video shows none of the pilots brought proper protection. Why does it rule out solutions in the south? The problem is that such solutions no longer fit the BFO values due to the 0.3Hz/°C correction for the SDU and the extremely low temperatures (at 11km height) at those latitudes, while the solution near Christmas Island fits well given the atmospheric temperature at 11km at that latitude.
@airlandsman
Did you make tests below minus 30 °C ?
@All, re cabin temperature.
Assuming a decompression or open outflow valves the packs could still be on and trying to maintain cabin temperature, pumping in warm air. So the temperature at the SDU in the roof may not be that low.
@Viking
Qualitatively working on many proposed flight paths, I do not see any problem with BTO/BFO data for the flight. I do think BFO may have drifted 3-4 units over the flight, but that would be normal drift for the quartz crystal.
In the active pilot scenario, most likely cabin pressure was restored, perhaps with intermittent bleed air. I don’t know if air flow can be maintained and heated without bleed air, or with less bleed air.
In any case, the data *looks* rock solid, poss with minor BFO drift. You have the burden of proof to demonstrate something is wrong and show a credible example.
Yesterday there was a serious incident with a Boeing 787 flying from Sydney to Auckland. Media in NZ and I think elsewhere in the world too, have been full of it all day.
The aeroplane was flight LA800, LATAM Airlines, en-route from Sydney to Santiago, with a stopover in Auckland, on Monday 11 March.
The media reports are very inconsistent. A sudden drop, sudden nose down, pilot reports all gauges going blank, [the unreliable] Geoffrey Thomas repots a sudden upwards movement before the “plunge”. The event lasted about 10 seconds, before the “pilots regained control”.
About 50 passengers injured with some hospitalised on arrival in Auckland.
Nothing untoward on FlightAware, or FlightRadar24, that I can see, but of course altitude reports in these apps will not be real time.
The aeroplane is still in Auckland and the FDR is to be examined.
It seems that turbulence is not the likely cause.
It would be nice to get some accurate information.
@Brian Anderson,
We may have to wait for FDR and (hopefully) CDR analysis. ?
@TBill
The slow drift of a few Hz is no problem.
@Brian Anderson
I experienced a similar (fortunately much milder) event as we passed Iceland many years ago. As far as I remember it only injured one stewardes (twisted arm or something like that). Many passengers got the coffee thrown in their face.
However there was a funny detail. Roughly 10 seconds before it happened, I said to the guy next to me that I could feel from the turbulence noise spectrum that something big was coming. Therefore we swallowed our coffee and grapped solid hold of the seat in front. After the event he was impressed – since he was an airline captain himself, and he had not expected it.
Perhaps a warning system could be developed based on this observation? A few seconds warning could prevent many injuries.
@All
In order to understand the complete picture we now have to take a step back and look at the two telephone-call-related data sets again. Apart from extremely different statistical BFO distributions, they actually look very similar. Both last close to one minute, and the time-distribution of data is similar. However, a closer inspection reveals some important differences. For the first data set at 18:40, two attempted phone calls came very fast after each other. To realize this, look at a plot of the first 10% of the time – some data points at 18:40 are practically occurring at the same delay-times as for 23:14, while some interleaving points appear. These additional points are the reason that there were significantly more BFO values at 18:40 than at 23:14 (almost twice as many).
Now we must remember that each time a BFO value occurs, the SDU pre-amplifier sent out a powerful pulse burst to the HPA and from there via the antenna to the receiving satellite. This produces heat at the ‘heat-release end’ of the SDU box. Therefore, one would expect almost twice as high temperature-rise at 18:40 compared to 23:14. This is clearly not the case as can be seen from the amplifier-power (C/NO). There are similar drops in the power-output due to heating of the pre-amplifier itself. How is that possible? Well, the heat dissipates in two different ways from the pre-amplifier. Normally the dominant dissipation is through the airflow generated by fans in the box. However, this is reduced roughly a factor 3-4 due to reduced pressure (roughly same factor as the increase in the time-delay discussed above), but simultaneously increased slightly due to a larger temperature difference (remember the temperature inside the pre-amplifier is much higher than on its surface – so the total difference only increases slightly). The other way is through the chassis. This is more efficient at low temperature. In practice, it looks like the four contributions (including almost twice the thermal energy at 18:40) effectively canceled each other leaving the power-curves practically identical.
A conspiracy theorist would probably ask if that was a coincidence or somebody planned everything carefully and told the telephone operator to call twice the first time (at 18:40) to make the data look similar the second time. We are not in favor of conspiracy theories, so therefore we choose to consider it a random event. However, a very unusual one because of the extremely short time (0.8 s) between the two attempts at 18:40. Just try with your mobile phone. If something goes wrong with getting a connection the first time, most people wait 4-8 seconds before trying again. For a fixed net telephone and/or high call-price typically longer. At 18:40, the operator waited less than one second.
@Victor
I do not recall if we discussed this short YouTube here, I find it fascinating. It is the account from Butterworth of a USA military jet aircraft advisor when he got to work on 8-March-2014, the jets were surprisingly flying search missions for MH370 over the Malacca Straits, which the advisor did not understand why.
Believe I may have inferred on Reddit that Malaysia obfuscated that they knew about the radar turnaround at IGARI, and somebody took issue with my wording, suggesting Malaysia did not know. Well, they looks like Malaysia had a strong idea.
Personal account starts at 2-min mark, before that they are critiquing comedienne Jocelyn Chia.
https://www.youtube.com/watch?v=xlJBS9yRVpU
@Viking
I think it possible that the second call attempt was actually automatically placed by the ground station in an effortt to improve QoS.
@TBill
I have a vague memory that Kate Tee had seen these aircrafts too.
Viking wrote ‘Now we must remember that each time a BFO value occurs, the SDU pre-amplifier sent out a powerful pulse burst to the HPA and from there via the antenna to the receiving satellite.‘
That is quite wrong.
The C-channels operate as SCPC, single carrier per channel (in each direction). The modulation and transmission of the C-channel carrier is continuous from approx 18:39:55 for the earlier call on channel 3730 and 23:14:00 for the later call on channel 3737 (until the minute long no-answer timeout exprires).
‘C channel. Circuit-mode single channel per carrier (SCPC) channel, used in both to-aircraft and from aircraft directions. This channel is time division multiplexed to provide a primary channel for voice or data traffic and a sub-band channel for signalling, supervision and data messages. The use of the channel is controlled by assignment and release signalling at the start and end of each transaction.‘
BFO is extracted by the GES channel unit, whether processing an ALOHA or TDMA burst (R or T-channel, resp) or SCPC, as each convolution block is processed. It’s perhaps a misnomer to prefix the SCPC FO with ‘Burst’.
Regardless, to write ‘the SDU pre-amplifier sent out a powerful pulse burst to the HPA‘ is nonsense. The SDU of the MCS-6000 system doesn’t emit any ‘powerful’ pulse, it generates the IF signal to the RFU which, in turn, upconverts the IF signal to L-band before amplification by the HPA.
@Don Thompson
From a system point of view you are probably right about the sequence. However, how do you explain the following details:
1) The HPA needs a significant input power since its amplification (in dB) is limited to moderate values.
2) How do you compensate for loss in various components, connectors and cables and still feed the HPA with sufficient power?
3) How do you account for the significant drop in received power during the one minute duration?
@TBill,
Referring to your linked video, Gonky is indeed a former Navy F/A-18 International Instructor Pilot. After leaving the Navy, he flew with the Malaysian Air Force as a Boeing Instructor Pilot. (Per bio).
His recollections don’t seem strong. However, his comment that aircraft were making search sorties from Butterworth the next day (Sat 8th March) shouldn’t be surprising. The clueless media throng gathered at KL were simply lost, unfamiliar with what they were dealing with, barely able to form one single cogent question.
From my conversations with people with direct, first hand, involvement it was known to DCA early on the 8th that the turnback had occurred. I have no specific details about when playback of the military air defence radar occurred but DCA’s Area Control and the TUDM/RMAF Regional Air Defence Operations Centre (RADOC) were then co-located in the same building at Subang Airport. I cannot conceive that, following establishment of the RCC on the morning of 8th March, that the DCA Area Control staff and TUDM staff had not initiated some comparison of what their systems had surveilled overnight by accessing recordings.
Indonesia began search sorties over the Str of Malacca using its unique B737-200 Surveiller aircraft, repositioned to Medan to conduct the sorties. It’s my hypothesis that Ms Tee’s story involved an encounter with an inquisitive Surveiller crew.
In summary: the Malaysian authorities knew, on Saturday 8th, that the turnback had occurred. Continued search for possible debris in the SCS should not have been surprising: at that time, the turn back may have been considered a consequence of some event that caused debris to fall into the sea. The media throng in KL were unable to divert their attention to what was happening to the west, their reporting was behind the curve. The Malaysian authorities (DCA, MOT, etc) were slow to divulge what was happening, I’d assume that they were attempting to constrain their narrative to what could be attributed to verifiable ‘facts’.
I’m not excusing anyone then involved but everything now known about that time has emerged in hindsight. Some choose to meld what later emerged with what was in the early days of the saga, errant speculation.
Viking asked ‘3) How do you account for the significant drop in received power during the one minute duration?
Because the AES and GES were negotiating optimal transmitted EIRP from the AES at 10 second intervals during the 1 minute period when channels 3730 and 3737 were active. Had the call been established, that repeated optimisation would have continued.
R.E. timing on the morning of March 8, according to a press statement of May 1, 2014 from Hishammuddin Hussein, we have the following (all local time):
7:24 Malaysia Airlines issued its first Media Statement.
8:30 Review of playback of military radar data
9:00 Air Force operations room notified
10:30 Hussein notified, who then notified Razak.
10:54 Military aircraft sent out.
The possibility of an air turnback was announced by Hussein at a press conference at 11:30 am the next day.
@sk999,
Thank you for that detail. The habit of variously publishing ‘Press Statements’ via Facebook, or Twitter, or mot.gov.my made cataloguing difficult.
In this case, it was Facebook
Don Thompson said: From my conversations with people with direct, first hand, involvement it was known to DCA early on the 8th that the turnback had occurred.
I had a long conversation with the NTSB representative that was sent to Malaysia to help them analyze the radar data. According to him, the military and the DCA were at odds about where to conduct the search. The DCA favored the South China Sea, where MH370 dropped off SSR. The military favored the Malacca Strait, as their radar data suggested a turnback over the Malaysian peninsula. You might recall the Malaysian military officer who first said there was a turnback, and later retracted his statement and said there was only a possibility of a turnback.
That same NTSB representative told me that the Indonesian military radar was not turned on. He reviewed the Indonesian radar data, and there were no targets. However, even known traffic was not captured. He thought the radar was turned off at night to save on diesel fuel.
@Viking said: The HPA needs a significant input power since its amplification (in dB) is limited to moderate values.
There is nothing preventing the HPA from having several stages of amplification to achieve whatever gain is required. In fact, some versions of the HPA tout a gain of 60 dB, which means only tens of microwatts of input would be required for an output of tens of watts. I don’t think there would be any correlation between duty factor of the transmitter and the stability of the OCXO, which would be a very poor design.
@Victor
That same NTSB representative told me that the Indonesian military radar was not turned on. He reviewed the Indonesian radar data, and there were no targets. However, even known traffic was not captured. He thought the radar was turned off at night to save on diesel fuel.
1. Which Indonesian Radar’s did he get to see:
(a) PSR’s or SSR’s or Both ?
(b) Military or Civil or Both ?
(c) Physical Locations of Radar Heads ?
(d) Time period examined (start UTC / end UTC)
2. Re turned off at night, was that generally known, presumed or suspected ‘around the traps’ (in aviation circles).
To follow up, this website:
https://says.com/my/news/malaysia-airlines-flight-mh370-missing
has many of the early stories. Later on Sunday (Day 2) at 1:54 pm, it was announced that the search had expanded to west of Penanag.
The story of the “Malaysian military officer” who allegedly retracted a previous statement r.e. the turnback can be found here:
https://www.gmanetwork.com/news/topstories/world/352163/malaysia-air-force-denies-tracking-missing-jet-to-strait-of-malacca/story/
It was Daud, and what he actually said was that he had never made any statement in the first place. [The newspaper Berita Harian apparently put words in his mouth.]
@ventus45: He was looking at the recordings of military radar data. I assume it was for the Indonesian Air Force’s Radar Unit 231 at Lhokseumawe in Aceh, Northern Sumatra. I can’t recall whether or not he specifically talked about this radar site. I didn’t ask him about the time interval, but I’m quite certain he reviewed the relevant time period.
@All
Thank you for the discussion above. I am thinking three countries probably saw MH370 heading towards Andaman Sea, not including Indonesia. Most of my contacts seem to feel Indonesia only had secondary radar coverage that night.
DonT good idea on what Kate Tee might have seen, and I assume the implication is, perhaps not the same night. Bottom line, right now we do not know of any eyewitness account that seems to actually be MH370.
@Victor Iannello
You say that the HPA may have 60dB gain in multiple stages. That is in principle possible, but as far as I remember such a gain was not available at the time INMARSAT designed and installed the MH370 system. In addition, it would probably not have been a good design. The reason is that going through all the losses and then amplifying from very low power would give poor signal/noise performance, since amplifiers were far from ideal in those days. Finally, there might have been power consumption, reliability and security issues.
For instance, the most powerful amplifier I used in 1990 had moderate gain (10-20dB), but several tens of W output. It contained a TWT amplifier, which needed kV input, and the voltage supply had the nickname the electric chair. I think that would not be safe on an airplane. Besides TWT’s are quite fragile in case of turbulence.
@370Location
I sent you a couple of emails. Did you get them?
@Viking said: You say that the HPA may have 60dB gain in multiple stages. That is in principle possible, but as far as I remember such a gain was not available at the time INMARSAT designed and installed the MH370 system.
Your memory may be mistaken.
Mike found a document for the Honeywell SD-700 SDU which is dated February 2001. That model replaced the model of the SDU installed in 9M-MRO. However, the document describes a 40-Watt (continuous) HPA with the following characteristics:
INPUT: 1626.5 MHz to 1660.5 MHz
Gain: 60 dB ± 1 dB with 0 dB back-off attenuation
OUTPUT: 1626.5 MHz to 1660.5 MHz
The RF signals in the HPA pass through five stages of class A amplification followed by two stages of class AB amplification. The signals then are routed through an output power detector after which they exit the HPA. The HPA output is connected to the diplexer and antenna subsystem.
I think any theory that hypothesizes that the drift of the OCXO is related to the heating from pre-amplification within the SDU is unlikely. The amplification for the SATCOM installed in 9M-MRO most likely occurs in the HPA, as it does for the SDU-700.
@Victor Iannello said: “their radar data suggested a turnback over the Malaysian peninsula”
“Suggested”? In my opinion it always was a clear-cut case for the Malaysian military. Their radar tracked MH370 before and after the loss of the transponder, including the U-turn at IGARI. Therefore they cannot say, they were not sure. This cannot be true.
DrB says: « Unfortunately, no distance-based and Doppler-based data can discriminate between two “mirror image” routes (N and S) with a line of symmetry through the satellite. The ranges, the range rates (i.e., the Doppler shifts), and the second derivatives of the range (i.e., the slopes of the Doppler shifts) are identical, so one cannot tell them apart. In effect, there are always two route solutions to match BTOs and BFOs, as I mentioned recently. When the satellite is over the equator, a North route and a South route are both possible. »
Yes, that’s what I thought.
But doesn’t Victor’s statement disagree ?
« the slope of the sawtooth would be different for a northern and southern path »
@viking
You do understand that the Inmarsat AMS(R)S operates (and has always operated) in the L-band? We’re not discussing Ka/Ku band aeronautical comms.
That solid state power amplifier design had evolved, in the late 1980s, to operate effectively in L-band spectrum was a key enabler for the Inmarsat aeronautical satellite system.
@Peter Norton: I agree that any path that satisfies the BTO values between 19:41 and 00:11 would show a nearly constant BFO slope of -2.6 Hz/min, whether the plane went north or south. The aircraft to satellite range in that time interval is closely approximated by a parabola, and the second derivate (the concavity) is constant. The uplink Doppler is zero at around 19:36 and decreases by 2.6 Hz/min after that time. At the time of the call at 23:14, the uplink doppler is (-2.6)(218) = -567 Hz.
@Victor
Are you seriously saying that he only saw one radar – Satrad 231 Lhok Semauwe (TRS 2215 R), and none of the others ?
The ‘shutdown to save fuel’ may be plausible for Satrad 231, but not for the most critical ‘northern outpost, ie Satrad 233 Sabang (TRS 2215 D), which was often touted by Indonesian Authorities as ‘strong’, and it is the most critical one to examine for the 18:20 to 18:40 time period.
Even Satrad 234 Sibolga (TRS 2215 D) may have seen something at extreme range.
Is there any chance you could find and talk to that guy again, and get any more answers to my original questions of March 12, 2024 at 7:13 pm above ?
@ventus45
Overnight/weekend shut downs to save fuel would be more likely for Satrad 233 Sabang; most supplies, including diesel, have to barged over to Sabang from Banda Aceh.
You will find that when it comes to the Indonesian military there’s is a fair gap between political rhetoric and practical capabilities.
@Brian Anderson, George G. Rumour has it that. “…. there may have been a problem with one of the pilot’s seats, in that the electrical adjustment motor unexpectedly moved the seat forward so that the pilot’s body impacted the control column, disconnected the A/P and forced the nose down.”
That would be novel.
@David
Moving flight deck seats and inflight upsets are not entirely novel. There was the incident with Royal Air Force Voyager ZZ333 back on 9 February 2014, where the Captain’s camera, which was on the arm rest of his seat, jammed against his side-stick controller creating some interesting negative G experiences for the pax and crew. The PIC was court martialed for it.
Mentour Pilot gives a very good account of it – https://youtu.be/Dl-Fl66Jfao
@Mick Gilbert
@David
RE: “Moving flight deck seats and inflight upsets are not entirely novel.”
In the RAF incident, the camera had been placed in the space immediately behind the side stick and in front of the armrest, such that when the Captain moved his seat forward, the armrest forced the camera into the side stick, with somewhat predictable results.
I’d be surprised if forward movement of the seat alone would have caused the pilot’s body to “impact” the control column in the LATAM incident, unless he’s a VERY large man. The seat is designed to move all the way forward without touching the control column, even when it’s occupied by a pilot. That was certainly the case with the B777 and I doubt the B787 seats are any different. If forward movement of the seat was responsible, I suspect there was some contributory factor that has not been revealed.
@ventus45: He was talking to the Indonesian military officers that were responsible for the radar. I can’t be sure which sites he visited, but he was quite confident that there was no data because the relevant sites were not turned on.
@all,
Wow, Starship seems to have slashed down near the 7th arc.
I wonder if we can learn anything from the future debris drift .
https://www.livescience.com/space/space-exploration/spacexs-incredibly-powerful-starship-lost-in-the-indian-ocean-after-reaching-orbit-for-1st-time
@All,
1. What if the amplitude of the saw teeth observed in the phone call BFOs is not a 5.0 Hz resolution in the SDU transmitter frequency control but rather quantization of the predicted nominal range rate to 1 m/s?
2. The Doppler equivalent of 1 m/s range rate is 5.49 Hz at the band center frequency of 1.646 GHz.
3. 1 Hz Doppler steps are the equivalent of 0.182 m/s range rate.
4. What if the transmitter frequency quantization is in range rate, not in Cosine(Doppler Angle) or in transmitter frequency?
5. Further suppose the resolution of the SDU transmitter frequency control is 1 Hz, which it must be not greater than, in order to create the observed inter-step slopes (and which don’t equal the actual Doppler slopes).
6. The 1 Hz steps seen in the first part of the 18:40 BFOs can’t be due to the actual Doppler, because that slope is less than 3 Hz/min (or only one 1 Hz step every 20 seconds). Therefore these 1 Hz steps in BFO seen every 3/4 second must be in the transmitted frequency.
7. The 18:40 BFO plot shows 1 Hz steps about every 3/4 second (i.e., we see four steps in 3 seconds, which is the period of the saw teeth).
8. Maybe the BFO steps are actually 5.49 Hz in amplitude and correspond to exactly 1 m/s in range rate.
9. Step amplitudes from 5.0 Hz to 5.5 Hz are acceptable fits to the phone call BFOs.
10. Next suppose the calculated transmitted frequency is an integer with 1 Hz resolution.
11. Further suppose the actual transmitter frequency reliably follows the value of the integer frequency command currently present in a register in the SDU.
12. Further assume the adjustment of transmitter frequency is done asynchronously with a given transmission. Instead, the frequency is being continually adjusted “in the background”, at least every 0.75 seconds and perhaps more often, and is not synchronized with any message transmission.
13. When a transmission is desired, the SDU simply transmits using the commanded frequency available in the register. So, it has 1 Hz resolution (but it also has 5 Hz steps).
14. Further suppose there is an error in the prediction of the pre-compensation nominal Doppler shift, such that the calculated nominal BFO slope is in error (being too large by 5X).
15. This calculation error creates an incorrect observed slope between the 5 Hz steps, but that error goes to zero at the beginning of each 5 Hz step.
16. In particular, the observed BFO slope between the 5 Hz steps is equal to the “correct” actual Doppler slope plus an induced slope error.
17. A non-accelerating route should have a very small slope in the observed BFOs between steps, typically about -3 Hz/minute at 04:29 and 23:14.
18. The observed inter-step BFO slopes are -13.6 Hz/minute at 04:29 and -15.0 Hz/minute at 23:14.
19. Most of the observed slope is probably caused by the SDU calculation error.
20. Correcting the observed inter-step slopes for the calculation error gives -2.6 Hz/minute at 04:20 and -4.0 Hz/minute at 23:14 for the actual Doppler slopes.
21. These actual Doppler slopes are consistent with straight and level flight at those times with no accelerations due to maneuvers (such as turns or speed changes).
22. The next question is, what error in the SDU calculation of the nominal Doppler shift could produce the BFO inter-step slope error and yet not accumulate an error larger than the 5 Hz step size?
23. Suppose each complete calculation of the predicted nominal range rate (perhaps done at regular time intervals) was rounded or truncated to 1 m/s. The equivalent Doppler shift of that resolution is 5.5 Hz.
24. Next suppose that, until such time as the predicted and rounded nominal range rate toggles from its last value by 1 step (of 1 m/s, or by 5.49 Hz), the SDU simply predicts the change in range rate based on the elapsed time between the times of the two most recent “toggles” of the rounded/truncated range rate.
25. So the SDU might do this:
(1) Current Range Rate = Range Rate 1 + (Range Rate 1 – Range Rate 2)*(Time – T1) / (T1 – T2)
where the most recently toggled value is Range Rate 1 at time T1, and
the toggle before that was Range Rate 2 at time T2.
26. This equation linearly projects the most recent range rate slope data until another toggle (i.e., a change in the rounded range rate) occurs.
27. Note that Range Rate 1 and Range Rate 2 are rounded to the nearest m/s.
28. The second term in the equation above is not rounded.
29. As soon as either, the (a) rounded range rate computed from the aircraft data, or (b) the extrapolated value of the range rate, changes from the previously stored value (by at least 1 m/s), the projection error becomes zero because the current value from (a) is inserted directly in the register controlling the transmitter frequency. Note that the extrapolated range rate can also trigger the change from the old to the new values from Step (a), although almost all the time the values will be unchanged. That is, the extrapolation can trigger resetting the rounded frequency, but most of the time it gets reset to the prior value.
30. Thus, the projection error cannot accumulate beyond a magnitude of 5 Hz. Every “toggle” of the compensation value reduces the extrapolation error to nearly zero.
31. This model produces the non-accumulating error feature we observe.
32. Next, what about the inter-step slope error (which is the range rate slope error in the extrapolation equation?
33. This model accounts for the 1 Hz steps seen between the 5 Hz steps, because the transmitter frequency control has 1 Hz resolution. Furthermore, I assume the calculation of the time-projection of the range rate is done to much greater precision than 1 m/s, using floating point arithmetic.
34. The commanded transmitter frequency, updated at least every 0.75 seconds, has 1 Hz resolution:
(2) Transmitter frequency = INTEGER[ (1.646E9/3E8)* Current Range Rate ]
where the Current Range Rate is a floating point number.
35. Suppose there was a 5X scale factor error in the second term of the range rate Equation (1). Such an error would mean the second term would have a large slope error, because the slope would be 5X greater than desired.
36. For the two phone calls at 04:29 and 23:14, the nominal range rates are both decreasing with time. So, in those cases Range Rate 1 will be less than Range Rate 2.
37. Thus, with a scale-factor error, but no sign error, we should see the Transmitter Frequency decreasing with time between the 5 Hz steps, and that is what we observe. So, the data for those two calls are consistent with a scalar error, and not a sign error.
38. Next, what about the 18:40 phone call? The nominal range rate here is also decreasing, but at a much higher acceleration than for the other calls. Note the much larger inter-step BFO slope of -96 Hz/minute measured at 18:40. In level flight with no maneuvers, the 18:40 Doppler is slowly decreasing with time at about -3 Hz/minute (and is not very different from the other two calls). At 18:40 some maneuver is apparently occurring during the call, and the combination of the maneuver and the slope error produces an observed -96 Hz/minute BFO slope.
39. Next consider the ratio of the observed slope to the predicted slope: At 18:40 we have -13.6 / -2.6 = 5.2X. At 23:14 we have -15.0 / -3.0 = 5.0 X. So, two of the phone calls, where we know or suspect there were no ongoing maneuvers, the impressed slope is 5X larger than the correct slope.
40. At 18:40, taking the measured slope of -96 H/minute and dividing by the same factor of 5 scale error gives -19.2 Hz/min for what would have been measured if the slope scale factor error is also 5X at that time. Since the expected BFO slope at 18:40 with no accelerations is -3.0 Hz/minute, the additional slope of -19.2 – (-3) = -16.2 Hz/minute is probably real and due to a maneuver. I find that a modest slow-down from 480 KGS to 470 KGS over one minute at 18:40 is sufficient to match this slope.
41. Perhaps the coding of Equation (1) somehow had an extra factor of 5 inserted (or more likely 5.49?).
42. Suppose instead of the term in Equation (1) of (Range Rate 1 – Range Rate 2) with units of m/s the software coder used (Doppler 1 – Doppler 2) with units of Hz. That would make the slope 5.49X too large, and this 5.49X is also consistent with all three sets of BFO measurements.
43. In summary, I have described a reasonable model for how the SDU might control the transmitter frequency which fully explains five observables: (a) the smallest observed frequency step is 1 Hz, (b) 5 Hz steps are impressed on the measured BFOs, (c) the 5 Hz steps appear much more frequently than expected, and (d) in between the 5 Hz steps the slope of the BFOs is 5X higher than expected, and (e) the errors don’t accumulate to more than 5 Hz.
44. To summarize, rounding of the range rate to 1 m/s precision produces the 5.49 Hz steps in transmitter frequency and in BFOs.
45. A coding error which uses the change in predicted Doppler shift rather than the predicted change in range rate causes the inter-step slopes of the pre-compensation Doppler to be 5.49X larger than the correct values. It is possible to correct this error post-mission.
46. There is no evidence of any ongoing climb, turn, or speed change occurring at 23:14 during MH370 or at 04:29 during MH371.
47. There is evidence for a slight slow-down of 10 knots / minute during the descent at 18:40.
@DrB said: 20. Correcting the observed inter-step slopes for the calculation error gives -2.6 Hz/minute at 04:20 and -4.0 Hz/minute at 23:14 for the actual Doppler slopes.
The uplink Doppler shift rate, calculated directly from the satellite-aircraft range fit from the BTO data, predicts -2.6 Hz/min after 19:41. Do you have the values reversed?
@DrB
@Victor Iannello
My apologies for being confused now. I’d be most obliged if you could help me understand:
I always thought the BFO data conclusively tells us MH370 went south (on condition the BFO data was not spoofed – in which case a mirror northern route is possible).
However @DrB’s comment seems to say the opposite, namely that the BTO and BFO data isn’t enough to distinguish between a northern and southern route:
« Unfortunately, no distance-based and Doppler-based data can discriminate between two “mirror image” routes (N and S) with a line of symmetry through the satellite. The ranges, the range rates (i.e., the Doppler shifts), and the second derivatives of the range (i.e., the slopes of the Doppler shifts) are identical, so one cannot tell them apart. In effect, there are always two route solutions to match BTOs and BFOs, as I mentioned recently. When the satellite is over the equator, a North route and a South route are both possible. »
These two assertions seemingly contradict each other, no ?
What I am misunderstanding here ?
And contrary to @DrB’s assertion above that “a North route and a South route are both possible”, he seems to have found a completely new way to distinguish between a northern and a southern path as @Victor wrote here:
« @DrB: That is VERY clever! […] I really doubt anybody has considered this before. So with that in mind, it would become harder to spoof the BFO with fake satellite inclination parameters so that a northern path has the BFO signature of a southern path, as the slope of the sawtooth would be different for a northern and southern path, even though the BFOs might be very similar. »
I took this as saying that you have found a completely new tell-tale sign that would give away a spoofing of the BFO data: The sawtooth would be different for a northern and southern path.
If I understand this correctly, you can now say conclusively that MH370 must have gone south because
(a) the BFO data matches a southern route much closer than a northern route (see chart above) and
(b) the BFO data can’t have been spoofed, because the slope of the sawtooth would be different for a northern path.
Is that correct ?
As I said before, if this is true, it would be a major break-through to eliminate the entire Northern hemisphere without a shadow of a doubt. Eliminating a northern path would help focusing the search efforts in the South (whereas currently there still are weekly podcast episodes promoting a Northern path). I must admit that I always found and still find Jeff Wise’s theories intriguing, but if the slope of the sawtooth would be different for a northern path, that would sound the death knell for all northern path theories, no ?
@Victor,
The numbers you quoted are the saw teeth slopes divided by 5.5. There is a second component on the slope due to the overall “tilt” of the sawtooth. To get the total BFO slope, you have to add tilt to the sawtooth slope / 5.5. I have done that as explained below.
The slope of the range rate is calculated to be equivalent to -1.7 Hz/min at 04:29 (using the ACARS route data), and -2.7 Hz/min at 23:14 (using the UGIB route). Note these range rate slopes have nothing to do with BFOs.
I also fitted the actual slopes of the BFOs, and correcting for the SDU calculation error and including the overall sawtooth tilt, the total range rate slope based on BFOs is estimated to be 0.0 Hz/min at 04:29 and -1.7 Hz/min at 23:14.
The differences between the two methods indicate the typical error in retrieving range accelerations from BFO slopes. Therefore, the typical error in estimating the range rate slope from a few cycles of BFO saw teeth appears to be 1-2 Hz/min. This retrieved BFO slope uncertainty puts an upper limit on the detectable maneuvers (i.e., accelerations) that might be ongoing during a phone call.
@All,
1. The Inmarsat GES in Perth measures the received frequencies (the BFs) from the SDU with 1 Hz resolution. That means there is never a measured BFO step of less than 1 Hz, even if the transmitter had its frequency controlled in steps smaller than 1 Hz. So, there is no benefit to make the SDU transmitter frequency resolution less than 1 Hz.
2. When the aircraft is at the gate, its velocity is zero, and the pre-compensation Doppler shift is zero and won’t change. It should never toggle. There should be no 5 Hz BFO steps at the gate.
3. Therefore, the BFO offset is accurately determined using the gate data. It is not corrupted by the SDU error, and there is no bias error to correct.
4. Whenever the aircraft is moving, the pre-compensation frequency will vary, and the 5.5 Hz steps will occur. That means that there are times when the transmitted uplink frequency is in error by 5.5 Hz, and there are times when the error is zero.
5. Therefore, so long as the slope of the nominal calculated range rate does not change sign, BFO errors between 0 and -5.5 Hz occur when the nominal range acceleration is negative. The RMS error is 5.5 Hz / SQRT(12) = 1.6 Hz. The mean error is -2.75 Hz.
6. Similarly, BFO errors between 0 and +5.5 Hz will occur when the nominal range acceleration is positive.
7. The BFO read noise is estimated by analyzing the residuals after fitting a saw tooth waveform to each BFO phone call data set. Their RMS values are 0.7 to 0.9 Hz. Thus, the typical BFO read noise is about 0.8 Hz RMS.
8. It is clear that the BFO probability density function is dominated in flight by the SDU coding error because its RMS error (1.6 Hz) is twice the read noise (0.8 Hz).
9. It is also important to understand that, when the aircraft is moving, the expected BFO value is biased by the SDU coding error. In fact, that bias is +/- 5.5 Hz / 2 = +/- 2.75 Hz, assuming random transmission times. When the nominal range acceleration (i.e., the nominal Doppler slope) is negative, the average of many BFO readings will be 2.75 Hz less than the “true” value.
10. When the nominal range acceleration is positive, the BFO bias is +2.75 Hz, and the average of many BFO readings will be 2.75 Hz higher than the “true” value.
11. These findings allow one to statistically compensate the measured BFOs (to remove the BFO bias caused by the SDU coding error) with a fairly high degree of reliability if the route is sufficiently well determined at a given time that there is no doubt about the sign of the nominal range acceleration.
12. One can in principle remove this BFO bias error and then re-fit route parameters, for instance, from 19:41:03 until 00:11:00. For this portion of the MH370 route, the range acceleration is always negative for fixed lateral navigation modes, and the ‘corrections” to be applied to the existing BFO values are to simply add 2.75 Hz to the BFOs from 19:41 through 00:11.
13. The sign of the nominal range acceleration after 00:11 is not highly certain because the aircraft may have slowed down upon first engine flame-out (circa 00:11), and course changes almost certainly occurred after 00:17:30.
14. It is not possible to correct all the BFOs individually, because we don’t know exactly when the BFO measurement occurred during the sawtooth cycle. We can only apply a statistical correction for the bias in the expected value. We also can’t remove the 1.6 Hz RMS BFO noise introduced by the SDU error.
15. My explanation of this major component of BFO noise finally makes understandable Inmarsat’s empirical criterion that the magnitude of the BFO error could be up to 7 Hz even when the route was well known. You can get up to +/- 5.5 Hz of BFO error from the SDU coding error, and then you have about a +/-1.6 Hz additional read noise (at 2 sigmas), making +/- 5.5 +/- 1.6 = +/- 7.1 Hz errors possible for a single isolated BFO even with accurate knowledge of all aircraft data.
@Peter Norton,
With a stationary satellite above the equator, it is impossible to distinguish between “mirror-image” routes N and S of the equator based on BTO (range) and BFO (range rate) measurements.
My new analysis technique for phone-call BFOs allows us to independently determine the second time derivative of range (i.e., the range acceleration, AKA range rate slope and AKA BFO slope) at several points along the route.
I never said one could distinguish N and S MH370 routes from the Inmarsat data by any means, new or old, because I know you can’t with a stationary satellite. The two routes can have the same ranges at the same times (producing equal BTOs), the same range rates at the same times (producing equal BFOs), and the same range rate slopes (producing the same sawtooth BFO slopes). The latter won’t have different signs for the BFO slopes of N and S mirror routes.
You can tell the aircraft went south by when and where crash debris were found in the Indian Ocean.
A second indicator of a southbound route was the sighting by Kate Tee.
A third indicator comes about because of the inclination of the satellite. It has a motion toward and away from the equator, so it has a N-S velocity component that induces a Doppler shift toward the aircraft which is asymmetric for N and S routes. The Z-dot speed of Inmarsat F-1 is -1.5 m/s at 19:40 and 81.9 m/s at 24:10. So these satellite speeds parallel to the Earth’s axis are sufficiently large to create a measurable difference in the BFOs for N and S mirror routes, especially late in the flight. More or less, the satellite is moving south at a good clip at 00:10 and reduces the net Doppler to a southbound aircraft in the southern hemisphere, but it increases the net Doppler to a northbound aircraft in the northern hemisphere. This difference in BFOs will be distinguishable in the fitted route parameters because it is on the order of a 100 Hz difference in the required aircraft-to-satellite Doppler to match the BFO values. What that implies is that you can’t fit the MH370 data set by simply changing all the S latitudes to N latitudes. The symmetry is broken, so the mirror image of the southbound route can’t fit the same BFOs. However, that does not prove there is not a reasonable fit for a northbound route which is different from the best-fit southbound route. I have not tried to fit any northbound routes, so I don’t know whether or not this is possible.
What the BFOs told us early on was that the aircraft was initially flying toward the satellite, and then it flew away from it. That it headed S when flying away from the satellite was an early assumption, not a conclusion.
If you want evidence, independent of debris reports and visual sightings, that the aircraft flew south rather than north, my third point above about the satellite velocity destroying the BFO symmetry is a solid argument, but modeling northbound routes to look for the absence of any reasonable route fit would need to be done to make that case solid. In other words, you still have to demonstrate that there is no flyable northbound route that reasonably fits the 19:41 to 00:11 data.
@All,
There is a typo in my 1:06 message above: The F1 Z-dot speed at 24:10 should be -81.9 m/s, not (+)81.9. It is moving “southward” at a significant speed.
@DrB
Regarding your 7:33 pm post: Interesting thoughts.
Looking at eq. 5.9 in the DSTG book: f/c = 5.49 m-1
Which is multiplied by the whole vector thing (the nominal range rate [m/s]) which you suggest might have some strange error pattern with an amplitude max. 1 m/s.
I’m wondering how the denominator is calculated (below the vector inner product) in eq. 5.9. And how the numerator and denominator are represented as numbers. If that is indeed how it was done. Is there perhaps a binary long division involved with limited precision and some rest term which accounts for the inter-step slopes?
But for any “model” to explain the saw teeth: how to prove it is correct with the little data we have?
@DrB wrote ‘36. For the two phone calls at 04:29 and 23:14, the nominal range rates are both decreasing with time. So, in those cases Range Rate 1 will be less than Range Rate 2.
04:29? The Stratos Log, showing the record of SATCOM traffic with 9M-MRO as it operated the MH371 and MH370 services does not show a SATVOICE call at 04:29. Did you mean to write 18:39?
In very simple terms, you are hypothesising a 5Hz ‘box’ imposed by the SDU’s doppler pre-comp while the GES channel units discriminate in 1Hz steps?
@All
I am sorry for being absent yesterday. I had some sheduled work to do. I will try to catch up today.
@Don Thompson
You are completely correct about the development of semiconductor-based amplifiers in the late 1980-ies. That is also the reason that I used examples of what was available in 1990. After 1990, a couple of things happened, partly due to the fall of the Berlin wall. Initially we found out that Russia had developed competing (but cheaper) technology based on radio-tubes, produced in Ukraine. Around 1992 Ukraine started exporting it (against the laws of Russia as far as I know). However, further development of western semiconductor technology outcompeted the tubes relatively fast. The 2001 model is certainly based on this more recent development.
@Victor Iannello
For the reasons described above, I am quite sure the older model of HPA in MH370 had less impressive specifications, and possibly a couple of weaknesses. Therefore, the SDU may have needed a pre-amplifier anyway.
More importantly, the temperature stabilization in the older model was almost certainly only using reading from one point, most likely at the last (most power consuming and hottest) stage. If both temperature and pressure fell, this causes temperature gradients, so things like gain-equalizers and the first amplifier stages got relatively cold, while the temperature of the last stage remained constant.
I am not saying temperature gradients change the frequency (stochastic realization shifts will be near zero by symmetry, since the SDU frequency was near the middle of the amplification band), but the output power is likely to increase (as observed, – and as I will discuss later).
@Victor Iannello
Concerning the frequency performance of the SDU during the 23:14 phone call, I am still convinced it was predominantly due to thermal effects. If they did not originate in the presence of a pre-amplifier, they may well have come from increased power consumption in the electronics during the phone call, e.g. from GPS-correction related calculations. I had problems with that kind of crosstalk during experiments in 1992.
@Don Thompson,
During MH371 cruise, between 04:29:15.419 and 04:30:15.413 UTC, there were 132 BFO measurements. I assumed they were during a 60 second satellite phone call. These are plotted in Figure F-4 in UGIB (2020).
You asked: “In very simple terms, you are hypothesising a 5Hz ‘box’ imposed by the SDU’s doppler pre-comp while the GES channel units discriminate in 1Hz steps?
In simple terms, yes. The SDU is artificially “dithering” the transmitter frequency in a way which is not generally recoverable. In some cases, and with additional knowledge of the route, we can figure out whether the “box” is above or below the correct frequency.
@DrB
I would put simulator flight south considerably over any unsubstantiated eyewitness sighting.
Peter’s need for flight south verification I am afraid mirrors even the most bright NoK. They are hearing a barrage of conspiracy theories which are attractive vs. the apparent truth.
@Niels,
You asked: “I’m wondering how the denominator is calculated (below the vector inner product) in eq. 5.9. . . . .”
I don’t know how the calculation is done in the SDU. It may be done differently than what DSTG did. The equations do not matter. What matters is that the nominal uplink range rate is quantized (converted to an integer) with 1 m/s resolution. That’s what creates steps of 5.5 Hz in the transmitted frequency.
You also asked: “But for any “model” to explain the saw teeth: how to prove it is correct with the little data we have?”
My theory of what is occurring in the SDU explains all the strange observable behaviors, which I previously listed. It matches the MH371 and MH370 phone call BFOs within 0.8 Hz RMS.
To judge for yourself, I have prepared figures showing the saw teeth fitted to BFOs in three phone calls. You can get it here.
@All
If we assume the unusual statistical BFO distribution at 23:14 is due to thermal effects, we are now ready to handle the entire BFO data set. At some time after 18:40 but long time (at least two hours) before 23:14, somebody must have opened a door or big hatch causing pressure-loss and temperature drop. Why would somebody do so? If the purpose was suicide it would have been infinitely more easy and convenient to just let the air out of the plane through a valve while leaving the heating on, passing out and die peacefully from shortage of oxygen. Therefore, we can only come up with one reason to open a door or hatch – the person wanted to attempt parachuting out of the airplane.
Where could that have happened? The only realistic place is over Bandar Aceh where there was plenty of streetlight and relatively flat ground to land on. This is exactly where Kate Tee observed a large airplane diving down to low height, and flying slowly. She even concluded that it looked like someone was going to jump out or emergency-land in Aceh. With the airport closed and the route passing perpendicular to the landing strip, this is ruling out the second option. Therefore, attempted parachute jump is the only option. This is consistent with the airplane flying low and slow with an open door over Bandar Aceh, and then returning to previous normal speed and height under pre-programmed autopilot control with the open door soon after passing a waypoint.
We only need to know the flying height and the thermal time-constant of the airplane to make a good estimate of the temperature (and pressure) as a function of time if it had an open door and continued towards Christmas Island. That is what we did in version III of our paper, and obtained good agreement with all BFO values and more importantly with the received power at the satellite (which nobody else have explained properly). Remember here that most microwave amplifiers of the relevant generation have an output power dependence on temperature around -0.05dB/°C, so a 20°C drop in temperature increases the power roughly 1dB. This explains the majority of the odd increase in received power with increasing distance to the satellite.
If you insert the pressure at 11 km height, it corresponds to a factor 4 on the time-constant, which is within the range 3-4 found from the BFO values at 23:14. Please remember the reference is a normally working pressure cabin at 18:40 and not conditions at sea level. Therefore, 11 km flying height at 23:14 is a good match.
As a final remark on the BFO errors, it is now possible to use an uncertainty of 4 Hz (as most people prefer based on previous BFO statistics, see e.g. the Holland paper) and still have excellent agreement. However, this only goes for the solution near Christmas Island. For the ‘consensus’ solution in the south things do not fit any longer, even with 7 or 16 Hz error. Basically this means that the data from the attempted phone calls rule out the ‘consensus’ solution entirely, and explains why the wreckage has never been found.
@All,
We don’t know if the SDU truncates or rounds the range rate to 1 m/s resolution. If it truncates the range rates, then one has to add 2.5 Hz to the BFOs from 19:41 to 00:11 because those data are biased low by ½ m/s.
If the SDU rounds the range rate value, then there is no bias in the BFOs, just additional “noise”. In that case, no offset adjustments are needed to the MH370 BFOs for the previous southbound route fits.
@DrB
I hope I understood your method correctly: I checked the second derivative of the range vs. t around 23:14 but to change the range rate by 1 m/s I would need about 188 s, which is about 10 times longer than we observe based on the step period. Now this is based on the real range function and the satellite is also accelerating -> I’ll try to construct the range function vs. stationary sat. position.
@Niels,
The range rate slope at 23:14 using the UGIB (2020) route is -2.70 Hz/min for the nominal satellite location and -2.53 Hz/min for the actual satellite. So, for the SDU to calculate a change of 1 m/s (equivalent to 5.5 Hz) in the nominal range rate should require 122 seconds. The period of the sawtooth fitted to the BFOs is 22.15 seconds, which is about 5.5X more often than expected.
The sawtooth inter-step BFO slope at 23:14 is one -5.5 Hz step amplitude per sawtooth period of 22.15 seconds, or -14.90 Hz/minute, if the overall tilt of the sawtooth is constrained to be zero. Because we have only about 3 cycles, the overall tilt is not well determined. Therefore, it is better to constrain it to be zero and allow the period to vary to find the best overall slope measurement. That measured slope is exaggerated by the SDU error by a factor of 5.5X, so we predict an actual change in the nominal BFO 5.5X smaller, or -2.71 Hz/min. This agrees closely with the -2.70 Hz/min of the UGIB route.
Viking: I see Don and Victor have already answered the question about the HPA gain and drive requirement. The HPA gain is digitally controlled (power agile). The nominal RF output (1645 MHz) from the seven combined channels in the SDU is nominally +15 dBm (31.6 mW). Therefore, there is nothing inside the SDU that could cause a “Key Down” thermal effect on the OCXO.
DrB wrote ‘During MH371 cruise, between 04:29:15.419 and 04:30:15.413 UTC, there were 132 BFO measurements. I assumed they were during a 60 second satellite phone call.‘
In the air-to-ground direction, during the period stated above, the aircraft transmitted a sequence of T-channel bursts that carried the 30min periodic FMS progress reports comprising four ACARS blocks destined for MAS OCC.
The T-ch speed and modulation was 1200bps, B-PSK. I believe you have double counted the number of independent BFO measurements concerned with those 8 separate T-ch bursts.
During the MH371 service all half-hourly FMS progress reports were received as expected whereas only one during the MH370 service.
@Don Thompson,
For that MH371 data, I observe that BTO and BFO values can change when two adjacent events have time differences of at least 0.16 seconds. These look to be independent. Some elapsed times are only 0.04 seconds, and in those cases the BTOs and BFOs don’t seem to change, so these pairs of events may have redundant BTOs and BFOs.
Using all the data, including some redundant events, has no material effect on my analysis.
@DrB
There are two points of interest that have always intrigued me as to the possibility of calibrating a particular hypothesis. One is the fact that the satellite is virtually stationary at about 19:41 [actually I think it is closer to 19:36], and the other is that it is very likely that the aircraft was flying tangentially to the satellite at about 19:52.
In the first case the satellite contribution to the BFO [Doppler] ought to be zero, and in the second case the contribution of the aircraft to the BFO [Doppler] ought to be zero.
Are these two points able to be used, even if a little interpolation of the BFO is necessary, since there is only one observation at 19.41, to help verify your hypothesis, or indeed to help establish the aircraft track at 19:41.
@DrB
For the T-channel traffic, in the period 04:29:15.419 and 04:30:15.413 UTC, there were 8 bursts. That means 8 independent BTO measurements. The deconvolution, FEC decoding, and descrambling involves sequential pairs of SUs, hence the a single BFO extracted for the pair of SUs and reported for the log entries of two SUs.
During that same period there are 8 R-channel bursts, from which independent BTO and BFO measurements are derived.
So, (8×8)+8 = 72 independent BFO measurements.
While I do agree that this analysis isn’t materially affected, I suggest it’s best to properly describe the observed events.
@All. Former Qantas pilot Mike Glynn, with Andrew Banks’s assistance, has drawn a direct connection between the southern Indian Ocean points recorded on 2nd Feb 2014 on Captain Zaharie’s home simulator and a plan by him to divert the 4th Feb 2014 flight he piloted to Jeddah, Saudi Arabia to the southern Indian Ocean.
That was about five weeks before MH370’s.
He suggests a possible reason for Zaharie not utilising that Jeddah flight with this aim was that a third pilot would be needed for that, as has been raised here.
That plan envisaged turning off the track to Jeddah at 10 deg N towards the US base at McMurdo Sound, Antarctica.
After fuel exhaustion it foreshadowed a long glide then a final dive from 4000ft.
For those that can access The Australian it is here:
https://www.theaustralian.com.au/inquirer/secret-flight-path-of-mh370-captain-zaharie-shah/news-story/011d30d8539a87b7639e74f324a4a339
Its title is, ‘Is this the smoking gun? Sinister details fuel MH370 captain focus’
then; ‘A Qantas pilot’s take on the chilling scenario suggested by independent research into Zaharie Shah’s home simulator.’ By Mike Glynn
It is an extract from a book by him, to be published shortly: ‘MH370: Answers From a Pilot.’
While I have copied it I gather that copyright extends to newspaper articles in Australia so am unwilling to post that, though I will send @VictorI a copy in case he can get around that issue.
Perhaps Mike Glynn or @Andrew might be able to help.
In providing an explanation as to what the purpose was of this simulation clearly his article’s interpretation is telling.
While the two points in the simulator suggest there was high speed between them, his planned flight to Jeddah did foresee a 90 miles glide after MEFE. Thus while the fuel for the MH370 flight was less, if a parallel that would have concluded with a long glide too.
A steep descent after MEFE or a long glide? Just the steep descent would be consistent with the final log on request BFOs, though both with a final dive from 4000ft.
Either way the simulation points to the likelihood that MH370 was piloted to the end and by whom.
@DrB
After recalculating the second derivative in another way I get much closer to your 122 s for the step period at 23:14 so that looks fine.
I would prefer if the factor 5.5 could be explained better though (than through a coding error, point 42 from your 7:33 pm March 14th posting)
@Brian Anderson
When I derive the sat-ac distance r from BTO measurements and then interpolate with a third order polynomial I get the function r(t), which has a minimum (dr/dt = 0) around 19:55:30 in my calculations. Is that how you approx. came to the estimated 19:52?
A follow up question there: It seems to me the latitude for flying tangential would depend on the track; did you derive the relation between the latitude and track in your analysis?
I’m currently analyzing 20 million randomly generated paths, and I could check how the distribution for LEP changes if I add this possible relation between track and latitude at 19:55:30 (of course with some error margin).
@David
I’m more than happy for Mike to clarify on this point but it appears that the “range ring” from 10N is based on an estimate of remaining fuel plus glide if you apply a “typical” fuel load for MH150, rather than the actual fuel at 10N of 62,700 kg per the simulation data point.
Probably also worth noting at least few other points. In the recovered sim session:
1. from 10N, the simulation aircraft was almost certainly flown south-east, back towards Banda Aceh rather than south towards 45S1.
2. from 10N, the fuel jettison function was almost certainly used, resulting in fuel exhaustion somewhere off the west coast of Sumatra.
3. it was only after fuel exhaustion had occurred, and the simulation aircraft had been in a brief unpowered descent, that it was manually relocated down to 45S1.
Deep Sea Vision wants to search for MH370. 60 Minutes Australia interview with DSV’s Tony Romeo on Sunday.
https://twitter.com/60Mins/status/1768774174984597598
PS DSV is talking to some of us.
@Brian Anderson and @Niels
It is an interesting subject you are bringing up. One that I had actually planned to postpone until later. Near 19:41 there is a significant degree of correlation between the flight route of MH370 and the ‘Eigen-movement’ (sorry for using a German-inspired word from quantum mechanics, but the meaning becomes clear below) of the satellite near its most northern extremum. The exact values on a minute timescale depend on which of two possible intrepretations you use:
1) Simple version (anti-correlation): The route soon after 19:41 had a slight (and smooth) change of heading (a few degrees). The resulting scenario sends MH370 to the ‘consensus’ solution in the Southern Indian Ocean.
2) Complicated version (correlation): The route soon after 19:41 contained almost a U-turn (180 degrees minus roughly the same few degrees). The resulting scenario sends MH370 towards Christmas Island.
Because of the correlation/anticorrelation issue it is almost impossible to decide which of the two is correct (like in quantum mechanics). My figure 2 is probably the only way – at least the only one I could come up with, – and it points to the U-turn.
If you look at the presumed Jeddah simulation on February 2, there is an anti-correlation with the Eigen-movement of the satellite (near its most southern extremum). In other words Shah chose the simple solution there. Perhaps he left that simulation to be found and mislead the investigators? That would also explain why he turned the SDU back on (a mystery for many investigators).
This thought has kept giving me the creeps for 10 years for the following reason: Exactly two years before Shah’s Jeddah simulation I gave the following presentation at a conference in Sydney Australia:
’Generating and Sensing Signals for Quantum Cryptography using Phase Encoding in Compact Silica-on-Silicon Mach-Zehnder Circuits with Bragg Gratings’, Presented as oral at the APOS conference, Sydney, Australia, January 31 – February 3, 2012
I had a printout of the presentation on my breakfast table for last minute rehersing on February 2. Guess who was also sitting in the breakfast room of the hotel (Mercure hotel near Central in Sydney)? Two tables from me was Captain Shah together with another pilot from Malaysia Airlines. As I picked up more food from the self-service table he could easily have walked past my table and seen a nice figure illustrating how to hide with quantum entanglement. In another figure was a discussion of possible thermal crosstalk and how to avoid it.
@Niels,
Yes, interpolating the BTO with a 3rd order polynomial and extracting the point at which dr/dt=0 is exactly the way I did it . . .many moons ago. It was the fundamental observation which led to my “mid-flight speed” hypothesis late in 2014. Based on the accuracy with which we knew of the data and measurements at that time, I can accept a couple of minutes either way.
I agree that the latitude at that time depends on the track, [or vice versa] and this is the interesting question. I was never able to derive it.
Hence the question for DrB, and my interest in your result too.
@Niels,
Just to clarify . . I reverse engineered a BTO for the time 19:01 too, on the basis that the track looked like it may have been pretty straight from that time, and that helped with the polynomial fit. I ignored BTOs before that time.
Those adjustments could also affect the precise time for the derived tangent time.
@Brian Anderson
Perhaps the relation between track and latitude is not too difficult: I would think that at t the velocity difference vector between sat and ac has to be tangent to the sphere around the satellite defined by r(t). This probably means that the in plane component of the velocity difference vector has to be tangent to the virtual ping ring which can be defined for given r(t) by intersecting the sphere with earth surface. As we have a polynomial r(t) we can define a virtual ping ring at any desired moment in time assuming there are no strange U-turns, for which the overall trend in BTO and BFO imo leaves little doubt.
@Viking
Re: “Two tables from me was Captain Shah together with another pilot from Malaysia Airlines.”
I am curious as to how you recognised the fellow some two years prior to his notoriety? And how you could tell that he was with another MAS pilot?
Re: MH371 BFO patterns?
In addition to the group of BFO values circa 0430 that Bobby analyzed, there were similar groups every 30 minutes (ACARS reports). I picked 3 more (during the stable cruise phase) to compare in this note: https://bit.ly/3TnPB1I
I find it difficult to see a consistent pattern (saw tooth or otherwise) across these samples.
@Peter Norton,
@All,
1. I have done a comparison of the MH370 BTO and BFO errors for northbound and southbound routes.
2. The results are available here.
3. I built a new flight model which assumes a constant ground speed at a constant flight level.
4. I ignored the local weather.
5. I assumed there was sufficient fuel on board to fly until 00:11.
6. I assumed an initial bearing using LNAV (i.e., following a geodesic route).
7. I solved for the 19:41 latitude, longitude, and ground speed which minimized the sum of the normalized BTO and BFO errors in units of sigmas.
8. For the BTORs I used one sigma of 29 microseconds.
9. For the BFORs I used one sigma of 2.4 Hz.
10. The normalized error is the RMS value of the residuals divided by the assumed standard deviation. So, a value of 1 is an excellent fit. A value of 2 is a low-probability, marginal fit. Values larger than 2 are considered to be unacceptable.
11. All southbound routes from 172-195 degrees had normalized BTORs less than 2 sigmas.
12. For the southbound routes, initial bearings from 175-187 degrees had normalized BFORs less than 2, and the 19:41 locations were reachable on time from the last radar contact position. The 00:11 latitudes of these acceptable southbound routes are from 30.4 to 37.1 degrees S.
13. Southbound routes from 188-195 degrees are rejected because the 19:41 location is too far south from the last radar contact position to be reached on time.
14. The primary discriminators for southbound routes are the BFO errors and the 19:41 location.
15. Because of the southbound speed of the Inmarsat F-1 satellite at the times of interest, the BFORs are highly discriminating between southbound and northbound routes.
16. No northbound route has acceptable BFO errors, with the best case being more than 20 sigmas. Therefore, no northbound routes can possibly match the MH370 BFOs.
17. The BTO errors for northbound routes are all greater than 1.7 sigmas. Therefore, all northbound routes have, at best, marginally acceptable BTO errors, and some routes are much worse.
18. Northbound routes from 300-330 degrees are too far north at 19:41 to be reached from the last radar contact position.
19. Northbound routes from 6-70 degrees are too far south at 19:41 to be reached from the last radar contact position.
20. Northbound routes from 331 to 6 degrees are reachable at 19:41, but their BFO errors are totally unacceptable, being more than 22 sigmas.
21. As shown in the figure, the northbound cases are plotted as red dots. None are acceptable, and by huge factors. The red numbers next to the points are the initial bearings in degrees.
22. As shown in the figure, the southbound cases are plotted as black dots, and the black numbers are the southbound bearings.
23. Acceptable southbound routes have initial bearings from 175-187 degrees, and their 00:11 latitudes are 30.4 to 37.1 degrees S.
24. In my opinion, this exercise demonstrates that no northbound LNAV route can possibly be consistent with the Inmarsat data.
The link to my N-S comparison plot seems to be broken. I’ll try it again. Look here.
@David
I’ll let Mike Glynn explain, if he’s willing. The theory is his work; I only reviewed the fuel calculations.
If Mike’s theory is correct, it explains the significance of the recovered simulator data and adds weight to the theory the aircraft’s disappearance was pre-planned. However, it does not necessarily explain what happened the night MH370 disappeared.
Oops! Wrong plot. I hope the third time is the charm. Look here.
@DrB: Yes, we’ve known since Inmarsat published the comparison of calculated BFOs for northern and southern routes that the measured BFO values favored the southern routes. The satellite inclination breaks the symmetry.
I think what @Peter Norton was referring to were these two papers I wrote in the spring of 2015 that discuss northern routes and how a SATCOM hack of a particular kind would give a northern path the BFO signature of a southern path. Note that those papers were written before any debris were recovered and before the captain’s simulator data were available.
https://www.dropbox.com/s/693pvqgqpawglj6/2015-04-29%20Northern%20Routes%20for%20MH370.pdf?dl=0
https://www.dropbox.com/s/0gav5kh74ll6xkd/2015-05-16%20Northern%20Routes%20and%20BFO%20for%20MH370.pdf?dl=0
For northern and southern paths that satisfy the observed BTO data, the uplink Doppler should match for all paths, and cannot be used as a discriminator (assuming the uplink Doppler could be extracted from the BFO data, which now seems very difficult). On the other hand, the AES pre-compensation values would differ. In the proposed hack, the pre-compensation algorithm is manipulated (by changing the satellite inclination parameters from zero) in a way that the BFO values of a northern path would be similar to a southern path.
@Brian Anderson,
@Niels,
1. Here is everything you wanted to know, and probably more, about aircraft routes which are tangent to spheres centered on the satellite location and the nominal satellite location.
2. The UTC when the satellite velocity (i.e., the range rate) w.r.t. the Earth is zero is 19:40:56. At this time the satellite is not moving in the Earth reference frame and the downlink Doppler from the satellite to the GES in Perth is zero.
3. The time when the aircraft range rate to the nominal satellite location is zero depends on the aircraft route. For the 180 degree UGIB (2020) route, this occurs at 20:03:00. This can also occur at 19:41:03 if on a course of 174.8 degrees.
4. The time when the aircraft range rate toward the actual satellite is zero depends on the aircraft route. For UGIB (2020) this occurs at 19:48:57. This can also occur at 19:41:03 when on a bearing of 178.1 degrees.
5. The time when the satellite range rate toward the aircraft is zero depends on the aircraft location. For the UGIB (2020) route this occurs at 19:54:57.
6. The time when the combined range rate between the satellite and the aircraft goes to zero depends on the aircraft route. For the UGIB (2020) route this occurs at 19:48:51. This can also occur at 19:41:03 when on a bearing of 178.2 degrees.
7. The time when the predicted BFO is minimized depends on the aircraft route. A minimum of 104 Hz can occur at 19:41:03 when on a bearing of 179.3 degrees, or a minimum of 83 Hz could occur at 17:48:56 if on a bearing of 180 degrees.
8. I don’t see from these results any new way to constrain the MH370 route besides what has already been done – by fitting the 19:41 location and bearing, as was done by UGIB (2020).
Victor,
I reread the second of your two papers, the one proposing the satellite inclination “spoof”. I have some minor comments:
1. Your wrote that “As part of the log-on process, an AES updates its System Table …” That is not necessarily true. The system table is in non-volatile memory and survives a power reset. The system table maintains a revision number. Only if the revision number is out-of-date is the table updated. (At least, this is true for the MCS-7200.) So the inclination will not necessarily be reset at the 24:19 logon.
2. I read Ruben Santamarta’s paper and watched his video on SATCOM vulnerabilities. For the Cobham Aviator 700 (which you pointed out provides Classic Aero service like the MCS-6000), he identified a possible security vulnerability in the Swift Broadband service. I don’t believe that an MCS-6000 supports Swift Broadband. Likewise, the backdoor in the MCDU interface involved a screen requiring the entry of a password. I couldn’t find a comparable screen for the MCS-7200, presumable the same is true for the MCS-6000. The only way I can see manipulating the system table is by uploading new software with custom modifications. But once you do that, you introduce the possibility of spoofing the BTOs, which are sacrosanct in the Jeff Wise universe.
3. I think Gerry Soejatman’s idea of spoofing the IRS data is more feasible than you represented. I would guess that both ground speed and ground track are used by the Doppler compensation calculation but are not used by the beam steering unit (which instead uses heading, along with other parameters). Gerry makes it sound like it is easy to reach into the E-11 rack and “Take out the feed cable …” for the IRS ARINC-429, but it isn’t that easy. These lines (there are two) are in the middle of a large cable bundle feeding a massive connector. Rather, I would think it easier to access in the MEC, where there is an Inertial Reference Bus ARINC-429 output on each AIMS cabinet that looks to be much easier to access. Well, these buses also go to the GPS receivers, so who knows what spoofing the data would do.
@Victor Iannello,
1. The pre-compensation of the expected uplink Doppler is computed in the AES based on the aircraft location and velocity vector and an ASSUMED nominal satellite location (with no satellite velocity information). That’s it.
2. The calculation of this uplink frequency pre-compensation does not depend on any communicated “satellite inclination” or other parameters besides the assumed (from a data table) constant satellite altitude, latitude, and longitude for the particular satellite in use. The possibility that the satellite inclination parameters were communicated from the GES to the AES was an incorrect notion early on about how the SDU worked, and which we now know is incorrect.
3. Therefore, you can’t “spoof” the SDU using satellite inclination transmitted to the AES because it does not get any of those parameters from an external source.
4. Even if you changed the internal SDU value for the satellite inclination, you are not affecting the actual downlink Doppler. At most you are changing the location of the satellite used to compute the pre-compensation Doppler shift.
5. To manipulate the BFOs using the SDU you would have to re-program the SDU, while in flight, so the nominal satellite had a velocity which was time-dependent in a way to make an actual northern route match the BFOs for a pre-planned southern route.
6. In my opinion, it would be virtually impossible to write the SDU code in advance so as to make an actual northbound route match a pre-planned hypothetical southbound route within a few Hz in the BFOs.
7. Another possibility that gets you part of the way there is to change the sign of the aircraft latitude and the sign of the N-S velocity component received by the SDU from the IRU. This also seems very difficult to do in flight without disrupting the FCS.
8. The notion that the BFOs can be manipulated in flight to match the predicted BFOs of a southern route seems like a fantasy to me. That’s not to mention the fact that nobody at the airline or at Boeing, and almost all of the Inmarsat personnel, had a clue what a BFO was in 2014, much less that it was being recorded.
@DrB
@Victor Iannello
et al
Re: “8. The notion that the BFOs can be manipulated in flight to match the predicted BFOs of a southern route seems like a fantasy to me. That’s not to mention the fact that nobody at the airline or at Boeing, and almost all of the Inmarsat personnel, had a clue what a BFO was in 2014, much less that it was being recorded.
That was a point made by Peter Foley during his lecture at the Royal Aeronautical Society in Canberra the other week.
@All
I have an essay (back in 2021) on the sim cases. My interpretation is it represents a flight south to SIO, based off the Jeddah flight. I found that the sim fuel loaded at N10 was exact amount needed to reach the Magnetic South Pole using at 180 South Mag path, and my working hypothesis is that was the basic structure from which variations for different MAS flights may have been worked out. Active pilot to end. We now know from ATSB these were types of MSFS temp files(thanks to Mick Gilbert working with ATSB to get more info a few years ago). (The data is still partially secret – I say apparently redacted – but ATSB has it and has given guidance.)
Pilot probably was not aware of the temp files, so the old idea of him intentionally leaving data to be found is less good…looks candid data. This is my opinion, but I believe Malaysia hid this data for the obvious reason it is highly sensitive, and I would speculate FBI advised Malaysia and presumably Saudi Arabia of the plan, which FBI would have to take seriously. As MikeG says 3 pilots to Jeddah, so if I am FBI, I am wondering how many involved? I further speculate if Malaysia had released the sim data, even as late as 2016, it would have been explosive news. Instead we had redacted unclear data, that Victor and Yves worked hard to interpret partial data.
That work by Victor stands the test of time, but we do have more data now, so I no longer think NZPG was the target…I know it looks like it…just bad luck the data lined up to NZPG, and Malaysia’s fault for redacting and holding secret the sim data. But it looks like Malaysia tried to give what they could, but it was too late to really allow factoring into prior searches.
@DrB
Many thanks for your March 16th 9:54 pm posting. To start with: I’m also not sure the tangent method could in the end help to select candidate paths or limit a distribution.
Typically, there is not a single minimum dr/dt as there could be different valid polynomial fits to the BTO derived data points r_i (t_i).
Hence, the question would be if one could define a range t_min – t_max for the time range in which the minimum typically should occur, and a connected set of polynomial functions. The method would then become something like: for a path to be potentially valid it should have a velocity difference vector tangent to a virtual ping ring at the aircraft position at t_0. The virtual ping ring connected to a polynomial minimum r(t_0) where t_0 is in the range t_min – t_max.
A first question therefore on your points 6: Is the path that gives a combined zero range rate at 19:41:02 with track 178.2 degrees indeed a valid path / could you share some more details?
@DrB said: The pre-compensation of the expected uplink Doppler is computed in the AES based on the aircraft location and velocity vector and an ASSUMED nominal satellite location (with no satellite velocity information). That’s it…The calculation of this uplink frequency pre-compensation does not depend on any communicated “satellite inclination” or other parameters besides the assumed (from a data table) constant satellite altitude, latitude, and longitude for the particular satellite in use. The possibility that the satellite inclination parameters were communicated from the GES to the AES was an incorrect notion early on about how the SDU worked, and which we now know is incorrect.
Sorry, but this is absolutely false. As explained in the paper, the satellite inclination and ascending node time are broadcast as parameters in the System Table. The SDU has the capability of using these parameters for its AES pre-compensation. As it happens, the inclination was broadcast as zero for IOR at the time of log-on for MH370, so the pre-compensation is calculated as if the satellite was above the equator.
Even if you changed the internal SDU value for the satellite inclination, you are not affecting the actual downlink Doppler. At most you are changing the location of the satellite used to compute the pre-compensation Doppler shift.
You are misunderstanding. There is no need to compensate or understand anything about the downlink Doppler, or any other component of the BFO chain other than components that are route specific.
To manipulate the BFOs using the SDU you would have to re-program the SDU, while in flight, so the nominal satellite had a velocity which was time-dependent in a way to make an actual northern route match the BFOs for a pre-planned southern route…In my opinion, it would be virtually impossible to write the SDU code in advance so as to make an actual northbound route match a pre-planned hypothetical southbound route within a few Hz in the BFOs…Another possibility that gets you part of the way there is to change the sign of the aircraft latitude and the sign of the N-S velocity component received by the SDU from the IRU. This also seems very difficult to do in flight without disrupting the FCS.
As explained in the paper, the BFO spoof only requires a change in two satellite parameters. Basically, the inclination parameter is set to twice the true satellite inclination. No other real-time or a priori calculations need to be made.
I don’t think the BFO was spoofed, but not for the reasons you listed which are mostly incorrect.
sk999 said: Your wrote that “As part of the log-on process, an AES updates its System Table …” That is not necessarily true. The system table is in non-volatile memory and survives a power reset. The system table maintains a revision number. Only if the revision number is out-of-date is the table updated. (At least, this is true for the MCS-7200.) So the inclination will not necessarily be reset at the 24:19 logon.
I understand this. The assumption is the spoof involved a change in the revision number of the table that is stored.
One question I’ve had that has not been answered is what occurs if the revision number broadcast by the satellite changes after log-on. Does the SDU update the System Table?
One penetration vector I considered long ago is whether an onboard GES emulator after log-on could override the actual P-channel signal and broadcast a System Table with fake inclination parameters with a new revision number. Then, the table could be broadcast with the same fake inclination parameters but with the correct revision number. With these parameters, stored, the fake inclination parameters would persist even when the true System Table is received because the revision numbers would match.
In any event, I don’t think the BFO was spoofed.
@Niels said: Typically, there is not a single minimum dr/dt as there could be different valid polynomial fits to the BTO derived data points r_i (t_i)
Looking at the polynomial fit of the aircraft to satellite range at 19:41, 20:41, 21:41, 22:41, and 24:11, the standard deviation of the BTO error from the fit is 74 µs (range error of 11 km) for a second-order fit, while that for a 3rd order fit is 10 µs (range error of 1.5 km). For the third order fit, the minimum range occurs at 19:55.
@DrB. You said: 13. Southbound routes from 188-195 degrees are rejected because the 19:41 location is too far south from the last radar contact position to be reached on time.
This is incorrect. The best-BTO fit no-loiter from a turn initiated off N571 takes you to 39.3S. There is ample opportunity to reach the requisite 1941 latitudes for 7th arc termini 38.8-39.8S without requiring speeds >M0.84 at FL340.
Why would you expect anything other than an imperfect polynomial fit if we know there were significant headwinds for last hour of all candidate paths thereby “distorting” the arc 5-6 portion? What’s more interesting is that fixed speed great circle path produces perfect fit to arcs 2-5 (but predicted overshoot of arc 6 to the tune of error ~330), illustrating need for slow down.
@Mick Gilbert
You asked how I could remember his face after two years? The reason was that he did something unusual and impolite. I had a foot injury at the time, so I was walking relatively slowly for 2 weeks. He took advantage of that and sqeezed into the line (jumping a long line) for the self-service table right in front of me. I almost dropped my coffee, and I got really angry, thinking what an arrogant bastard. Therefore I remembered his face.
Concerning identification, both men had pilot uniforms. In addition, there were several women in stewardes uniforms in the breakfast room at the same time.
Initially (in 2014) I was a bit uncertain if they were from Malaysia Airlines or some other Asian carrier. Therefore I picked the same hotel when I had business in Sydney again in 2016. It was very slear that Malaysia Airlines used this particular hotel as their hub in Sydney, and there were no staff members from other airlines there.
I know it looks like a conspiracy theory, but I can prove everything. There were roughly 150 people who attended my talk later that day (and many noticed my foot problem), one of them gave me a lift back to the hotel twice (same reason). In addition, I have a copy of the receipt from the hotel, proving that I did indeed have breakfast that morning at a cost of 25 AUS$. Some mornings I had a burger across the street instead, since it was faster and cheaper.
@All
Concerning a possible U-turn soon after 19:41.
What if I can prove that an airplane did indeed make a U-turn exactly at the right spot at the right time using other independent information (not from Inmarsat)? I guess that would change the situation.
@Victor
There was a (hopefully obvious) mistake in my post 🙁 “.. minimum dr/dt =o”
I also found the minimum range for the 3rd order polynomial around 19:55. However the low sd in BTO error for the third order polynomial might suggests there is some room to slightly morph the function and hence move the minimum. The question what kind of function to expect would also be relevant. It’s interesting but I’m not sure if I follow much further down this avenue..
@Niels,
You asked: “A first question therefore on your points 6: Is the path that gives a combined zero range rate at 19:41:02 with track 178.2 degrees indeed a valid path / could you share some more details?”
Regarding point 6, the relative range rate between the satellite and 9M-MRO is zero at 19:41:03 on a bearing of 178.170 degrees at the UGIB route location of 2.9350N 93.7875E using 480 kts ground speed. That is not the best 19:41 location for an entire route that also minimizes the relative range rate at 19:41:03. Fitting a route so the 19:41:03 relative range rate is equivalent to less than 1 Hz, and minimizing the BTORs/BFORs, I get a 19:41:03 bearing of 178.28 degrees and a position of 2.91N 93.61E. The ground speed is 470 kts, the normalized BFORs are 1.00 and the BTORs are 1.61 (which is a bit large). This is a marginally valid route. I don’t see any reason for the relative range rate to be near zero at 19:41:03 except coincidence.
@Paul Smithson,
You said: “This is incorrect. The best-BTO fit no-loiter from a turn initiated off N571 takes you to 39.3S. There is ample opportunity to reach the requisite 1941 latitudes for 7th arc termini 38.8-39.8S without requiring speeds >M0.84 at FL340.”
As far as I know, no one has demonstrated a flyable route from the last radar contact at 18:22 that seamlessly connects to the 188-195 degree bearing routes while satisfying ALL the intervening 18:25-18:40 satellite data and which has sufficient fuel. Sure, if you ignore what happens between 18:22 and 19:41, you can connect those routes, but I don’t recall seeing a route example that fits all the BTOs and BFOs. If I am wrong about this, please provide a link to such a route. The fuel is also a huge issue, because you can’t achieve 00:17:30 MEFE if you cruise continuously after 18:22. You need a period of fuel savings, such as a HOLD or some legs low and slow.
David says: « I will send @VictorI a copy in case he can get around that issue. Perhaps Mike Glynn or @Andrew might be able to help. »
Maybe I can help with this link to an archived copy of your newspaper article about the flight sim:
https://archive.is/h5XKs
@DrB says: « I don’t know how the SDU computes the pre-compensation Doppler shift. »
@Victor Iannello:
Doesn’t Thales help with the search for MH370 by providing the required information about the SDU ?
Can’t the IG obtain the required information from Thales rather than the hard way through number crunching ?
You all put a lot of effort into something the manufacturer could provide, one would hope.
@Peter Norton: Thank you!
@DrB. I don’t take 1822 as a hard data point but prefer to start path models at the 18:28:15 arc in vicinity of N571. As long as you include turn dynamics (AOB 15) you will find that pretty much any prior with single turn will end up on a bearing between 189 and 191 dependent on where arc 1 is crossed. You asked for a full route. As previously proposed try MEKAR-IGEBO-RUNUT-end at FL360 M0.840. Alternatively, for FL340 models start at 18:35:00 with BTO=12160 anywhere between 7.2N and 6.2N; or for FL360 models start at 18:35:00 same latitudes, on BTO=12130. All offer excellent Bto fit with great circle, constant M. Fuel adequacy is another matter but this conundrum afflicts all solutions, including UGIB, unless unusual fuel saving actions are inserted.
Viking says: « why he turned the SDU back on (a mystery for many investigators) »
@Andrew:
AFAIK there are 2 ways to turn off the SDU:
• circuit breaker in the EE bay
• isolating the left A/C bus in the cockpit (which seems to be the consensus here)
There is no other way, correct ?
I would assume flying sans left A/C bus to be quite stressful due to the barrage of triggered warnings and (I assume) lost A/P (which incidentally also explains why MH370 was piloted manually post IGARI) ?
So wouldn’t it be very obvious for the pilot to reconnect the left A/C bus (and thereby the SDU) for a less nerve-wrecking hours-long flight into the SIO with A/P and without all those distracting warnings ? Or would the pilot have been able to disable all those warning ?
This would easily explain why the SDU was turned back on (but not really why it was turned off in the first place).
sk999 says on March 16, 2024 at 9:59 pm:
> I would guess that both ground speed and ground track are used by the
> Doppler compensation calculation but are not used by the beam steering unit
> (which instead uses heading, along with other parameters).
Could this divergence be used to determine whether the BFO data was spoofed ?
> DrB says on March 16, 2024 at 7:14 pm:
> 1. I have done a comparison of the MH370 BTO and BFO errors for northbound and southbound routes.
Many thanks for your work and sharing your results.
> 13. Southbound routes from 188-195 degrees are rejected because the 19:41
> location is too far south from the last radar contact position to be
> reached on time.
I would caution against relying too heavily on the radar data.
Can we know for sure that both the location and the timestamp are precise ?
> 24. In my opinion, this exercise demonstrates that no northbound LNAV route can possibly be consistent with the Inmarsat data.
While a straight path seems logical to me for an hours-long flight over the absolute nothingness of an ocean, I would not assume a straight northbound path over land since it needs to take into account, among other things, radar coverage and minimizing the possibility of detection (which can also be observed post IGARI).
> TBill says:
> March 15, 2024 at 10:20 am
>
> Peter’s need for flight south verification I am afraid mirrors even the most
> bright NoK. They are hearing a barrage of conspiracy theories which are
> attractive vs. the apparent truth.
@TBill:
First of all, thought-terminating clichés such as “conspiracy theory” should not have a place on a thoughtful scientific blog. Conspiracies occur all over the planet, every minute, every hour. Thus whether or or not a theory involves a conspiracy says nothing at all about its validity.
Second of all, your statement doesn’t do me justice. What is bad about “verification”? Science is based upon verification, not guesswork. I never said I deem a northern path likely (though I freely admit being intrigued by Jeff Wise’s theory due to it’s sophistication). But the point is this:
A northern path (with tampered BFO data) has not yet been ruled out mathematically.
Only by circumstantial evidence.
So however unlikely, apparently it’s possible that MH370 flew north.
Wouldn’t it be great to eliminate the entire northern hemisphere definitely once and for all by demonstrating mathematically that a northern path is impossible ?
As you say yourself, some NoK may still cling to that tormenting hope. And there is a Netflix documentary, a weekly podcast and articles swirling around, nurturing this possibility and creating doubt about finding the plane in the south. So if that’s where the plane truly is (in the SIO), then all this creates a lot of distraction for the search efforts. That’s why I said, a mathematical proof for the impossibility of BFO spoofing and thus for a northern route would help focusing the search in the south.
@Paul Smithson
What is a typical 19:41 lat for the route ending S39.3?
@DrB
Please correct me if I am wrong. My understanding is:
(1) A stationary satellite doesn’t tell us whether MH370 went north or south:
« With a stationary satellite above the equator, it is impossible to distinguish between “mirror-image” routes N and S of the equator based on BTO (range) and BFO (range rate) measurements. »¹
(2) But our moving satellite does tell us MH370 went south, because due to its N-S movement “the mirror image of the southbound route can’t fit the same BFOs”.¹
(3) So the only way MH370 could have flown north (despite the incompatible BFO data for a northern path) is by spoofing the BFO data (i.e. Jeff Wise’s theory based on Victor Iannello’s “brilliant work showing how simply it could be done just by changing a single parameter in the SDU”²)
I think the above points are nothing new but have been settled years ago.
What’s seems to be new is @Victor Iannello’s recent statement above:
« @DrB: That is VERY clever! […] I really doubt anybody has considered this before. So with that in mind, it would become harder to spoof the BFO with fake satellite inclination parameters so that a northern path has the BFO signature of a southern path, as the slope of the sawtooth would be different for a northern and southern path, even though the BFOs might be very similar. »³
I took this as saying that based on your new analysis (specifically the different sawteeth for a northern and a southern route) “it would become harder to spoof the BFO” as @Victor said. Maybe not only harder but outright impossible? If so, could the slope of the sawteeth be a definite proof for either a northern or a southern flight and thereby settling this issue once and for all? This would be an enormous breakthrough! When I read @Victor’s posting, I thought you were onto something very exciting, that could toss the possibility of BFO spoofing out of the window.
If not, then why did @Victor say, “it would become harder to spoof the BFO” ?
What makes BFO-spoofing harder? And why?
And in what way would it become harder than “just changing a single parameter in the SDU” ?
What additional efforts would be required ?
–––
¹ https://mh370.radiantphysics.com/2024/03/05/ocean-infinity-proposes-new-search-for-mh370/#comment-36239
² https://www.jeffwise.net/2017/01/05/zaharie-shahs-secret-psych-evaluation/comment-page-1/#comment-200608
³ https://mh370.radiantphysics.com/2024/03/05/ocean-infinity-proposes-new-search-for-mh370/#comment-36116
@Peter Norton: My statement was not correct. We originally thought the sawtooth was the results of AEC pre-compensation held constant and the BFO slope between jumps was essentially the changing uplink Doppler. If this were true, then the uplink Doppler would essentially be the same for all paths that satisfy the BTO criteria, which means my statement was wrong. But things are even more complicated because the BFO slope between jumps appears to not be the rate of change of the uplink Doppler, i.e., there might be other factors that contribute to the BFO slope, whether intended or not.
So, perhaps Bobby can still extract useful information from the BFO clusters, but I think it will be more difficult than I originally envisioned.
@Niels.
Start 183500 at 6.7N 95.0711E. Fl340, M0.836, initial azimuth 190.128. I get 1941 latitude -2.3688 and longitude 93.467. My Bto all calculated at geo height 35000ft. 7th arc 39.39S. Let me know if that works for you.
@Peter:
Re: “Can’t the IG obtain the required information from Thales rather than the hard way through number crunching ?”
Believe me, I am trying. I contacted a Thales engineer I met in London in 2017, but he has not responded. I am continuing to seek the critical information we need through several other contacts. Hopefully, one will pan out soon.
In the mean time, here is what we know so far:
1. The Racal Prototype hardware flown on 3 planes in the 1989-90 time frame used a TX up-conversion chain that had “…100 MHz and 200 MHz local oscillators for up/down conversion, frequency selectable in 1 Hz steps to counteract the effects of Doppler…”.
2. We know more details about the SD-700 SDU. It had an up-conversion scheme using a 209 MHz Channel Synth, 164 MHz Doppler synth, but with 10 Hz steps, and a final mix using a 1270.08 MHz LO to the 1645 MHz TX band. Details here: https://bit.ly/3v5QaFz
3. The later SD-710/720 and the companion HS-720 had provisions for 1 Hz Doppler compensation steps.
4. We also know the Doppler update rate was software selectable at installation to be either 30 or 60 Hz (default) on the HS-600 (High Speed version of the SDU).
Obviously, both the Doppler synth resolution (1 Hz, 5 Hz, 10 Hz, ???) and the update rate is important to know when try to decipher what BFO patterns mean. From other documents, I think it is safe to assume the update rate is OTOO 50 Hz, but the SD-600 Synth resolution is still uncertain.
@Victor Iannello:
thank you for the clarification.
If as a byproduct of your work, you and @DrB end up disproving a northern path with BFO spoofing, please let us know.
I know most of you don’t see this as necessary, given the debris and the flight sim data, etc. but it would settle the north/south-question for good and put it to rest.
@ALSM:
> Believe me, I am trying. I contacted a Thales engineer I met in London in 2017,
> but he has not responded. I am continuing to seek the critical information we
> need through several other contacts. Hopefully, one will pan out soon.
Thank you for the feedback. It’s good to know that you are on to this. Best of luck.
It’s somewhat unexpected that it’s so difficult to get them to talk. One would assume that the manufacturer would readily help with the search efforts … but maybe even for them it’s not easy to come up with the information for an older product of theirs.
@Paul Smithson
The 19:41 position you mention should be feasible. However I also tend to think the fuel needed will be a problem, especially if you apply FL340 and M084 all the way.
I’ve worked on a method to estimate an upper limit for available fuel at 19:41 by connecting N571 18:30 position to different Lat, Lon combinations at 19:41 (I calculated a complete 2D grid) by a TT straight segment in holding mode at the FL that best matches the distance to travel in 71 minutes.
Even with this “soft” fuel limit applied it seems hard to get beyond S38.0. There are some paths left but the cumulative probability gets really low south of S38. I’m working on a short summary which I can hopefully share by Tuesday or Wednesday. Without any fuel limit applied paths ending close to S40 would imo not be impossible with an early turn, no loitering and staying at high altitude cruise.
@Paul Smithson The 19:41 position you mention should be feasible. However I also tend to think the fuel needed will be a problem, especially if you apply FL340 and M084 all the way. I’ve worked on a method to estimate an upper limit for available fuel at 19:41 by connecting N571 18:30 position to different Lat, Lon combinations at 19:41 (I calculated a complete 2D grid) by a TT straight segment in holding mode at the FL that best matches the distance to travel in 71 minutes. Even with this “soft” fuel limit applied it seems hard to get beyond S38.0. There are some paths left but the cumulative probability gets really low south of S38. I’m working on a short summary which I can hopefully share by Tuesday or Wednesday. Without any fuel limit applied paths ending close to S40 would imo not be impossible with an early turn, no loitering and staying at high altitude cruise.
from my earlier comment: « a mathematical proof for the impossibility of BFO spoofing and thus for a northern route would help focusing the search in the south. »
Just to illustrate my point:
“Ocean Infinity wants to scan the seabed again. Should it?“
Peter Norton asked:
“Could this divergence be used to determine whether the BFO data was spoofed?”
Not that I can think of.
posted 1 hr ago:
> I think even Victor says that the data gaps are not sufficient to hide the whole
> plane, but he thinks pieces of it might be there. Where that would leave the main
> wreckage I don’t know.
@Victor Iannello:
Does this accurately portray your position (I assume not) ?
If not, maybe you want to set the record straight ?
@Peter Norton
RE: “There is no other way, correct ?”
Correct, unless you pull the cables out of the box itself. There is no ON/OFF switch.
“I would assume flying sans left A/C bus to be quite stressful due to the barrage of triggered warnings and (I assume) lost A/P…”
There would be some initial EICAS alerts plus a Master Caution, all of which can be cancelled. Some systems would be lost, but the major systems have a high degree of redundancy, powered by other sources. The autopilot would not be lost.
@Andrew:
Thank you. What practical problems for a pilot would result from disabling the left A/C bus ?
> “The autopilot would not be lost.”
Sorry for my error. Why is the A/P not lost?
From AF447 I remember that the A/P disengaged because of the degradation from normal law to ALT2.
Wouldn’t disconnecting the entire left A/C bus in a B777 also result in degradation of the flight control law to secondary mode, which according to Skybrary results in loss of A/P ?
« Boeing Secondary mode is somewhat similar to the Airbus Alternate Law. When the PFCs can not support Normal mode operation due to internal faults or to loss of information from other aircraft systems, they automatically revert to Secondary mode. Reversion to Secondary mode results in the loss of the autopilot and the pilots must control the aircraft manually. »
@Peter Norton: Do you think my making a comment here will “set the record straight”? If only it was that easy.
@Peter Norton
Yes, the B777 autopilot would be lost if the primary flight control system (PFCS) degraded from normal to secondary. However, de-powering the L AC bus does not cause the PFCS to degrade to secondary mode because of the redundancy designed into the system. Without going into massive detail, essential elements of the system have several identical components powered by different sources. If one falls over, another continues to provide the data the PFCS needs to continue operating in normal mode.
In the AF447 accident, the flight control system degraded to alternate because it lost airspeed data when all the pitot probes were momentarily blocked by icing. That’s a very different scenario to the loss of AC power on one side.
@Peter Norton. Re the Mike Glynn article and your, ‘Maybe I can help with this link to an archived copy of your newspaper article about the flight sim: https://archive.is/h5XKs.
Thanks though that doesn’t open, at least for me.
‘The Weekend Australian’ has told me that they would ‘entertain’ releasing the article from copyright next Monday (here): ie the two days’ elapsed period so far is too short.
Supposing they do and no alternative comes up in the meantime, I will post it then.
@Mick Gilbert. Yes a response from Mike Glynn to what you raise re the simulation would be helpful.
About Peter Foley’s presentation, that is not yet up loaded to the Royal Aeronautical Society’s Australian site and I know not when it will be. Did he have anything more to say that is relevant to a future search, or accident cause, that you can impart?
@Victor. The interview of Tony Romeo, CEO of ‘Deep Sea Vision’ and ex USN on ‘60 Minutes’ Australia.
To access that, some might we willing to brave ‘9now’, which offers free access on disclosure of e-mail address. Enter without a log-in, select ‘News and Current Affairs’, ’60 minutes’, the ‘Latest’, then click on the last segment at 42:06. Then fidget through about 10 unexciting ads.
Finally for those that have had the tenacity, success enters.
Alternatvely, here is the final URL in case that should help, though I believe that will require you to open an account, albeit free, and with challenges as above yet to come:
https://www.9now.com.au/60-minutes/season-2024/episode-7
However, to save that struggle, I present you with my summary, at no cost.
The program indicates:
– that he has a Hugin 6000, shown utilising a standard stern ramp. However the ‘hangar’ of the vessel shown has room for just the one and the vessel is small, trawler size, without boomed cranes. Possibly he has more AUVs and bigger vessels: he says he can cover 4 X the area searched so far.
– he located what may be Amelia Earhart’s Lockheed, though that is still to be confirmed. Once done he says he would “look around and engineer a solution to bring it up.”
– He makes no mention of an ROV and since the (maybe) Lockheed image appears to be from sonar from above, it may be he didn’t have one or the cranage, at least there and then.
-as to MH370, “finding the black box will unravel everything”.
-he has modified the Hugin he says, which has 36 hrs endurance and searches 50m above the seafloor.
– judging by what is on the map shown, by my eye his search would focus on about 32½˚ to 35˚S, 93½˚E. Map annotated SIO then, NOAA, USN, NGA and GEBCO.
Speaking of GEGCO he like others may need prior bathymetry to conduct a search in a new area, though AUVs in place of towed might lessen that.
– he adds that planning is underway and his company, “has been asked to submit a search proposal to the Malaysian government.
– as to whether he would go ahead without Malaysian approval he said he. “would like the Malaysian government to participate a little bit. It is a very expensive search.” So whether he would charge for his search, as distinct from expect a reward, is unclear.
– it may be that he could conduct a supplementary search of selected areas ie outside the main search.
-also, while beyond ’60 Minutes’ scope, there is the issue of quality management and or sonar calibration, though OI has not disclosed either (so far as I am aware).
– were I the Malaysians I would favour disclosure of those and any bathymetry, whoever did a search, and also the outcome if not found, ie including the find-probability-if-there realised, area by area, particularly where steeply sloped and disclosing spots missed. Whether these should be left to searchers is a good question but they are important background for further searches.
-naturally also not covered by ’60 Minutes’ is the scale of any recovery operation, and about which the NOK might be of divided opinion. The scale is evident from ATSB planning for that (eg floating accommodation at the site for many investigators, observers, coroners,the law etc).
@David
G’day David,
Peter’s presentation was very straight up and down, very facts-based. He is a very good presenter.
Peter kicked off with an acknowledgement of the contributions from all the relevant parties (IG got a mention). The presentation itself was structured in basically a chronological fashion; stepping through the basic details of the flight, the initial search efforts either side of the Malaysian Peninsula, the shift to the SIO and the Australian involvement, the surface search, the preparation of the underwater search area, the DSTG work, the selection of Furgo, the first hit out, the drift work subsequent to wreckage finds, analysis of the flap, First Principles Review, the OI work, and the conclusion addressed the newly proposed OI search area.
Unsurprisingly, there was nothing “new” presented.
It is a pity that it hasn’t been uploaded yet, not so much for this group but for the many people who seem to be only now taking an interest in it. It would be a go-to piece for anyone who wanted a roughly 50 minute precis; a terrific foil to the Netflix show and other associated ramblings.
Separately, I got the opportunity to meet and chat with Mike Glynn, and a few of the ATSB people. I had a good yarn with two of the ATSB fellows who had basically been with it from when the balloon went up. I was struck by the magnitude of the task they were dealing with in the early days when the surface search shifted to SIO, and how quickly newly emerging information superseded what was previously the “best” information; a real moving feast.
It was pretty clear that both Peter individually, and the ATSB collectively view it as unfinished business.
@Peter Norton
Apologies. I was reacting to a panel discussion with Blaine, Grace and Naren.
What I was trying to say:
An enormous human and scientific effort has yielded hard evidence that MH370 crashed in the Southern Indian Ocean (radar, Satellite, debris finds).
Failure to find MH370 in a specific SIO search area does not weaken the hard evidence. Rather the hard evidence is insufficient to make any guarantee whatsoever for finding the aircraft.
David says:
> @Peter Norton. Re the Mike Glynn article and your, ‘Maybe I can help
> with this link to an archived copy of your newspaper article about the
> flight sim
> Thanks though that doesn’t open, at least for me.
Sorry, I don’t know why the link doesn’t open for you.
I just verified again, it does work for me.
Here are some alternatives, which all work:
https://archive.fo/h5XKs
https://archive.ph/h5XKs
https://archive.today/h5XKs
https://archive.is/20240315174529/https://www.theaustralian.com.au/inquirer/secret-flight-path-of-mh370-captain-zaharie-shah/news-story/011d30d8539a87b7639e74f324a4a339
Do any of these work for you ?
If not, let me know and I’ll try to find another solution.
@Andrew:
Thank you for your insights and correction of my misconception.
I had always thought that flying without the left A/C bus would cause all sorts of problems and would be very impractical. But if A/P is not lost and all warnings can be cancelled, it may be less outlandish than I thought.
What would be the biggest practical problems (if any) for a pilot when flying without the left A/C bus ?
@TBill:
Thank you. No problem. And I agree with your latest comment. But:
> “hard evidence that MH370 crashed in the SIO (radar, …)”
We have radar evidence for a crash in the SIO ?
> Victor Iannello says:
> Do you think my making a comment here will “set the record straight”?
> If only it was that easy.
@Victor:
I was just trying to help.
Maybe a direct reply over there ?
I actually do think he’ll take that into account.
@Peter
The Mike Glynn article version you posted yesterday worked for me. Thank you.
@Amelia Earhart’s plane:
The sonar image seems to show the plane in one piece under water.
Isn’t that almost impossible after ditching ?
@Victor Iannello:
@Mick Gilbert:
What do you think about Mike Glynn’s flight sim article ?
To me it seems the article mixes facts and speculation which are presented as facts.
Example:
« He created a circle called a range ring centred on 10N. »
I have never heard of that before. Is this a proven fact ?
@Peter Norton: No, not proven, but certainly possible.
@Viking
Just an aside re the two simultaneous phone call attempts:- at the time that info became available I saw it as a deliberate attempt to call both the cockpit phone and the cabin phone so that if one picked up, the other would keep ringing.
I speculated that the ops personnel making the call had a suspicion that the plane had been deliberately diverted and wanted feedback from both pilots and cabin staff and that they were following some SOP to do thst.
@Peter Norton. Response to one of those 5 URLs by my contraption is, “This site can’t be reached” and “archive.ph unexpectedly closed the connection”. Others the same, including that earlier, except the “ph” changes.
Whatever, I see that @TBill now has access. Assuming that applies widely I will leave aside my impending post.
Thanks.
Is it possible that the debris of mh17 flight are implanted to prove mh370 distruction while it still stand somewhere in miliyry airbase
@flatpack
I like your thinking, but how could ops suspect diversion? If they had their thinking caps on, the fact the 1840 calls went thru would be first hint plane still flying.
I did not realize there was two calls until Viking mentioned it.
@flatpack and @TBill
Sounds like an interesting possibility. I need to think a bit more about it.
@All
Concerning a U-turn soon after 19:41, let us start with ‘the smoking gun’. Then we can go for the ‘bullet hole’ and the final proof afterwards:
A geostationary Japanese weather satellite far out over the Pacific Ocean took a picture containing short contrail stripes of both a northbound path (of MH370?) near the southern tip of the longest island in the Andaman Sea and a corresponding southbound leg of a route (turned slightly less than 180 degrees) near the next island @20:30. Between these two, there is a third contrail-segment from another airplane.
A low-altitude Chinese infrared satellite caught the heading segment of contrail from the same (other) airplane further south around 19:50 proving that it was southbound, most likely heading for Singapore based on further contrail pieces seen by the Japanese satellite. More importantly, the Chinese satellite also caught a U-turn (possibly the final major turn for MH370?) along the edge of a cloud far up in the Andaman Sea. The position agrees within 5 km with our theoretical prediction of a U-turn. It is a nice camera shot since the airplane is around 75% through the U-turn. This is a smoking gun for our solution, since U-turns are extremely rare under normal flying conditions over the open sea, – in particular crossing an airway with risk of a collision.
This is the same as finding a smoking gun at a crime scene. Now we just need a bullet hole and a fingerprint to have a final proof it was MH370 making a U-turn.
The phone call family members of passengers recieved after the incident supports my point of view.
@flatpack
Would you care to elaborate on the process undertaken by a ground station “to call both the cockpit phone and the cabin phone“.
According to my understanding, what happened that night
Someone hijacked the cockpit, while other one went down to switch off the communications and after plane went off the radar , they took the plane to some desired place.
Interesting – Flaperon Barnacles.
https://www.youtube.com/watch?v=BaW4mWw_6mY
https://www.youtube.com/watch?v=7EG-tBmXFOE
@Peter Norton
RE: “What would be the biggest practical problems (if any) for a pilot when flying without the left A/C bus ?”
A bunch of things would be lost if the left AC bus were unpowered, but most of them shouldn’t cause too much concern. The main problems would be the loss of TCAS and the Satcom SDU.
@Peter Norton
G’day Peter, regarding Mike’s article mixing facts and speculation which are presented as facts, the range ring is speculative, for sure.
Suffice to say that the data recovered from the Captain’s home flight simulator is easily one of the most misunderstood, and consequently misrepresented, items relating to the disappearance.
@Mick Gilbert
There is no real ‘process’ as such, merely placing two calls in quick succession. If one call is picked up in the cockpit, the other should keep ringing in the cabin, if the cabin terminal is still powered.
@flatpack
What makes you think that you can direct a satellite call to the pax cabin? Do you think it’s set up like an extension?
@Ajay
Realistically, it is the reinforced cockpit doors that tell us the the Captain probably did it. The Capt had means, motive, opportunity, and we have to add to that since 9/11, Fort Knox cockpit barrier to better enable a rogue pilot.
I personally wish we could be blame it on unknown perpetrator, because I have no interest in blaming pilot…my only gripe is that someone in the cockpit can do this so freely. Most deniers want to rule out 3rd party hijack, and go with fire or mechanical, because if it was nefarious, the who-done-it answer is almost undoubtedly ZS. Cockpit door is virtual AI computer, answers that question automatically.
In any case I can accept your scenario, and this person was active whole flight and probably flew at least 150 nmiles from Arc7 to hide the plane. Well planned, and somewhat similar to sim cases. Wink, Wink- but not our pilot.
@All
What do the ‘bullet hole’ and the ‘fingerprint’ consist of?
A satellite picture IR@22:00 from the French weather satellite Meteosat 7 over the Indian Ocean delivers both. The previous pictures (between 19:00 and 21:30) are unfortunately absent because the satellites solar panels passed through the earth shadow, so correction for an hour wind drift is necessary. Slightly north of the position from where Kate Tee observed an airplane diving towards her there is a so-called fall-streak hole in the thin, hazy cloud-cover, proving that an airplane did actually dive at that position (the ‘bullet hole’). In addition, the contrail passing through the hole ‘inverts’ in infrared. Before the hole, it is thin and white in IR (normal for a plane flying above the clouds at night) while after the hole it became thick black (clearly visible) indicating that the airplane was now below the clouds where the contrail absorbs heat from the ocean and cools the cloud above making the IR image turn black with time. When the contrail reached land in Aceh, there was no longer any cloud cover, and it turned white again. This is practically a fingerprint. The only problem is that Kate got the time one gybe wrong as we already suggested in our paper. If her gybe-number were correct, there would most likely not have been any trace of contrail or drop-hole left at 22:00. Instead, there might have been something in the Japanese satellite picture at 20:30, but there is nothing at that position at that time. The southbound contrail stops near the second Andaman Island as one would expect.
Until recently there was also a Japanese satellite picture available from 21:00 (unfortunately now deleted) showing a much smaller fall-streak hole and white contrail until a few km into Bandar Aceh confirming the exact time the airplane passed. This means that it could only be MH370, thereby confirming the fingerprint. There is no longer any doubt Kate Tee saw MH370 flying low and slow over Bandar Aceh.
With correct gybe assignment by Kate Tee, the independent group would probably have solved the mystery several years ago. This is the reason we always urge students to write their observations in the logbook before they forget important details. As a last detail, we would like to point out that with the contrail below the thin, hazy cloud the optical contrast becomes very different, so Kate Tee thought it was smoke instead of just a normal contrail.
For those who might like to regenerate Japanese satellite pictures (Himawari 7) from the Internet Archive try this link: https://sc-nc-web.nict.go.jp/wsdb_osndisk/shareDirDownload/03ZzRnKS#
However, I warn you that they are in a horrible format and all text is in Japanese. I got copies from another source (collaborating university in USA). Unfortunately deleted several years ago (sorry, I forgot to copy the link and the 21:00 picture, so it is lost forever).
Some METEOSAT 7 pictures are still available from Weather Graphics, the rest from other places.
@Viking
Can you please share all the relevant images you refer to and that still exist, and indicate all “drop holes”, “contrails” etc you mention clearly in the images?
@Andrew says:
> A bunch of things would be lost if the left AC bus were unpowered,
> but most of them shouldn’t cause too much concern.
> The main problems would be the loss of TCAS and the Satcom SDU.
IMO both are not problems but advantages for someone who wants to disappear
(which seems to be the case here).
Ok, so if there are no major inconveniences for piloting the aircraft and as an additional bonus the hijacker gets rid of TCAS and the SDU, I see for the first time how a case could be made for intentionally disconnecting the left A/C bus (if it’s possible the hijacker could have had or obtained any knowledge of what systems would be lost).
@Mick Gilbert says:
> the data recovered from the Captain’s home flight simulator is easily
> one of the most misunderstood, and consequently misrepresented, items
> relating to the disappearance.
Good day to you too, Mick.
What’s your current assessment and did Mike Glynn’s article change it in any way ?
G’day Peter,
Regarding, “What’s your current assessment …”
The most simplistic and widely put about interpretation of the Captain’s home flight simulator data is a continuous and contiguous “suicide route” constructed by joining the locations recorded for six flight files that had been recovered; what I’ve always referred to as the “string around the map pins” approach. Simplistic, somewhat damning, and, to my mind, wrong.
When you understand how the recovered flight files were created, and you incorporate ALL of the data, it becomes evident that there are three distinct elements or phases to the session.
In the first phase of the session, the Captain was almost certainly simulating the take-off and departure of his then upcoming flight to Jeddah on MH150; both the take-off time and the fuel load are there or thereabouts for MH150. This was a route that he hadn’t flown for some three months. The data shows that after what appears to have been a fairly standard departure from KL, the Captain simulated what might have been some sort of in-flight upset (we see meaningful changes to the Maximum and Minimum G-force recorded) that was then followed by his turning the aircraft around. This all occurs as the aircraft was crossing over from airway B466 to N571, but before it had reached VAMPI.
This is fairly important, I think. At this point in the session, with the aircraft still in the Strait of Malacca, not having yet reached N571, he was turning the aircraft away from a flight path further up the Strait and likely either heading back to his departure airport, KL, or possibly heading to the then nearest suitable airport, Medan, Indonesia. Simply joining this point to the next point over in the Andaman Sea doesn’t make a lot of sense in terms of it being either continuous and contiguous.
That first phase of the sim session ended there with the simulation aircraft just shy of VAMPI at FL400, turning left through 255/256° in a 20° AOB turn.
After that, the aircraft was relocated manually using the Map function to the eastern edge of the Bay of Bengal, roughly equidistant from Port Blair and Car Nicobar. While the Captain had flown through that region on the MH17 return flight from Amsterdam to KL just two weeks prior, it is probably worth noting here that the aircraft may not have been repositioned with any degree of specificity – it is not on an airway but it is close to the E90-N10 grid intersection that’s shown on the simulator map.
From the Bay of Bengal location the aircraft was then flown back towards Indonesia/Malaysia such that the NAV1 and NAV2 navigation aids autotune to the Banda Aceh VOR. This is important – the aircraft did not fly south towards the next sim point in the SIO, rather it flew south-east possibly still tracking back towards its destination (KL or Medan) from the previous phase in the Strait of Malacca.
Notably, during this phase the simulator aircraft was almost certainly jettisoning fuel such that it reached fuel exhaustion somewhere near Banda Aceh, Indonesia. It began a brief unpowered descent after fuel exhaustion had occurred. The second part of the sim session ended at this point, with the aircraft having reached fuel exhaustion somewhere near north-western Sumatra.
Again, it is probably worth stressing here that the simulation aircraft reached fuel exhaustion somewhere in the northern hemisphere, probably just off the north-west coast of Sumatra. This is contrary to the commonly put about notion that the aircraft was flown to fuel exhaustion deep in the Southern Indian Ocean.
Something else worth noting here, this phase ran for about 42 minutes elapsed sim time. Elapsed sim time is not necessarily the same as “real world” elapsed time, as sim time can be compressed by accelerating the sim, and stretched by pausing. But it should be noted that on the day that the flight simulation was undertaken, Captain Zaharie took a 45 minute phone call from his first cousin and aircraft engineer, Zulhaimi Bin Wahidin, commencing at 9.49am. Given the similar duration for the fuel exhaustion flight leg I have wondered whether the two might have been coincidental – for example, he started the fuel jettison and was then distracted by the phone call and only got back to the simulation when the tanks were dry. There’s no evidence that the two are linked but it would be remiss not to mention the possibility.
After that second phase, the simulation aircraft was then relocated to the Southern Indian Ocean about 940 nm south-west of Perth, Australia. Here again, it is probably worth noting here that the aircraft has been relocated close to a grid intersection that’s shown on the simulator map, this time E104-S45. Shortly after that relocation, after another brief period of unpowered descent, the aircraft’s altitude was manually adjusted down from its then current 37,650 feet to 4,000 feet.
Taken in its entirety, is the sim session suspicious? Probably, yes. Could it be construed as practising a diversion south into the SIO to fuel exhaustion? I don’t think so, not in any practical sense.
Regarding “… and did Mike Glynn’s article change it in any way ?”
Mike certainly makes an interesting observation, one that hasn’t been touched on before. I don’t think that it rises to the standard of “smoking gun”.
@Niels
Please send me an email, so I can send pictures to you.
@Viking
I expect I’m not the only one interested to see evidence/data supporting your claims.
You explained you release your findings step by step to have discussion on each section. We can only have a proper discussion here if you share the supporting data/evidence/methods that go with the claims. Can’t you share relevant images and/or parts of images through a cloud link? Or write a summary on this part including images and share that to start with?
@Mick Gilbert
“That first phase of the sim session ended there with the simulation aircraft just shy of VAMPI at FL400, turning left through 255/256° in a 20° AOB turn.”
How far short of VAMPI ? ( lat / lon )
@Victor Iannello
After spilling iced-tea into my laptop several weeks ago which trashed my harddrive, I find that several files I was working on were not on my last harddrive image back-up.
There was a paper published by a working group (Inmarsat?) showing communications traffic from a 777 parked on the tarmac from 16/09/2014 14:46 to 17/09/2014 09:13.
There were two figures showing the BTO error, Figure-1, and Figure-2 and an analysis that justified the 59 microsecond error that subsequent investigators have used in their models.
I need Title, Authors and download link so that I can cite the paper in the study I am working on.
Any help would be appreciated…
Regards,
George
@George Tilton: You may be referring to a paper entitled “Burst Timing Offset (BTO) Characteristics” published on Dec 23, 2014, by the ATSB, which described work performed by the MH370 SATCOM Working Group.
@Victor,
Thanks!
I will google that now that I have a title.
@Peter Norton
In case you are interested, I have very short YouTube showing how easy it would be to isolate LEFT BUS to depower SATCOM. I made the video a few months ago for Jeff Wise, in his podcast he told listeners this was a very difficult task. You be the judge.
https://www.youtube.com/watch?v=SmP1t2cOTJk
@All,
Measuring the slope of the pre-compensation Doppler shift imposed on the AES uplink transmitter frequency is possible during several periods when a sufficient number of independent BFO measurements were made in rapid succession during Flight MH370.
These periods occurred circa 18:40 and 23:15 UTC.
The observed BFO slopes are unexpectedly large, causing saw teeth to appear in the observed strings of BFOs in three MH370 and MH371 cases I have studied. Three effects are observed in each BFO sequence. First, there are discontinuous steps in frequency of about 5.5 Hz. These appear to be due to the SDU using a nominal range rate variable which is quantized in units of 1 m/s. Second, these steps occur about 5.5 times more frequently than one expects (due I think to a coding error in the SDU). This conclusion is based on the predicted pre-compensation Doppler slope during the MH371 example when we have simultaneous recorded data for track, speed, altitude, ROC, etc. Third, the exaggerated slope in the BFO data does not accumulate a large BFO error, because the 5.5 Hz steps occur whenever the accumulated error reaches 5.5 Hz, triggering a reset to a small error and creating the “steps”.
The steps seen in the BFOs can occur in less than 0.2 seconds, demonstrating they are not due to actual aircraft acceleration or to thermal effects in the OCXO used to establish the base transmitter frequency.
All the examples I have studied appear consistent with this model of frequency control in the SDU.
While I believe the SDU code contains an error, the fact that it exaggerates the BFO slopes is a blessing in disguise. That allows me to measure very small Doppler frequency slopes, and this fact allows me to place rather tight constraints on the aircraft accelerations during the two MH370 phone calls. These constraints then make it possible to discriminate certain routes at those times.
My theory of how the SDU actually works is unproven. It is a logical explanation for the observed behaviors. I note that we don’t need to recover any part of 9M-MRO to test this theory. All that is needed is to inspect the code listing for the SDU software version used by 9M-MRO, which should still be available today from the manufacturer. I leave this exercise to others.
To figure the constraints on the aircraft accelerations during the two MH370 phone calls, I first need to estimate the error in the observed BFO slopes. At 18:40 the observed “baseline” BFO is 88.9 Hz and the slope is -103 Hz/min. At 23:15 the observed baseline BFO is 219.5 Hz and the slope is -14.9 Hz/min. Correcting for the slope error factor of 5.5 X, the observed BFO slopes were actually -18.7 Hz/min at 18:40 and -2.7 Hz/min at 23:15. Based on the sensitivities observed when fitting a sawtooth waveform to the measured BFOs, I estimate the fitted period of the sawtooth waveform cannot be in error by more than +/- 10%. That, in turn, constrains the slope error to also be +/- 10% of the measured slope. Therefore, at 18:40 the actual BFO slope error is +/- 10% of -18.7 Hz/min or +/- 1.9 Hz/min. At 23:15 the slope error is +/- 10% of -2.7 Hz/min or +/- 0.3 Hz/min.
In order for a MH370 route to be consistent with these constraints, at 18:40 the predicted BFO must match 88.9 Hz within about 3 Hz (at one sigma) and the BFO slope must be consistent with -18.7 Hz/min +/- 1.9 Hz/min. Thus, the BFO slope should probably lie between -16.8 H/min and -20.6 Hz/min.
At 23:15 the BFO must match 219.5 +/- 3 Hz (one sigma) and the BFO slope is expected to lie between -2.4 Hz/min and -3.0 Hz/min.
I will address the 23:15 case first, because it is simpler to analyze, with no expected or proposed major turns or climbs. The first question is, what limits can be placed on LNAV routes in terms of ground speed changes or track changes?
For the UGIB (2020) route at 23:15, which is on a track of exactly 180 degrees at LRC at FL390, I predict a BFO of 218.4 Hz and a nominal uplink Doppler rate of -2.7 Hz/min. Thus, both 23:15 route conditions are easily satisfied by the UGIB (2020) route. Next, I evaluated the sensitivity of that route in terms of track change (true bearing rate) and speed change. The sensitivity is ¼ kt/min of speed change per -0.3 Hz/min in nominal uplink Doppler. Therefore, the acceleration in ground speed is very tightly constrained to be a fraction of a kt/min. Certainly, no manually-induced or even FCS-induced speed setting changes were underway at 23:15. The ground speed was remarkably constant during the phone call then. The sensitivity to track changes at 23:15 is +0.025 deg/min per 0.3 Hz/min. Therefore, the track bearing was also remarkably stable at 23:15, changing by less than 0.025 degrees during the phone call.
The next question is do other proposed routes also satisfy the 23:15 BFOs and slope?
I analyzed two cases. First, I looked at the “39S Route” flying a geodesic with an initial bearing of 191 degrees at 19:41:03 from 3.50S 93.28E at a ground speed of 482 kts at FL360. The predicted location at 00:11 is near 38.88S. At 23:15 this route had a predicted BFO of 226.9 Hz, which is 7.4 Hz, or more than two sigmas, above the predicted value. This is a very unlikely match. The predicted nominal Doppler slope with no accelerations is -3.0 Hz/min, which is a reasonably acceptable match to the observed value of -2.7 Hz/min +/- 0.3 Hz/min (at one sigma). Next, I analyzed the Doppler rate due to the changing track of the geodesic route (which is about 0.018 deg/min). This bearing rate induces a nominal Doppler rate of about -0.3 Hz/min, so it is not possible to exclude this route based solely on the Doppler slope. Its two main deficiencies are the poor BFO match and a lack of fuel.
Second, I looked at a 186 degree route, ending near 36.61S at 00:11, starting at 0.77S 93.63N at 19:41:03, and flying at FL390 with a constant ground speed of 480 kts. The predicted 23:14:30 BFO for this route is 223.4 Hz at 29.14S 90.37E at a bearing of 186.867 degrees. The predicted BFO is 3.5 Hz high, which is acceptable. During one minute of flight the geodesic’s bearing increases from 186.867 to 186.876 degrees. The predicted nominal uplink Doppler slope is -2.7 Hz/min, which is quite acceptable. The sensitivity to speed changes is the same as the 180 degree route, being about ¼ kt for a -0.3 Hz/min slope. The sensitivity to bearing changes for the 186 degree route is also nearly identical to the 180 degree route, being about +0.025 deg/min per 0.3 Hz/min. The changing geodesic route bearing is only about +0.009 deg/min at 23:15, so one can’t distinguish between a 186 degree True Track path or a geodesic by using the Doppler slope method.
To summarize, assuming my theory of the SDU coding errors is correct, the 23:15 BFOs imply a bearing rate limit of 0.05 deg/min and a ground speed acceleration limit of 0.5 kts/min, both at two sigmas. The 180 and 186 degree routes fit the BTOs, BFOs, and the 23:15 BFO slope very well. The 191 degree route has a large, and therefore low probability, BFO error at 23:15, but it fits the BTOs and the BFO slope acceptably well.
It is clear, because of the very low speed change and bearing change limits, the aircraft was being autopiloted and probably was following a true track or LNAV geodesic route with normal automatic cruise speed control. No turns or climbs/descents were underway then. I also doubt the aircraft was in a constant heading mode for lateral navigation. Certainly, this would not be sufficiently stable in bearing to demonstrate such tiny BFO slope errors.
There is the question of whether a magnetic track route can be excluded. A magnetic track route is highly unlikely for two reasons: (a) the BTO/BFO residuals don’t match the data nearly as well as the “straight” routes, and (b) the true bearing in the SIO would change rapidly with location. For example, the UGIB (2020) route has a magnetic declination which changes by 8 degrees between the 23:15 and the 00:11 locations. That is an average rate of 0.143 deg/min, which is equivalent to a BFO slope of +1.7 Hz/min. This high positive slope is well outside the measured slope of -2.7 +/- 0.3 Hz/min. Therefore, I can say with confidence that the MH370 BFOs at 23:15 are incompatible with constant heading (either true or magnetic) or magnetic track navigation.
The MH370 route possibilities at 18:40 are more complex because of potential descents and/or turns. I’ll post my analysis of that data when it is complete.
@Niels and @Victor Iannello
I have spent most of the day preparing 4 files with illustrations that I can upload for further discussion. Please tell me where you would like them.
1) Overview picture showing where the different evidence is collected (PDF around 150kB)
2) Interactive Powerpoint presentation illustrating results from the low flying Chinese IR satellite (2.5MB)
3) Picture from Himawari-7 (relatively poor quality) showing the 3 contrail segments. The U-turn is also left as a faint ecco near the NW corner in agreement with the Chinese data (PDF around 150kB)
4) Interactive Powerpoint presentation illustrating results from Meteosat-7 near Bandar Aceh @22:00. Here you can see a faint white IR contrail from the north leading into the the fall streak hole in the middle, followed by the black contrail leading to Bandar Aceh. I use the near-IR picture as a reference. Due to lots of water vapor in the muggy night near the equator, the near IR only shows contrail evidence when the airplane is flying high (beginning and end). Actually it is possible to see the airplane starting its dive and retuning to normal height after Aceh (new to me). This file is around 2MB.
Both interactive files are userfriendly, since you only need to push ‘arrow down’ to toggle with the speed you prefer.
@TBill
For what it’s worth, your video appears to show a fault in the software modelling. The EICAS messages that appear in the video when the left AC bus is isolated include HYD PRESS DEM L. That is not correct; it should read HYD PRESS DEM R. The left hydraulic system demand pump is powered by the right AC bus, while the right system demand pump is powered by the left AC bus.
@Andrew
OK thank you. That would be PMDG777 issue. But they were close, it sounds like. I am using the so-called STEAM version I do not know if they keep it current.
What is the implication of losing that hydraulic system?
@ventus45
I estimate that the turn away from VAMPI commences around 10nm before the waypoint, at about 6°3’37″N, 97°42’01″E.
@DrB
OK I am listening…
@Mick Gilbert
Thank you for that position.
@TBill
RE: “What is the implication of losing that hydraulic system?”
The hydraulic system is not lost in this scenario. The L & R hydraulic systems each have two pumps: an engine driven pump that operates whenever the respective engine is rotating, and an electric pump that operates when system pressure is low or when the control logic anticipates high system demand (take-off and landing, for auto speedbrake operation and braking). In this scenario, the R demand pump is inoperative but the associated engine driven pump is operative.
The L & R system demand pumps don’t normally operate during the cruise, so the loss of the R system demand pump has no consequences for the cruise phase unless the associated engine driven pump or engine were to fail. In that event, the R system would be lost, but the other hydraulic systems would provide enough redundancy to keep all the flight controls powered, except for two spoilers (#3 and #12 on the L and R wings respectively).
On reflection, there are two problems with the sim software. It not only shows the wrong hydraulic pump as inoperative, it also displays an EICAS alert when there shouldn’t be one. The HYD PRESS DEM L/R EICAS alert should only be annunciated when the pump output pressure is low and the pump is commanded on. The pump shouldn’t be commanded on in this scenario, so there shouldn’t be any associated EICAS alert. There would (I think) be a status message on the lower EICAS.
You might notice another EICAS alert: HYD PRESS PRI C1, which indicates the Primary C1 pump in the centre hydraulic system is inoperative. Again, that has no consequence during the cruise. The centre system has four pumps, two electric primary pumps that operate continuously and two air-driven demand pumps that operate during periods of high demand. If one of the electric pumps fails (in this case the C1 pump), the remaining pump continues to pressurise the system, with assistance from the demand pumps as required.
@Andrew
Thank you very much for that explanation. Interesting to hear low consequence during flight. LEFT BUS isolation is almost essentially how to turn off SATCOM and not have any other issues. At IGARI, I personally suspect more power was cut off, but certainly LEFT BUS off is apparent minimum requirement.
@All,
1. In order for a MH370 route to be consistent with the satellite data constraints, at 18:40 the predicted BFO must match 88.9 Hz +/- 3 Hz (at one sigma) and the predicted nominal Doppler slope must be consistent with -18.7 Hz/min +/- 1.9 Hz/min (at one sigma). Thus, the nominal Doppler slope should lie between -16.8 Hz/min and -20.6 Hz/min (at one sigma, or with a 68% probability).
2. The UGIB (2020) MH370 Route at 18:40:00 has the aircraft near 7.52N 94.57 E on a 270 degree course while descending at 1,800 fpm and at 250 KIAS at a Flight Level of 203. The predicted BFO then is 86.7 Hz, which is an acceptable match. Thus, both the BTOs and the BFOs circa 1840 are good matches to the Inmarsat data [as previously demonstrated in UGIB (2020)]. What about the Doppler slope at 18:40? Does it match the apparent -18.7 Hz/min?
3. The answer is “yes” because the predicted nominal Doppler slope for this route is -17.1 Hz/min. That is close to the “observed” value (after correction for the scale factor error in the SDU) of -18.7 Hz/min. Therefore, the predicted Doppler slope for the UGIB Route lies within the one-sigma error bars, indicating good agreement.
4. So, there is no discrepancy in the UGIB (2020) route at 18:40 in terms of matching all three observables (BTO, BFO, and nominal Doppler slope). An exact match to the observed Doppler slope occurs when the ROD is 1,980 fpm, but this produces a mismatch in BFOs. The optimum combination of Flight Level and ROD occurs at FL = 271.8 and ROD = 1,711 (when the course is 270 degrees).
5. The UGIB (2020) route circa 18:40 is in a FLCH descent at 250 KIAS while tracking at 270 degrees. Note that, as the aircraft descends, the true air speed decreases with time because of the increasing air density. The TAS change over one minute depends on the Flight Level. The Flight Level is also decreasing with time because of the descent. So, here we have a situation wherein the descent rate is primarily constrained to make the BFOs match, because the BFO is highly dependent on the ROD, whereas the nominal Doppler slope does not depend at all on the ROD, since the SDU ignores vertical speed. So, circa 18:40 we have a fairly accurate estimated position, and we know the airspeed is close to 250 KIAS. The ROD may then be determined to make the BFOs match, given a course, and the Flight Level will determine the rate of change in true air speed during the 1-minute phone call. So, we see that the observed BFO slope provides some additional information about the Flight Level. There is an assumption here that we know the course (bearing). If I use 270 degrees, per UGIB, then I find the following at 18:40:00:
Latitude 7.52 N
Longitude 94.57 E
Track 270 degrees
Calibrated Air Speed 250 KIAS
Delta SAT 10 C
True Air Speed 345 Kts
ROD 1800 fpm
Flight Level 203 hft
Predicted Nominal Uplink Doppler 558.2 Hz
Predicted BFO 86.7 Hz
6. At 18:41 I predict the following:
Latitude 7.52 N
Longitude 94.47 E
Track 270 degrees
Calibrated Air Speed 250 KIAS
Delta SAT 10 C
True Air Speed 335 Kts
ROD 1800 fpm
Flight Level 185 hft
Predicted Nominal Uplink Doppler 541.1 Hz
Predicted BFO 86.9 Hz
7. Both BFOs are within 2.2 Hz of the measured “baseline” BFO value of 88.9 +/- 3 Hz. Therefore, the BFOs are an acceptable match.
8. The predicted nominal Doppler slope is (541.1– 558.2) = -17.1 Hz/min, which is a good match to the “measured” value of -18.7 Hz/min +/- 1.9 Hz/min.
9. So, we see that no adjustment in the proposed UGIB route is needed to match the BTOs, the BFOs, and the nominal Doppler slope.
10. I can get acceptable matches to all observables for courses from 240 degrees to 300 degrees, but some bearings in this range have abnormal descent rates. For example, the ROD at 240 degrees bearing is only 930 fpm, which is abnormally low. At 300 degrees bearing the ROD is 2,400 fpm, which is abnormally high. At 2,000 fpm the best-fit course is 282 degrees with FL = 241. At 1,800 fpm, the best-fit course is 273.6 degrees with FL = 259. At 1,600 fpm, the best-fit course is 266 degrees with FL = 301. At 270.0 degrees course the best-fit ROD is 1,711 fpm with FL = 272. Therefore, the course can’t be far from due West in order for the ROD to be in the “normal” range circa 1,700-2,000 fpm.
11. The nominal uplink Doppler slope observed at 18:40 of -18.7 Hz/min cannot be matched by level flight or any climb or descent in level flight. There must be an acceleration occurring to cause the nominal uplink Doppler slope to become significantly more negative over time.
12. For example, level cruise at a bearing of 205 degrees provides an excellent match to the BFOs, but the nominal Doppler is only -3 Hz/min. Therefore, the benefit of knowing that the Doppler slope is much larger than 3 Hz/min then is that ALL routes with constant track and constant speed may be excluded at 18:40.
13. One way to match the Doppler slope is through an ongoing deceleration. One possibility is to undergo a slow-down of -16.6 kts/min on a track at 202.6 degrees, with no turn and no descent. But why would the aircraft be slowing down in level light? This does not make sense. There are more probable solutions which do not require “non-standard” flight.
14. It is a necessity for the aircraft to slow down during a significant descent from cruising altitude. The UGIB Route demonstrates how a significant Doppler slope is created during a descent because the aircraft is slowing down (not because it is descending). We already know that if we only need to match the BFO value, then one cannot distinguish bearing and ROD. One bearing with no ongoing descent can match the BFO. A different bearing with an ongoing descent can also match the BFO just as well. However, those two cases will have very different Doppler slopes. Only the descent case will have a large-magnitude Doppler slope, because of the deceleration in horizontal speed which occurs at constant KIAS.
15. There is only one other option left to explore. Consider an ongoing turn at 18:40. What turn rate would produce a matching Doppler slope, and is this a reasonable value? One can get a good match to all observables with a bearing which is 205.2 degrees at 18:40 and 203.8 degrees at 18:41. That is a (left) turn rate averaging only 1.4 deg/min over one minute. This turn rate is much too slow to be a standard turn, and it is much too high to be induced by a constant-heading or a magnetic track mode. Therefore, I conclude that a normal ongoing turn or choice of lateral navigation mode cannot match both the Doppler slope and the BFOs circa 18:40. Only an ongoing (standard) descent can do that,
16. If one assumes the aircraft is being flown in a standard manner, regarding autopiloted turns and descents, the only match to all the data circa 18:40 is an ongoing descent with a slow-down in true air speed while on a westerly course, as proposed by UGIB.
17. Southwesterly routes using standard lateral navigation and with no descent cannot match the 18:40 Doppler slope and are therefore excluded.
@DrB: One possibility is the descent at 18:40 was not in FLCH mode but in VS mode, which means there was thrust added to match the MCP speed and the descent rate was controlled with the elevator. This could explain the low variation in BFO values compared to what was seen during the descent of MH371, as the vertical speed is directly controlled, and the vertical speed has a large influence on the BFO.
Why would the pilot choose a VS descent rather than FLCH? Well, if the cabin was depressurized, a rapid descent would rapidly increase the pressure, and this would hurt the pilot’s ears because of the differential pressure across their ear drums. By lowering the descent rate, it gives more time for the differential pressure to equilibrate, and there is less pain.
Private pilots in unpressurized cockpits understand that high descent rates lead to ear pain. With passengers onboard, descent rates are typically limited to around 500 fpm, which also means the descent angle can be quite shallow for faster planes. An experienced pilot can tolerate higher descent rates, but the VS mode would still reduce the pain.
@Victor Iannello
Since you did not come up with any suggestions to where I could upload the four files, I decided to do something different today. Inspired by observation of black contrails through thin clouds, I tried looking for shadows of the contrails I found near Christmas Island. The visible contrails are difficult to see, so few people believed me. However, what about the shadows?
I used the IR camera in Meteosat-7 as a thermal camera. It is somewhat cumbersome, but it produced beautiful black contrail shadows on top of the cloud cover. In some places, the original white contrail is also visible.
This opens a completely new possibility. By simple measurement of the distance to the shadow combined with knowledge of the solar height it is possible to calculate flying height, above the (relatively low) cloud cover. Hereby, I have now shown that the flight height decreased from 8.5 km near Christmas Island to 0 km when the second engine flamed out. Simultaneously the speed fell to below 400 km/hour. I guess the clouds were at around 2 km, so the total height is close to my original estimate (11 km).
I cannot imagine any pilot would allow such a scenario without immediately turning around and try to land at Christmas Island. Since no other airplanes were lost that night, there is only one possible explanation. The contrails were from MH370.
Now I guess you would like to have the original Meteosat-7 IR data. Send me an email and you will get the two relevant files (500KB each).
@Victor Iannello,
It is possible to match the average of the 18:40 BFOs with a 500 fpm ROD, a course of 215 degrees, and a Flight Level circa 250. However, the nominal Doppler slope is only -5 Hz/min, which is far short of the -18 Hz/min needed to match the observed saw teeth in the BFOs.
As I said before, the only “standard” flight maneuver which provides the correct Doppler slope is a slow deceleration, but at 500 fpm that deceleration is still far too low. I don’t know any other method of matching the Doppler slope than the deceleration occurring during a circa 1,800 fpm ROD at constant KIAS. I’m happy to evaluate all suggestions, but 500 fpm does not match. One needs about -10 kts/min in ground speed to produce the apparent Doppler slope.
@DrB
Bobby, regarding “46. There is no evidence of any ongoing climb, turn, or speed change occurring at 23:14 during MH370 or at 04:29 during MH371.” above, how can you be sure about that for MH371?
There are a series of six position reports in the 04:29:19 – 04:30:13 transmission, covering 04:04 – 04:29 in five minute intervals. The aircraft most definitely makes a heading change in the course of that 25 minute period preceding the transmission. There’s also evidence of minor ground speed fluctuations up to about 3 percent during that period.
https://www.dropbox.com/scl/fi/vt6khdr2rsyj2ykcx0p4q/MH371-04.04-04.29-position-reports.-jpg?rlkey=eodovmo4gao3be4yqkatktkb3&dl=0
At the time the transmission is being sent the aircraft is some 90-odd nm south-south-west of Hong Kong. I don’t think that you can eliminate the possibility that the aircraft was at that time turning right towards one of the waypoints on the HK/Sanya FIR boundary. (Andrew might have a view on this.)
Separately, there appears to be a lot predicated on there being an SDU coding error. If there was not the sort coding error you envisage, what are you left with?
@Mick Gilbert,
A plot of the MH371 ACARS latitudes and longitudes from 04:24:01 to 04:39:01 fall on a “perfect” straight line with a fitted R^2 = 0.9999. Therefore, there is no evidence for any course change in that time period. The reported altitude is unchanging at 40,000. The CAS during this period varied from 250.4 to 253.8 kts. Therefore, there were no turns, no climbs, and only small air speed changes, and these were even smaller in the one-minute period from 04:29 to 04:30.
The SDU coding error produces a very unusual signature. It is highly distinctive in both amplitude and period. I doubt there is a different type of error which would have the same “signatures” in the three cases I studied but which would be somehow incorrectly corrected in any of them by the method I used.
If someone could get the code listing from the vendor, we would soon know whether or not my explanation for the unexpected BFO behavior is correct.
@Mick Gilbert
It is no surprice there are a couple of heading changes for MH371 near Sanya. The reason is they not allowed to fly near the large Chinese Navy base in Sanya.
@DrB: I am not suggesting that the descent was at 500 fpm. I understand that does not match the BFO. I am trying to give a rationale for a VS descent rather than an FLCH descent to produce a fairly steady vertical speed to match the fairly steady BFOs.
Viking said: Since you did not come up with any suggestions to where I could upload the four files, I decided to do something different today.
I did not think you were serious when you asked me this. A simple Google search shows that there are so many possibilities: Dropbox, Google Drive, One Drive, Box, and others.
@Viking
The area in question is inside the Hong Kong FIR. It is nowhere near the PLA-N base located near the city of Sanya at the southern end of Hainan Island.
@Mick Gilbert
Aircraft normally enter the Hong Kong FIR from the Guangzhou FIR at a designated waypoint. Once picked up by Hong Kong ATC, overflying aircraft are often cleared direct to a waypoint on the ‘exit’ side of the Hong Kong FIR, subject to traffic.
I don’t know what route MH371 was flying, but from the chart you posted it looks as though the aircraft was tracking direct to IKELA. There are no routes any further West of IKELA, so a right turn doesn’t look very likely.
@DrB
Thanks Bobby, yes, I can see that now.
Are you seeing similar saw-toothing in the BFOs for the 04:49-05:00 transmission when MH371 is established on L642?
@Andrew
Thanks for that Andrew. Somewhat dimly, I had not accounted for the position reporting subsequent to the 04:29-04:30 transmission. From those it appears that the aircraft eventually joined L642 and crossed the FIR boundary at EPKAL.
@Mick Gilbert
EPKAL/L642 makes sense. Aircraft headed to Singapore/Kuala Lumpur/Ho Chi Minh from the Hong Kong area normally go that way.
@Victor
Is there a way to mitigate the ear pain on descent such as ear plugs? Do military pilots have that issue?
Anyways, interesting idea. But I apply it differently. Right now my flight path is level at 1840 (turn southerly). I have long considered 400-ft/min descent after twilight and before 2314 call. Previously I had no particular reason for slower than FLTCH other than it fit the data depending on scenario.
@Mick Gilbert,
There is only one BFO in the MH371 period from 04:49 to 05:00, so you can’t possibly discern a sawtooth then.
However, from 05:29:16 to 05:30:17, 132 BFOs were recorded, maybe half of which are independent (per Don Thompson). A plot of the next dense grouping of MH371 BFOs circa 05:29 is available here.
This plot again shows the saw teeth. The first two cycles are quite plainly seen. The third cycle is muddled, perhaps by changes in the actual uplink Doppler. A vertical speed of only 1 m/s will induce a BFO shift of a better part of 5.5 Hz. So, there might have been some turbulence circa 05:30:00 that distorted the third cycle of the sawtooth waveform in the BFOs.
@Victor Iannello,
@Andrew,
How is the desired vertical speed entered for VS descent mode?
What is the minimum resolution, or step size, in entering the vertical speed?
@DrB: You dial in the vertical speed on the MCP in increments of 100 fpm, in parallel with an airspeed setpoint.
Using FLCH, the thrust is full for a climb and idle for a descent, and the elevator controls the airspeed. In VS mode, the thrust is adjusted to control airspeed and the elevator controls vertical speed. This means that if a sufficiently high VS is selected for a climb, the speed will decrease towards stall. Similarly, if a sufficiently high VS is selected for a descent, the speed will increase towards overspeed. I think for large altitude changes, it would be rare to use VS for the climb or descent. But considering the stability of the BFO values, and the potential to slow the descent to mitigate ear pain, I thought the VS mode might have been selected at 18:40.
Re ear pain…The relatively rapid depressurization circa 17:21 was probably an ear issue too. I believe the plane was probably repressurized circa 18:23, when the Left Main AC bus was repowered, and the altitude was probably still closer to 40K ft. Of course, these are still “working assumptions” until Thales or Honeywell, or Inmarsat confirm the SD-600 Update rate and Step Size. In addition, we need to understand the algorithm (firmware filter) that takes as input the raw nav data updates and drives the Doppler Synth.
@DrB
@Victor
Adding to Victor’s comments re FLCH SPD mode, the autothrottle advances or retards the thrust levers to provide 500 FPM vertical speed for each 1000 feet altitude change, and the AFDS attempts to reach the MCP selected altitude within two minutes if able. Otherwise, the autothrottle uses idle or climb thrust to reach the MCP selected altitude.
I agree that it’s rare to use VS mode for large altitude changes, but it’s certainly possible, particularly if the pilot wants to limit the rate of descent, for example. It’s also rare to use VS mode for a climb, because of the risk of speed decay if the pilot sets a rate of climb that is too high for the available thrust.
@TBill
The most effective way to mitigate against ear pain on descent is to use the valsalva manoeuvre, which is usually done by closing one’s mouth and pinching one’s nose shut while expelling air, as if blowing up a balloon. The increased pressure inside the airway then forces air into the Eustachian tube, equalising the pressure on either side of the eardrum. Yawning can also help equalise the pressure by opening the Eustachian tube. That’s why it’s very important not to fly with a head cold, because the Eustachian tube can become blocked by mucus, which prevents the pressure equalising and can lead to eardrum rupture.
Military pilots suffer the same problem, but they are trained how to use the valsalva technique to keep up with the pressure changes.
@Victor Iannello,
@Andrew:
Thank you for addressing my question.
If my BFO analysis is correct, at 18:40 the ROD must have been close to 1800 fpm. If so, then why would a pilot who desired such a (high) descent rate ever use VS rather than FLCH? There would seem to be no significant advantage in this case.
The relatively benign change in the 18:40 BFO baseline, after removing the saw teeth, has always been a concern for FLCH mode. However, personal experience in commercial airliners demonstrates a wide variety of bumps and lurches during descents. Occasionally the descent is very smooth, with no discernible turbulence. Maybe a similarly stable atmosphere explains the steady BFO baseline seen at 18:40, even if using FLCH.
I am also thinking that the pilot might have been anxious to restore atmospheric pressure after laborious breathing using the supplemental oxygen mask for more than an hour. That might have motivated a high descent rate, despite ear pain, after diversion and depressurization circa 17:21.
Another motivating factor might be the pilot’s desire not to overshoot IGOGU for the south turn, while still passing under the minimum flight level for N571. You could not do both using a 500 fpm ROD after the right lateral offset was completed.
@DrB: If the descent was started at cruise altitude, a more typical FLCH vertical speed at the top of descent might be around 2,400 fpm, so setting the vertical speed to 1,800 fpm might have been less painful at the start of the descent.
Certainly, the turbulence caused by convection is much less at night than during the day, so perhaps that partially explains the steady BFO.
I never proposed a long 500 fpm descent. It doesn’t match the BFO and it would be way too shallow for an airliner. I only referenced it because a pilot in an unpressurized plane might limit a descent rate to 500 fpm as a courtesy for passengers. I know for me flying alone, my descent rate is limited more by concerns I have about shock cooling the engine.
All. Malaysia now expects OI search proposal early May. Search area apparently undecided as yet:
https://www.thestar.com.my/news/nation/2024/03/23/ocean-infinity-wants-another-crack-at-finding-flight-mh370
@ David
If Malaysia are to accept a Search, it might possibly involve finishing off anything left over or overlooked due to difficult Terrain in the original search area, however minute that may be, so they can be sure it’s not there, as a lot of work went into it previous. Then look at scanning elsewhere. Would be nice to see both @DeepSeaVision and @Ocean_Infinity and anyone else willing to Search, work together not competitively. It would be important to find Answers by working together.
@Re: Victor
Those with long memories may recall Victor has one very early flight path memo 180S w/ 100-ft/min descent for the whole flight. Good memo actually, as most all are.
@TBill: Thank you for reminding me of that paper from 2016, which can be downloaded here:
https://www.dropbox.com/s/zynbvxyv22r49ny/2016-06-25%20Possible%20MH370%20End%20Point%20North%20of%20Search%20Zone.pdf?dl=0
Using a magnetic heading of 180S, a descent of 100 fpm (the minimum possible via autopilot), and a speed of M0.84/310 KIAS, it’s estimated that the plane would cross the 7th arc at 31.5S latitude. The descent is required so that below the altitude cross-over of 31,560 ft, the TAS (and GS) continuously decrease as the flight path curves to the east due to winds and magnetic declination.
@Anyone
Do you ever think all the dark Blue areas outside the Box from the Bayesian will ever be searched. Imagine searching elsewhere and then in many years time finding the Wreckage there. I suppose the more that take part in searching the quicker of eliminating certain areas.
@Mick Gilbert
You asked “What makes you think that you can direct a satellite call to the pax cabin? Do you think it’s set up like an extension?”
I think that both the cockpit and cabin telephony units can make calls and receive calls made to the aircraft. If seat-side handsets were fitted and enabled (which they were not) then first class pax could also have made calls using their credit cards. I seem to remember that the SDU itself could handle up to five simultaneous calls.
To answer your question, I do not think that you can direct a satellite call specifically and solely to the pax cabin, which is why I speculated that MAS ops staff had made two calls simultaneously.
But I could be wrong.
@all
A major issue in defining the search area for MH370 is properly estimating the available fuel amount at 19:41 UTC, which is important for selecting feasible flight paths from 19:41 – 00:19 UTC. Attached a summary of the approach I have been developing focusing on the interval 18:30 – 19:41. This will be part of the bigger paper I’m working on, based on analyzing 20 million (TT + GC) random generated possible flight paths. Comments and suggestions are welcomed.
https://www.dropbox.com/scl/fi/k303ha1uh9llai7tdu7ka/MonteCarlo_v10.pdf?rlkey=b1hlxbtf27ecd9voumb1ail4o&dl=0
@Niels
Thank you for the paper. This is also a focus of mine. I currently make the assumption that fuel efficiency was a high priority. I want to go back to 1720-1825 period to see what the minimum fuel (worse case) use might have been. I tend to assume MRC/high altitude for the portion you are looking at. I believe it is possible Rt GEN was disengaged at IGARI (DFDR off and save fuel). I consider the goal may have been fuel remaining Arc7 (after SATCOM Off), if we had a case of intentional hiding of aircraft.
@TBill
You said: “I tend to assume MRC/high altitude for the portion you are looking at.” Do you typically combine that with a rather southerly 19:41 location then?
@flatpack, @mick Gilbert
Ground to air SATVOICE calls may be addressed and routed within the SDU by ‘called terminal ID or by ‘Q No‘. The latter is a priority mechanism.
In the case of all three ‘Call Announcement signalling units transmitted from the GES to 9M-MRO, two at 18:39 and one at 23:13, the called terminal ID was defined as ’00’ and the Q No/priority (10/0xAh) is translated as ‘Other “safety” C channel request/assignment signalling.
The implication of both these parameters, called terminal ID and Q No. is that the call would be routed to the Audio Management Unit, with a call alert within the flight compartment only.
@Don Thompson
Thanks for that.
Do you have a view as to why there were two calls made simultaneously?
@Niels
I favor the nominal 180S path at the start which vary from approx. 0 to 3N at 1941. I favor approx. 1.5N at the moment. In any case, the 180S path requires some “loiter” or delay, which I believe may have been going around Indonesia FIR but at some point approaching ISBIX for the FMT turn south. I am similar to UGIB except the report opts for descent/slowdown, but I believe the UGIB report notes that there is choice for high altitude fly around with similar fuel consumed. I also question the right offset from N571, I think it could have left offset to ANOKO. I can go either way on it, but why is MH370 wasting fuel/time to jog North? I am saying the deliberate hide case is save fuel. Also I question LRC speed to Arc7. Hitting Arc2 at 0-1N allows MRC cases. Note that the deliberate case allows maneuvers before Arc6.
@All
I recently put together a 60-sec video on YouTube to show the MH370 IGARI FlyBy as discussed recently by Mentour Pilot…previously @sk999 has a 2020 report and 2023 YouTube video, I believe Victor covered in 2019. Apparently at the instant IGARI was reached, and BITOD was the next waypoint, is when the Xponder was turned off, incompletely for a moment. This video is just supplemental and it attempts to show exactly how the pilot would experience waypoint changing at IGARI.
https://youtu.be/GeO6eEMdAdk
(caution always rely on real MH370 data for exactness, the PMDG777 simulator is just a reasonable facsimile but does not cut corners exactly the same etc.)
@Niels,
Your Comment#36390 of March 26, 2024 at 3:02 am
In the attachment you wrote:
“Fuel amounts saved at descent and
consumed extra to ascend back are assumed to more or less cancel each other.”
This seems a bit like getting something for nothing.
The assumption would result in a slightly relaxed constraint on fuel available for the remaining flight.
@Niels
I think you will find that there is a significant net loss in both time and fuel for a given distance in a descent / climb scenario for MH370.
The only way I can conceive to make it ‘equalish’ would be to have an aircraft that had a low idle descent rate and a very low idle descent fuel flow rate as well, coupled with a very rapid climb rate at a moderate climb fuel flow rate (both relative to maintaining level cruise flight values).
This requirement is more consistent with a powered glider than an airliner.
The U-2 might just manage it, but I doubt any airliner would even come close.
(@TBill)
Perhaps you could simulate a descent from say FL350 to FL200 and then an immediate climb back up to FL350 (recording the details of ROD, time, distance and fuel for descent, and ROC, time, distance and fuel for ascent), then compare the results to maintaining FL350 cruise for the same distance ?
The figures would be very interesting. I have a BOE gut feeling the fuel loss could be more than 50 percent, and the time/distance made good loss could be more than 10 percent.
@GeorgeG, Ventus45
It is a good point to consider more in detail. My approach was to find an upper limit to available fuel and rather be a bit generous (not to cut out paths that could potentially be relevant) given also possible error margins in the overall fuel consumption modelling. It is interesting to compare my table 2 with table D.1 of UGIB to have some reference.
The biggest issue(s) are actually for those lat/lon points where I have extreme low FLs (like 100). For most of those there is a considerable “overshoot” as the 18:30 and 19:41 positions are too close -> the straight line connection is really very hypothetical there. Also, for holding configuration, the fuel consumption per unit time is lower at FL200, so I’m considering to apply that as a FL lower limit for all 19:41 positions (as I was trying to estimate the most fuel efficient way to connect 18:30 to 19:41).
Note that the descent/climb considerations are less of an issue towards the southerly 19:41 positions, which are most relevant for the southerly cut-off of 7th arc positions.
@TBill
MRC paths at around 180 deg TT occur in my calculations, however not in high numbers / probability, please see link. I mainly see LRC and ECON modes, some M082 and M084, and MRC at low intensity.
https://www.dropbox.com/scl/fi/1tvkfpubmnrr3jjw5ausi/Thrust_Modes.pdf?rlkey=l4q1fou6rg9yut4iqc0xetrcm&dl=0
@Niels
Are you assuming no pilot maneuvers to Arc7?
@Ventus
UGIB report compares fuel for different Arc1 to Arc2 scenarios, with the rule-of-thumb that flying more distance at high altitude comes out about the same as descent/ascent, assuming the same Arc2 end-point (approx 3N for UGIB). However, what I would say is that it is possible to get further south with the same fuel, say down to 1N. UGIB is taking a bit extra loiter as needed to allow hitting Arc7 at LRC speed. My pilot is looking to end of flight, and wants to get as far South on Arc2 as possible, consistent however with apparently not wanting to violate Indonesian airspace (until Indo cannot see MH370 anymore).
All the creativity UGIB puts into Arc1 to Arc2 (one engine etc), I see as not coming unto play until after Arc5. I see Arc5 to Arc6 the way everyone else sees Arc1 to Arc2 (many options).
@ventus45
RE: ”The figures would be very interesting. I have a BOE gut feeling the fuel loss could be more than 50 percent, and the time/distance made good loss could be more than 10 percent.”
The -200ER performance manual shows the following for a descent from FL350 to FL200 followed by immediate climb back to FL350 vs continuous cruise at FL350:
Descent (210 T, M0.84/310 KIAS):
Dist. = 49 NM
Time = 7 min
Fuel = 110 kg
Climb (210 T, 310 KIAS/M0.84, ISA+10°C)
Dist. = 65 NM
Time = 9 min
Fuel = 1750 kg
Cruise (210 T, M0.84, ISA+10°C)
Speed = 494 KTAS
Dist = 114 NM
Time = 14 min
Fuel flow = 6750 kg/hr
Fuel = 1575 kg
The difference between the descent/climb figures and continuous cruise is 285 kg (18%) and 2 min (14%).
@Andrew: Using your numbers, the descent/climb case only uses about 3% more fuel over the 16 minutes required compared to the cruise case for the same duration. So if there was a deviation from straight flight over those 16 minutes, i.e., the cruise case took a longer path to traverse the same distance as the descent/climb case, the fuel consumed between the two cases could be fairly close.
@Andrew
Thank you for that – very precise.
I must get a new envelope !
AIN has an interesting article on GNSS (GPS) jamming and spoofing, which appears to be increasing in frequency.
https://www.ainonline.com/aviation-news/air-transport/2024-03-04/gnss-jamming-and-spoofing-events-present-growing-danger
@All
I have been busy checking my flight path generation procedures more extensively. Previously I had for example compared my results for LRC, CTT180 deg with the UGIB main candidate path and found a good match. In the past I had compared some different track angles with results generated by Richard Godfrey while still using excel based calculations, also with good match. Using those old excel sheets as a reference, all together I had good trust in my LRC, CTT results. I now tried to verify LRC+LNAV and ECON mode flight paths as well as, which is not that easy as there are no flight path details shared by others to compare with. The best I could think of for the moment was trying to match some entries of table 8 in the paper “Blowin’ in the Wind” (Richard, Bobby, Victor 2019). For all but one of cases considered, I could reproduce the regions of interest. I found that there was a strange deviation in my ECON Mach numbers, which in the end I could trace back and fix. Mach numbers would typically be too high, resulting in a southerly shift of some 7th arc latitudes. The consequence: I had to recalculate about 3 million flight paths, which I completed in the past week. Through the link I share some of the new results. Note that flight levels are between 310 (index 0) to 410 (index 10).
The good news is that results (focusing on fig. 5) imo seem to match a bit better with latitude distributions found by others before. Details of procedures etc. and discussion of results will be part of the paper I started writing. The bad news is that when comparing with S33-S36 as focus area, my results now indicate that this area might be a bit narrow both at the southern and northern ends, at least when considering SatCom data + “soft” fuel limit only.
As a final step in the calculations, and based on previous work by @sk999, I have generated 50 different sets of BTO and BFO data by adding random errors to the measured data (uniformly distributed within 1 sigma for all BTO and BFO data points). And will now repeat the complete statistical analysis for these 50 different data sets. This will typically take two/three days in calculation time.
The 7th arc latitude distribution is expected to widen when considering different data sets for BTO and BFO generated through depositing random errors on the measured values.
https://www.dropbox.com/scl/fi/pa1f6bb968atjtrzfni2r/update.pdf?rlkey=e6kkat2hi8g32nx7l07xet4ah&dl=0
@TBill
Indeed, this is all based assuming no pilot inputs after at least arc3. I omitted the 19:41 BFO value from my analysis as there has been quite some discussion in the past if a climb was still ongoing. I omitted the 00:19 BTO and BFO values from my analysis because of assumed prior engine flame-out. To estimate 7th arc positions, in the numerical integration for after arc 6 I extended the preceding autopilot settings.
@Niels
I assume Arc3 to Arc5 can be modeled maneuver-free, but Arc2 to Arc5 looks mostly constant allowing for BFO allowance for a maneuver resolving at Arc2. I was responding to your earlier comment about MRC, I believe MRC probably works well until Arc5, but then Arc6 is reached too soon unless there were slow down/maneuvers, which is my belief.
@TBill
I’m performing some runs where I remove the arc 6 BTO and BFO and the 23:11 phone call BFO from the statistics. Indeed MRC looks more prominent. I’m plotting arc5 latitude probability distribution, and path probability weighted thrust mode occurrence vs. arc5 latitude. To compare I also plot arc 5 latitude probability distribution and the thrust mode occurrence vs. arc 5 latitude for the original statistics. Plots follow tomorrow; it’s getting late here.
@TBill
The path probability weighted flight level occurrence suggests especially higher flight levels, so in the 380 – 400 range, when I filter on MRC mode only. Is that what you expected? Lat index 0 -> S15, Lat index 30 -> S30, FL index 0 -> 310, FL index 10 -> 410.
https://www.dropbox.com/scl/fi/nuprieleuyyzr6kjoznq8/fl_arc5_arc5stat_10_-8_newprob_MRC_68221.tif?rlkey=zk7s6gzlh3ck5gyeogdeo244m&dl=0
I have to double check before sharing more results. Did you ever look at M084 towards arc5?
@Niels
…been eclipse chasing…get back to you in a few. We had clear skies in Arkansas…TX and NY had some cloud issues.
@TBill: In Roanoke, Virginia, the eclipse was 88%. Near the peak, I was able to capture a short video during a hole in the clouds.
https://twitter.com/RadiantPhysics/status/1777420870203408814
@TBill, VictorI
Nice! I remember seeing the 1999 eclipse from somewhere high up in the French Alps. Interesting experience.
@all
Digital Malaysian Military radar data retrieved.
For your info, it took us some time to verify the data and here we are with a completed detailed analysis of the unpublished radar data coming from the Malaysian Military tracker. Malaysian acting Minister of Transport committed to publish this data in one of the press conference in March 2014. Here it is via an indirect channel.
This original data recently discovered by us is the only one known so far in numerical format with geodetic coordinates of the points on the track from Kuala Lumpur until 18h22 UTC. Altitude information is provided until reaching IGARI only. No time stamps are provided, thus estimation of instantaneous velocities would be pure speculation 🙂
The important conclusions confirm our analysis on the reconstructed trajectory (reported at http://www.mh370-caption.net). They are:
1- the aircraft was manually piloted from IGARI until Pulau Perak Island (at least). Noticeably, the manual U-Turn infered and simulated by us in 2022 at a banking up tu 38° is perfectly matching the numerical data.
2- an approximate vertical profile could be determined thanks to the cross-analysis of Malaysian radars coverage leading to a better knowledge of this segment of the flight. In particular, an emergency descent would have been simulated confirming the depressurisation of the airplane as hypothesised in our previous report.
All this consolidates our own conclusions made in our previous report.
Some of you in the past had concurring conclusions (Mike Exner on manual piloting for example, ATSB in one their early report for the FL etc.)
The report, data and an accompanying video are available at https://www.mh370-caption.net
Have a good reading 🙂
JLuc Marchand
@JLuc Marchand
https://www.mh370-caption.net/wp-content/uploads/Digital-radar-data-KLumpur-LSTRP-analysis.pdf Page 91
“At 18:15:48, when the last ADS-B message was received from Emirates EK343, the situation is illustrated in Figure 103”
EK343’s last known reported Position on flight aware going up the malacca Straits was at 02:07 anything after that seems estimated. Fig 103 Doesn’t illustrate what the sentence says about last know position. Do you have other Position info specifically for this, we are not Privy to.
@JLuc Marchand:
Please provide the link to the Geoscience Australia website where you obtained the radar data.
I’ll review your full report when I return from our eclipse adventure. It was mostly cloudy in Leakey, TX, but we did get a few good looks during totality. I did look at the altitude time series. Your new altitude profile is definitely wrong. We know from the KB civil PSR data that the altitude at KB was ~40,000 feet.
@JLuc Marchand,
Thanks for your detailed analysis. You’ve shown, what to me seems a classic hypoxic flight profile. Perhaps a phugoid motion just after the IGARI turn and then a meandering autopilot ‘off’ flight path. All seems to point to a hypoxic ghost flight with no one at the controls.
There is absolutely no reason a nefarious pilot would consider the need to take the autopilot off.
The fact that the autopilot is off screams out loud that mh370 was a mechanical failure issue.
@Tim: Despite claims to the contrary from Malaysian investigators and others, there is no definitive evidence that the autopilot was disengaged at the turnback just after IGARI. In fact, the “entry” and “exit” waypoints match a plane banked at 25 deg at FL350 and M0.81. All that is required is to start the turn before reaching the entry waypoint. You can clearly see this in this figure:
https://www.dropbox.com/s/6py60qljvxg05a6/2019-12-18%20Simulated%20Turn%20at%20IGARI.png?dl=0
There is also no evidence for hypoxia. The deviation from “straight” flight is due to radar noise, which is evident from the short time constant associated with the many of the jumps.
@JLuc Marchand: Thank you for the JSON file that was used in creating the figure shown on the Geoscience Australia website. This is helpful in creating a more accurate representation of the flight path from the military data that was previously only graphically represented. However, as you know, there are some important limitations of the data:
1. No timestamp data is available
2. The azimuth and range data for each target is not available
3. There is no altitude information
Therefore, any conclusions about altitude and speed are difficult if not impossible to determine from that data set in isolation.
On the other hand, we can get information about speed and altitude from the civilian radar data near Kota Bharu radar site. Looking at the range, azimuth, and timestamp data across the Cone of Silence in combination with the meteorological data, it is possible to determine the relationship between hypothetical altitude and speed. That relationship says that at a pressure altitude of 39,300 ft (geometric altitude around 41,500 ft), the groundspeed would be 525 knots, the true airspeed would be 500 knots, and the Mach number would be 0.87, which is Mmo. Those values are all within the capabilities of the aircraft.
By contrast, at your proposed pressure altitude of around 30,000 ft (geometric altitude of around 31,800 ft), the groundspeed would be 565 knots, the true airspeed would be 560 knots, and the Mach number would be 0.92. That speed is unrealistically too high. I really don’t believe that altitude is correct.
@JLuc Marchand, thank you very much for sharing this data. Preliminary observations:-
1.I share Victor’s view that the “air turn back” can be very well-approximated by a 180 degree turn at normal AOB if we anticipate some coasting/overshoot on the radar depiction of this turn as well as allowing for rather poor angular discrimination at this range. I’ll post separately the .kmz of my reconstruction of that turn and the speeds/AOB involved.
2. The excel file that you have linked does appear to include the “military” data and not only the primary radar from Kota Bharu and Butterworth. This is because a) we can see that the gap between those two radar records now has datapoints b) the periodicity of the datapoints beyond Penang is consistent with the rotation speed of the military radar ~10s rather than the civil primary radars ~4s.
3. There is very close agreement in track between this “military radar” and the primary radar trace that ALSM had previously shared.
4. It is pretty obvious where a datapoint is missing from the military trace. Counting the last 20 points and allowing for one obviously missing, I compute a groundspeed of 493kt. From the Butterworth radar I’d estimated about 507-509kt, declining slowly – so it is quite a bit lower. It would be useful to obtain the precise cycle period of Western Hill radar – is it exactly 10.00s?
5. The LKP position in this file corresponds very well with figure 4.1 of Bayesian Methods since the end of that track coincides conveniently with a distinguishable feature on the google earth seafloor depiction. There’s no doubt in my mind that this is the same dataset.
6. This is obviously a different dataset from that used for the “LIDO” slide insofar as that included intermittent radar returns all the way out to the LKP with exception of the lancuna around VAMPI.
@JLuc Marchand: Numerous times in the report there are claims similar to this one: “The aircraft was manually piloted as the lateral navigation characteristics of its trajectory do not match the LNAV capability of the auto-pilot.”
For some reason, you consistently ignore the more likely case that the autopilot was engaged in heading (or track) select mode, with waypoints displayed on the MAP used as a reference for navigation. After the turn around Penang Island, LNAV was engaged either by joining a leg between waypoints, e.g., FF04-VAMPI, or by joining a radial inbound to VAMPI. (Yes, I know that FF04 is associated with an instrument approach.)
This was described in detail in a report from 5 years ago:
https://mh370.radiantphysics.com/2019/01/12/mh370-flight-around-penang/
I see nothing in the data from which you can decisively conclude there was any hand flying. The track segments are basically straight, i.e., consistent with occasional small changes in heading, and the jumps in radar target position occur over time intervals that are too small to be physically realistic, i.e., the jumps are consistent with radar noise.
Victor Iannello wrote ‘However, as you know, there are some important limitations of the data:
1. No timestamp data is available
2. The azimuth and range data for each target is not available
3. There is no altitude information
Therefore, any conclusions about altitude and speed are difficult if not impossible to determine from that data set in isolation.‘
Precisely what I came here to comment after considering J-LM’s comment earlier, above.
What is being claimed by J-LM as ‘Digital Malaysian Military radar data retrieved‘ looks to be merely a JSON representation of the KML data that described the track depicted in Google Earth in Fig 2.1 of DSTG’s Bayesian Methods in the Search for MH370. The JSON data scraped from Geoscience Australia provides nothing new or credible whatsoever.
@Jean-Luc Marchand @Victor
What caught my eye is the 31 deg bank circle in Fig 51.
Recently someplace I read that 30 deg bank was pre-programmed in a B777 for some holding patterns. I do not know if the the U-Turn could be explained by entering/exiting a tight holding pattern.
Maybe Andrew can comment, but it seems like holding pattern banks are pre-programmed and not impacted by the Bank Angle setting (max 25 deg).
@TBill
Yes, the MCP-selected bank limit is ignored whenever LNAV is engaged. The bank limit in that case is determined by the FMC.
The commanded bank angle in a holding pattern is variable, depending on aircraft speed and wind. The FMC computes a bank angle to keep the aircraft within ‘protected’ airspace, and the AFDS then tracks the holding pattern using a maximum bank angle of 30°. The maximum bank angle is typically commanded at higher altitudes.
The “new” radar data is simply extracted from the ArcGIS map provided years ago by Geoscience Australia.
It’s an impressive collection of work.
To get the radar JSON data, start at the MH370 Data Release – Basemap page and look for Radar Polylines.
Following the links gets to an SQL Query page.
For Object Ids enter: 2,3
(2 is Air Defense Radar Path, 3 is Flight Path)
Also click on ReturnZ=True and change the format to JSON.
Click GET et voila
For anyone scratching their heads (like me!):
The Radar Polylines page mentioned in @370Location’s comment above is reached by selecting the “Flight Path” link in the “Layers” section of the MH370 Data Release – Basemap page.
The SQL Query page is then reached by opening the URL on the right hand side of the Radar Polylines page (https://services1.arcgis.com/wfNKYeHsOyaFyPw3/arcgis/rest/services/RADAR_Polylines/FeatureServer/1) and selecting “Query” under “Supported Operations” at the bottom of the page.
@All
So far I see, comments are mainly on the executive summary and not on the detailed analysis. So the substantive content seems missed and distracted by these light-hearted comments.
As a preamble, remember that this data does NOT come from radar sensors but is the result of radar tracker processing. Thus the track is made up of calculated points and not echoes.
If you read the report, you will see that the data includes points produced out of military data and probably also of some civilian approach radars, as the Minister said in March 2014.
A few words nevertheless to help you arrive at the heart of the analysis which includes 3 objectives (without immediately going to “calculation” as some would be tempted to do):
1- To check data quality -> done successfully, this is the best digital data on the military track and it can be used numerically with confidence
2- Determine how the aircraft was controlled -> done, it was under A/P LNAV function up to IGARI, manually afterwards to Pulau Perak at least
3- Extract altitude information if possible -> done, altitude information can be extracted at certain locations thanks to a cross-checking with extra data on radar horizons
Considering the comments so far, do I feel the syndrome “not invented here, thus not worth considering…” ?
Any brick of data is worth considering… but I can’t force anybody 🙂
@Victor
– “…no timstamps, azimuth, range and no altitude…”
Yes of course and that’s what we explain all along the analysis. But for lateral navigation quality evaluation, the ground projection is fully sufficient. Talking about azimuth and range is not relevant (see preamble above)
– “…conclusion about speed not possible…” : excatly, that is why we placed several caveats in the doc about the temptation to compute instantaneous velocities. But you probably did not reach them yet at that time.
– “…conclusion about altitude not possible from this set in isolation…” : Exactly. But this is the object of the 3rd objective by using additional information computed separately. The different radar coverage horizons have been used to cross-correlate the data especially in key points luckily provided by the gaps in the track. Entry and exit of the gaps allow to infer the altitude approximately.
Again, it appears that you made this comment without having reached this section of the report.
– “…the ground speed would be 565kt ..” : where did you find this value in the report? maybe from your previous computation based on uncertain ASTERIX time stamps? We clearly show that instantaneous speeds cannot be inferred from any radar data as they all went through a processor which “timely restamped” the data (this is also true for the ASTERIX data by the way). Here we estimate that the aircraft could have possibly reached 530kt max using long time intervals… please refer to Tables 8, 9 and 10.
-“…says that a pressure altitude of 39,300ft …” : If the aircraft were at FL390, meaning a climb, there would have been no gap in the military track after the U-turn whatsoever as both Western Hill and Bukit Ibam would have got the aircraft in their line of sight. In that location, the flight was globally crossing the radar horizons towards their decreasing altitudes, thus a gap means a descent and it cannot be otherwise. In addition, in the south of Penang, the copilot’s phone was detected… thus FL300 or FL320 max.
Here again, it appears that you did not reached this section in report, yet.
– “…you ignore the case was the autopilot was engaged … VAMPI” : In the report we evaluate a lot of times the possibility of piloting via Select Heading on the MCP (and even waypoints in FMC). We show that it does not fit at all or it would mean changing the heading every minute and half or so … which is a sort of manual piloting :-). Would you seriously see an experienced pilot doing so ? The oscillations of the track are much better explained by manual piloting. From Pulau Perak to VAMPI? maybe, but the track inferred from the LIDO image points shows a different picture : a non-linear oscillating trajectory globally south of N571… the LNAV function of the autopilot would do much better than this i.e. flying exactly on the route in straight line.
– “…track segment are basically straigth…” : “basically” ? are you serious ? It is time to use number crunching and distance measurement, and simply compare the perfect linearity of the path before IGARI achieved under LNAV of the A/P and compare it with any segment after IGARI as we did … These segments are obviously not “straight”. Please Victor, would you be so kind to explain why the track is as it is ? especially if under A/P LNAV.
@Paul smithson/@TBill/@Andrew
-“…180 turn at normal AOB…” & “..31° bank angle…”: Figures 51 and 52 show that it is impossible. The illustrative circles are at fixed banking. But an aircraft does not jump to 25° or 31° at once. The examples are there to show the boundaries. In contrary, the yellow curve is a turn up to 38° that we simulated, it started from 0° and had to increase to 38° to match the track. The circles don’t match even the one at constant 31°. To achieve this turn in matching the digital data you must bank up to 38° which is only possible manually. The automation would never accept such a banking.
-” …precise cycle of Western Hill…” : see preamble, points are NOT radar echoes, so not relevant. In the report the points are timely separated in a repetitive pattern of approx 9/9/12 sec. Remember, the data unveiled here are NOT from the radar heads.
@Don Thompson
– “…JSON data … provide nothing more new …” : First, my pleasure to have provided you with this data :-), did you have before such precision in a numerical form?. For the format, the database allows you to select the one that fits you best. Second, nothing new ? this is the only numerical data on the military track. Contrary to what you think this data was at the source of the creation of the DSTG image (and not the contrary). Several reasons for this. 1) the presence of coasting at several locations indicates an outpout of a radar tracker. A simple digitisation of the DSTG image would not reveal this coasting. 2) the DSTG mentions interval of about 10sec, we found that the pattern is 9/9/12sec actually which is in fact more realistic and probably explains why the DSTG had to reinforce their Kalman Filter as they did not notice it. 3) This is the closest data one could get on the Military data and several indication being them in the data or one other document give hints on their genuine origin.
@370 Location
– “click GET et voilà” : Bingo ! well done 🙂
And let’s give credits to trise5631 who pointed me to this database !
Remark: this is the only post I will produce on the analysis as the provided report is fully self-understandable and the reasoning could be redone by anybody who reads it completely 🙂
@JLuc Marchand: We have so many disagreements that I won’t try to reach consensus. And it’s not because I have not read the report, nor is it because the data “was not invented here”, as you boldly and incorrectly claim. I’ll provide a short list:
1. You try to determine altitude by estimating the radar range of Western Hill and Bukit Ibam. However, we don’t know the radar timing gate for Western Hill (which would limit the range), we don’t know if Bukit Ibam was operating, we don’t know the refraction index, and we don’t know what altitude was assumed when the range/azimuth data was converted to latitude/longitude. Yet, we have the timestamp, range, and azimuth for the civilian radar data from Kota Bharu. (Don has extracted the times as stamped by the radar head, which should be accurate.) Looking at the KB radar data near the cone of silence, we can calculate the groundspeed as a function of assumed geometric altitude, and using the detailed radiosonde data, we can infer the true airspeed, Mach number, and indicated airspeed. At FL393, the Mach number would be 0.87. At FL300, the Mach number would be 0.92. Unless the civilian data is wrong, the plane did not fly at FL300 near Kota Bharu, which means your assumptions about radar range and operational sites are wrong. Also, for all we know, some of the jumps in position seen in the military data could be due to jumps in the assumed altitude, as we KNOW that military altitude was recorded, and we KNOW that altitude data is erratic.
2. You conclude that the plane MUST have been hand flown after IGARI. I view that as only a possibility. I’ve shown (and provided the relevant links) that prove the turn after IGARI and the turn around Penang could have been flown with the autopilot engaged. And yes, small changes in heading are consistent with the autopilot in heading select mode, where the pilot is making small adjustments to fly near whatever waypoints he is using for navigation on the MAP display. Heading select provides much more flexibility than LNAV, as changes in heading can be made quickly without fiddling with the FMS.
3. The LIDO radar image is consistent with military radar data that is noisy and unreliable. I’ll add that I talked with the NTSB representative that inspected Malaysia’s military radar and concluded that the equipment was not properly maintained and even had birds nesting in the antenna.
4. You asked, “Would you see an experienced pilot doing so?” There are as many opinions about the way a pilot would have flown MH370 as there are pilots. Flying with the autopilot engaged in high workload situations (such as during and after the turnback) is easily justified. The same for navigating in heading select mode (although I do believe the civilian radar data after Penang shows a “classic” intercept of an LNAV leg as you would see in automated flight).
@Jean-Luc Marchand
I have no philosophical problem with A/P off, I actually favor that. I am just trying to understand what are the new facts vs. interpretations. The SIR report of course already says probable manual sharp U-turn at IGARI, based on simulator studies. Some here disagree with the SIR on this interpretation.
I read the paper of JL Marchand and I agree with a lot of his method.
As I read your comments, it seems you didn’t read fully his report. It seems you have a preconceived vision of what can be done with it but you didn’t pay attention to the caveats clearly written.
I found JL very cautious with what can be assessed from these data.
His study give a new vision of the aircraft flight path over Malaysia.
Let me notice that some papers of some studies you have produced for years are not perfect and especially from aeronautical point of view.
If the JL study is not perfect, yours were not too.
All elements pointed by JL are well sourced and carefully analyzed.
You don’t agree with some points of view of JL and especially when you consider that this plane could have been manually piloted.
But you refused to admit that radar track is not sufficiently reprensentative of a LNAV track.
IG doesn’t have the monopoly of the technical study in this affair.
@TBill: Nobody disagrees that a manual turn would be required if the start and end of the turn occurred as simulated in the SIR. However, using only those two positions, and incorporating the uncertainty of those positions, an automated turn with bank set to 25 deg is certainly possible, despite the certainty that some ascribe to hand flying at the turn.
https://www.dropbox.com/s/6py60qljvxg05a6/2019-12-18%20Simulated%20Turn%20at%20IGARI.png?dl=0
@Gilles Diharce: Did you read my comments?
I can’t PROVE the plane was not manually piloted. I just believe there is not conclusive evidence it was.
And I explained why the assumptions required to extract altitude may not be correct, especially since the result is in contrast with the civilian data near Kota Bharu, where we have the range, azimuth, and timestamps (all of which are missing in the new data set). Do you believe the civilian radar data is incorrect?
And I agree that the plane was not flown in LNAV mode, except after Penang. Where have I said otherwise? Why do you and others equate autopilot with LNAV?
I get the same accusations about piloted flight after fuel exhaustion, even though I have said MANY times that I don’t dismiss that possibility. In fact, I have been recommending search areas that INCLUDE that possibility for years now. What I HAVE said is it is prudent to start the next search by completing what was not already properly searched close to the arc before going wider.
ok Victor.
Of course, extracting altitude is difficult but the method of JL can be interesting and viable in some points.
This aircraft was piloted after IGARI and after the U-turn, the track is not consistent with a LNAV, TRK or HDG mode. I am aware of the difference between autopilot and LNAV. We can have a different assertion of this portion of flight. Data are not sufficient but we can try in depth study as JL did.
After PENANG, the aircraft is still piloted and a VOR radial outbound is similar with what we observe. Of course we can’t be 100% sure. But I agree that the track is more linear in the straight of Malacca.
For the end of the flight, I agree we must include what was not properly searched but you should avoid searching inside the arc. It is less likely. Why not expanding the search area outside the 7th arc?
@Gilles Diharce: To accept that the aircraft flew at FL300 over Kota Bharu requires the civilian radar data to be wrong. I don’t think this is likely. There is a much higher probability that one or more of the assumptions in JL’s analysis are incorrect. I’ve already presented some possibilities.
After Penang, I don’t think an outbound VOR radial was flown, as there is not a match to a VPG radial or any other VOR radial. Rather, it looks like an LNAV leg was intercepted and joined. 5 years ago, I proposed the leg was FF04-VAMPI (FF04 is the final approach fix to ILS 04), but an inbound course to VAMPI could have been programmed in the FMC, and that course could have been intercepted and joined (similar to intercepting and joining an inbound course to a VOR) using the autopilot, as I described in this post: https://mh370.radiantphysics.com/2019/01/12/mh370-flight-around-penang/
You disagree that the portion of the flight path after the turnback and Penang could have been flown using the autopilot in heading select mode. I will attempt to resurrect the work I did years ago which shows this could have occurred, i.e., the path matches well to a series of segments of nearly constant track. If navigating in heading select mode using waypoints shown on the MAP, you would expect occasional small turns, as over time, the changes in atmospheric conditions would cause the aircraft’s projected track to slowly drift away from the target waypoint. Even if using track select mode, the track can only be selected in increments of 1 degree, so you would expect small corrections.
As for the proposed search area, I’m not against biasing the search area towards the outside of the 7th arc, as I agree that that side of the 7th arc is more likely.
Victor – Thank you for all your research. Can I suggest you provide an update on your conclusions as it is difficult to gain from a simple reading of the posts and the various comments. Could I also suggest you analyse the Sergio Cavaiuolo interpretation of the Inmarsat data (2014-2022) which indicates a path west towards the Maldives? His substantial reports are linked from my blog. Also, check out my video summary of what I believe happened to MH370 here: https://youtu.be/a3x4Ka_ra7w?si=DyYuXhg8SNOPa78z. Thank you. Michael, Australia.
Jean-Luc Marchand – CAPTION:
1. You state that the KB and BW civil radar data we made available was “processed data”, not raw data from the radar sensors. That was true for the original files we made public. But a year later we were able to obtain the raw radar sensor data (raw range and azimuth direct from the radar heads).
2. The raw KB and BW radar head data (shared here) allowed us to compute the altitude at Kota Bharu quite accurately. It was ~40,000 feet as reported here (and by several others): https://bit.ly/3HhSzzj. It was certainly no where near 30,000k feet. This method was far more accurate than inferring an altitude from low elevation radar coverage maps. As Victor noted, that method ignores atmospheric refraction and other weather related affects on radar signals.
3. The assertion that the altitude at Penang must have been no more that 30-32,000 feet because the co pilot’s cell phone could not have been received from an altitude of 40,000 feet is not true. As I demonstrated in this paper (https://bit.ly/2DRZjo9), an altitude of 35,000 (or 40,000) feet at Penang was consistent with the required link margins.
4. You also assert that we have not had the military data in digital form before your paper. Again, that is not true. In 2016, Steve Kent (sk999) digitized the same plot you scraped the JSON file from. His data is available here: https://bit.ly/3JfHYpw
ALSM,
The file you linked to is not a digitized version of the ATSB track (that’s in a different file) but rather my digitization and numerical integration of Fig. 4.2 from the DSTG book giving the Kalman-filtered ground track and speed, starting from the last ACARS point. Thus, it is a completely independent of the ATSB track other than that both are derived from the original military radar data. Aside from the hairpin turn and the slow turn around Penang, the two routes match pretty darn well (as I showed in a report from Nov 20, 2016, which can be found here:
https://docs.google.com/document/d/1o_Rtk4QHrnCyJ-aKZ48XO4mxcW42QGY5m2fOVokVSTY/edit?usp=sharing
)
Jean-Luc,
Nice work pulling out that JSON file. The raw data are a bit of a mess, and I concur with the choices you made for pasting them together. Regarding reconstruction of timestamps, I note that in the Royal Malaysia Police Report, Folder 4 – SKMM Analaysis, page 21, Diagram 3, as part of the analysis of the cell phone signal from Fariq Hamid, there is a map showing the path of MH370 passing south of Penang Island, and there are 4 points with timestamps labeled. I had digitized those positions some time ago, and they match points in your table but with 3 timestamps that are 1 sec later than yours and one that is 1 sec earlier.
It is clear that, contrary to many people’s expectations, primary radar is hardly a precision tracking mechanism.
sk999:
Thanks for the clarification, Steve. The important point I was trying to make is that, as you state, “…both are derived from the original military radar data.”
@JLuc Marchand. Thanks for your reply.
You said: “…precise cycle of Western Hill…” : see preamble, points are NOT radar echoes, so not relevant. In the report the points are timely separated in a repetitive pattern of approx 9/9/12 sec.
Can you tell me how you derived/estimated periodicity of 9.75s?
If these points aren’t radar echoes (or, more precisely, a graphical representation of processed discrete radar returns) what are they?
Out of interest, having applied 9.75s periodicity instead of 10s, I get a groundspeed across the peninsular from just after KB to just before crossing the coast of 528kt, which corresponds very closely to our earlier estimate of speed at the end of the KB primary radar record.
The repetitive pattern of 9/9/12 seconds every 30 seconds, indicates to me, that the antenna rotation speed was most probably 20 rpm.
Thus there would have been ‘a sweep past the “target”‘ every three seconds.
This would mean that for the ‘9’ second data, ‘only one in three hits’ was ‘logged’, and for the ’12’ second data, ‘only one in four hits’ was ‘logged’.
Does anyone have an explanation for this apparent pattern ?
@ventus45,
Suppose the radar antenna rotation speed was 20 RPM, so new data are available every 3 seconds. Next suppose the data logger, which is not synchronized with the antenna rotation, sampled every 10 seconds. Then you would generally save the radar data with spacings of 9, 9, and 12 seconds every 30 seconds.
@Victor
A somewhat less obvious part of the debate relates to power settings at IGARI. Assuming deliberate diversion, the pilot had a minimum choice of LEFT BUS isolation to depower SATCOM, which seems (see further above) to have relatively little impact on aircraft. Alternately I believe Blelly/Marchand propose a much more extreme approach of all power turned off (LEFT BUS, LEFT XFER BUS, RT BUS, RT XFER BUS). In the latter case, RAT deploys, A/P is not available, and also believe DFDR/CVR are depowered.
Mentour Pilot has recently seemingly embraced the more drastic power-off scenario, which some (eg; JJWise) critique as hard to believe. To some extent we are hearing defense of the complete power-off scenario.
My opinion, we do not know exact power strategy the pilot may have chosen. I tend to favor an approach between the two extremes, if feasible, leaving LEFT XFer BUS powered will allow AutoPilot operation and RAT would not deploy. CVR would operate but could be erased later. Believe if it was me, I’d go ahead and disengage RT GEN totally if I could, as well as RT XFER bus to depower DFDR and allow improved fuel efficiency. Presumably bleed air is off at this point, further increasing engine thrust.
Under the more unorthodox power scenarios, the normal operating envelop is presumably expanded due to increased thrust. Estimating fuel consumption debit vs. normal CI=52 (~MRC) is difficult.
@Michael Organ asked: Could I also suggest you analyse the Sergio Cavaiuolo interpretation of the Inmarsat data (2014-2022) which indicates a path west towards the Maldives?
Sergio uses network timestamps recorded in the database to calculate the round-trip time (and distance) from the GES to the satellite to the aircraft and then back. However, those timestamps have neither the accuracy nor the resolution to precisely compute the round-trip distance. On the other hand, the BTO methodology has been shown (using known positions) to have a round-trip accuracy (1-sigma) of 29 µs.
If Sergio wants to prove his method works for MH370, he can apply his methodology to MH371, for which the position and timestamps are known. He’s never supplied this analysis despite the request. On the other hand, the range calculations using BTO have been demonstrated on MH371 and many other flights.
@TBill: We can be quite sure that the left main bus was depowered. To extend that to depowering the right main bus and the left and right transfer busses is a guess supported by no evidence but justified by what some claim “a pilot would do”. Good luck trying to reach consensus. It certainly makes zero sense to me that a pilot flying an FBW aircraft wishing to complete a mission would deliberately disable all forms of electrical power generation except for a small propeller spinning in the slipstream. If the claim is it would increase the thrust, what would that gain you? The speed would still be limited by aerodynamic considerations (a small amount greater than Vmo or Mmo, whichever limits the speed). A climb could be completed faster at higher thrust and perhaps could fly a bit higher, but the claim is the aircraft descended from FL350 to FL300, and the speed is not limited by thrust at lower altitudes.
@airlandseaman
Mike, regarding your point 3 above about the FO’s cell phone registration, something that was pointed out to me by a telecommunications engineer who had worked for Celcom Malaysia, was that the BBFARLIM2 antenna is tilted upwards by a few degrees in elevation to provide coverage of the rising terrain to the south of station. This is unusual for an urban area where down-tilt is routinely used to control the coverage footprint.
Most attempts at determining the likelihood of connection/registration do not account for the tilt-up.
A friend has just referred me to this interview. Oh dear . . a collection of thoroughly debunked “evidence” woven into his own conspiracy theory.
US based observers here may already be aware. Victor and Mike will certainly be aware, having been quoted.
More misinformation, more conspiracies, No understanding of stuff that is real.
https://rumble.com/v4paq53-the-truth-about-flight-mh370-decoding-a-decade-of-deception-redacted-with-c.html?utm_source=newsletter&utm_medium=email&utm_campaign=Redacted%20News
@Brian Anderson: I stopped paying attention long ago.
@DrB
Thankyou, that makes sense.
So does that imply that the only way to get an overall ‘smoothed’ trajectory would be a four step process:
(a) start with one point, then create 30 second segments, and average them.
(b) then increment by one data point, and then create new 30 second segments, and average them.
(c) then increment by one more data point, and then create new 30 second segments, and average them.
(d) then finally, take the newly created interlaced segments, and average them ?
@ventus45
@DrB
RE: “The repetitive pattern of 9/9/12 seconds every 30 seconds, indicates to me, that the antenna rotation speed was most probably 20 rpm.”
Isn’t 20 rpm a bit fast for a long-range surveillance radar?
@Andrew
It is high, and that is the core of the problem that needs to be resolved. It directly implies that the ‘data set’ is a compilation of hits from multiple radar heads (long thought to be the case anyway) that were simultaneously tracking the target some of the time, and perhaps only one radar head at different times. For the Penang turn and beyond, there are definitely two candidates, Butterworth and Western Hill, but we ‘supposedly’ only have Butterworth data, and it is definitely NOT a 20 rpm system.
@ventus 45,
You said: “So does that imply that the only way to get an overall ‘smoothed’ trajectory would be a four step process . . . .”
The 10 second updates of the logged radar data are usually adequate for identification, tracking, and intercept planning purposes. I am not sure what purpose you have in mind for a “smoothed” trajectory. If it is to estimate ground speed, then you won’t be able to do that precisely unless you have the data collection time stamps for each logged location estimate. If you only have the 10-second resolution time stamp of the data logger, then you will be limited to +/- 5% peak error in speed precision using locations which are 100 seconds apart. To get to +/- 1% in average speed peak error, you need logged locations which are logged 500 seconds apart. If the logged data included data collection time accurate to 1 second, then you could reduce the time intervals required to estimate ground speed by 10X.
@DrB
What I am trying to do, is ‘nail’ the GS between each of the data points for the segment post the Penang Turn, after it straightens up heading for Vampi, and in particular, from there to the last Butterworth hit near Pulau Perak.
@Victor
@JLuc
Re: Holding Pattern to achieve IGARI Turn
I have done some FSX PMDG777 cases for UTurn, and it seems to verify a tight Left turn via HOLD at 30deg bank is quite easy to initiate with the AutoPilot. I assume the pilot has a number of AutoPilot choices: accept VNAV at CI52 which slows down to about 240KIAS (which will give some radius reduction). Or possibly stop the VNAV and steeply ascend. Optional manual flight period at end of turn, to exit the Hold.
Background- a few weeks ago, I started looking into HOLD patterns to describe end-of-flight. Believe there is an argument that MH370 could have entered the water in a descending circle banking right-wing-down. From that study, I learned that B777 HOLD pattern allows 30 deg Bank, I realized this might help explain IGARI turn as well.
Hi,
Long time….
In the past month, since the announcement that OI proposed a new search based on new evidence (?), and the publication of the new study by Marchand/Bliley I have tried to refresh my memory on the subject.
I hope that you will forgive me if I have missed some important information and will correct me.
I am impressed by the detail in the M/B paper, but I find the selection of an altitude of about FL300 to be contrived and not consistent with what an experienced aviator would do. But, it has set me thinking about the profound effect that one or two assumptions has on the details of the end game.
As one would expect, I continue to test other hypotheses against the one which I cannot find a reason to abandon:
The pilot did it and was in control until the FMT
The bulk of the route was at a high altitude (~FL400) and a
maximum range Mach speed
It is likely that most of the flight before the FMT was using
automated systems.
After the FMT the flight was until about the 6th arc.
A glide descent should be considered to be at least as possible
as an uncontrolled descent.
At one point I was in a position of defending the path after the FMT as being TT, but having read all of the discussion regarding flying with the A/P, I feel more comfortable suggesting that the FMT was made with a manual turn at IGOGU to 186T to overfly ISBIX.
In the simulation that I previously published in this blog, the end point was estimated at 00:11 at 89.2, 36.6 with about 100kg of fuel remaining. (Dr.B confirmed that this path was plausible)
The hypothesis that the APU could have been started prior to the second engine fuel exhaustion/shutdown is an interesting variation on what has previously been discussed. This would seem to be a more orderly process than waiting until all engine power was lost. But it does mean that if the glide started at FL400 and not FL300, the path is significantly extended and there is no reason to presume that the azimuth did not change.(Most likely to the SE.)
The M/B discussion of the FMT focuses on avoiding possible radar detection, but any turn to the nominally 180 track to fly approximately on the boundaries of FIRs is inconsistent with the behavior during the period prior to the known last radar contact when the pilot would be certain that, at least post facto, the military radar data would be carefully studied. So, if he was not detected in real time, the other radar data (India, Indonesia…) would clearly identify the 180 path.
On the other had, a track to ISBIX would give little hint as to whether there were subsequent turns.
Why do I interject now?
The approach that we (Barry, George and I) followed the path of fitting flight dynamic calculations to the SATCOM data with an absolute minimum of assumptions. Actual computed tracks were used to determine the FMT time and exit azimuth and showed regular behavior about the single maximum probability. This contrasts with the official and most other studies which need to introduce delays and path deviations. That is not to say that these studies are not correct but seem to me to be heuristically less probable.
It would be a pity if the area near the original IG estimate were to be omitted from the search area.
I worry about that last radar contact, where the data was shown in the LIDO presentation and never subsequently acknowledged.
With my deepest appreciation to the members of the group that have stuck with the project over the years…..
Sid
@Sid, likewise I cleave to yhe notion tgat the best qualifier of terminus latitude is a “simple” path with no need to invoke loiter. In contrast to early work by IG, Dstg and others, I find that the bto optimized path fit is in the ~39S zone rather than 38S zone. I think the chief explanation for this difference is that both IG and DSTG looked for a probability peak that combined BTO and BFO residuals, and the latter biases the results northward. The secondary reason is the omission of turn dynamics, which has the effect of shifting BTO optima by about 0.2 degrees of latitude at the terminus.
@Sid Bennett
@Paul Smithson
Welcome back Sid and Paul.
It is heartening (to me) to finally see some serious reconsideration of the “deliberate, pilot controlled to the end” scenarios.
It is noteworthy that although the early DSTG/IG work was centered on 38S, (prior to the general trend to ‘drift north’) all of the ‘pilots/navigators/mission planner’ types (beginning with Captain Simon Hardy https://davidlearmount.com/2016/12/14/mh370-where-to-go-for-the-final-sweep/) have been consistent in their view that 9M-MRO lies just a little further SW.
https://twitter.com/Ventus_45/status/1774552872669340080/photo/1
https://twitter.com/Ventus_45/status/1774552872669340080/photo/2
@All
I have now written up the new discoveries as an update to my original paper. Please choose the new version IV:
http://arxiv.org/abs/1811.09315
For those who do not have time to read a 30-page scientific paper, I recommend reading the appendix on ‘The Back Swan’ first. It contains most of the important novelty, including a completely new discovery, I have not discussed in this forum yet.
https://www.gbnews.com/news/world/mh370-search-hope-ocean-infinity-malaysian-airlines-flight
@all. Some time ago in re-arranging its website the ATSB omitted access to ‘Images’. These included photos of recovered wreckage and details of some damage to the part right outer flap part will not be found elsewhere.
Though not a priority this access will be restored on the new site.
In the meantime they are available at:
https://webarchive.nla.gov.au/awa/20181011032436/https:/www.atsb.gov.au/mh370-pages/resources/images/
Oct 2018’s is the most complete page.
@All – Below taken from recent post on airline ratings is very confusing being a regular reader of this blog.
“Prof. Simon Maskell, from Liverpool University, is a scientific advisor to Ocean Infinity and was also in attendance at the meeting. Simon leads a team investigating the possibility of using WSPR to detect and track aircraft. Simon plans to add the WSPR data to the particle filter developed by the Australian Defence Science and Technology Group (DSTG) described in their book titled “Bayesian Methods in the Search for MH370” in order to refine the new MH370 search area.
The new search for MH370 is expected to start in November 2024 and the process to approve the search on a no find no fee basis will take three months.”
Interesting (acoustics).
https://www.nature.com/articles/s41598-024-60529-1
@ST: I do not understand why those promoting using WSPR data for tracking MH370 and other aircraft don’t first perform the basic experiments that might persuasively demonstrate that low power HF signals can be scattered off of aircraft over a distance of thousands of kilometers and can be detected by discriminating the scattered signal from the direct (non-scattered) signal and from the background noise. The main pieces of equipment to properly perform these experiments are an amateur-radio software defined radio (SDR), an HF antenna, and software to store and spectrally analyze the IQ data stream from the SDR.
Radar theory predicts what is claimed is physically impossible. The experimental results collected by other investigators like Nils Schiffhauer and me are consistent with the theory. Instead of performing basic experiments to explore the underlying physics, the WSPR tracking proponents continue to try to statistically prove there is some weak correlation between the data in the WSPR database and the position of aircraft distant from either the transmitter or receiver.
Unless there is some new, previously unidentified physical phenomenon in which aircraft scatter of HF signals is more easily detected for weak signals over long distances than for strong signals over short distances, there is no physical basis for their claims. Whatever statistical evidence they claim they have is only evidence that they have introduced bias (perhaps inadvertently) into their methods of selecting and analyzing data.
It is possible that OI and Malaysia will give the WSPR hotspot (wherever that is, as it continues to move, despite the heralded precision) a look since it is relatively small. That might occur before or after scanning the more sensible search area that is discussed in the article above.
If OI is inclined to search small areas outside of the main search area, I would suggest searching Ed Anderson’s spot, as that is a location along the 7th arc that is associated with an acoustic event, whether or not it was caused by MH370. In contrast with WSPR tracking of MH370, Ed derived the location of the acoustic event using known physical principles.
@ to whoever able to reply:
2 questions about sonar imagery:
(1) I am wondering how debris can be spotted against a nonuniform background such as rocky terrain like in this image ?
Here is an image showing anomalies/possible debris on a uniform background (inside the red circle), but I doubt that these anomalies could have been detected on the rocky terrain (outside the red circle).
Do you know how Fugro/OI dealt with this ?
(2) 6 years ago I was trying to collect sonar images from airliner crash sites but could only find those of Air France 447 and Egypt Air 990. Do you know any other?
I couldn’t locate it on the net, but there must be an image of the Egypt Air 804 crash site, because Deep Ocean Search wrote here:
« With the agreement of the Egyptian Civil Aviation Authority and the French Safety Investigation Authority (BEA), DOS have supplied the Australian Transport Safety Bureau (ATSB) with side scan sonar data of the MS804 crash site in order to assist with the on-going MH370 search. »
PS: sorry to those who I have not yet replied to. I was hospitalized last month. I’ll try to catch up when possible.
@Victor Iannello:
At the risk of being ridiculed, any chances that in images #2, #3 and #5 of your post
https://mh370.radiantphysics.com/2023/11/02/man-made-objects-detected-on-seabed-are-possibly-from-mh370/
E33.7875° is a typo for
E93.7875° ?
I guess chances are low that the same error is present in all 3 images AND nobody caught it, so this must be some alternate form of notation I am unfamiliar with ?
@Peter
I am thinking South African Airways Flight 295 had images
@Victor – Thank you for your inputs and detailed thoughts. Your comments are very insightful.
Hope and pray for the sake of the families that the next search is successful and time, effort and money are spent in a way that expedites results and gives closure.
@Peter Norton: You are correct–the longitude should be E93.7875°. You are not the first to find the error, and I have been delinquent in making the correction.
Entertaining!
https://x.com/JustXAshton/status/1788323068886028691
@Victor
Thanks, I needed a good laugh!
@Victor
There is a mistake in the video version Ashton quotes plane heading for Langkawi should be Penang. Believe Langkawi was Christopher Goodfellow’s early fire theory before we knew the radar data and plane was actually headed for Penang.
For the casual readers who wonder what is wrong with the crazy story, we do not have a single confirmed eyewitness for MH370 after take-off. There is no evidence of fire or mechanical problem. The evidence, we do have, is radar data, Inmarsat data to SIO somewhere, debris finds confirming SIO crash site, and I would include simulator data to SIO is *probably* evidence, that has been largely downplayed because is was disclosed too late to influence opinions hardened long ago.
I am also critical of MH370 social media sites, most of which have been started to cultivate myths and obfuscate the truth. Many of these sites are closed and we are not allowed to disclose contents, but the purpose is to infuse participants with denial and myths about MH370. I do not think we have a single social media site dedicated to nefarious pilot to the end of flight in the deepest darkest SIO, which is probably actually what happened, sorry to say.
@TBill: Ashton was corrected previously about Langkawi versus Penang, which he uses interchangeably. That’s only one of MANY errors in his work. Some of us have tried to inject reality and technical accuracy into his narrative, but it is useless. He has a strong following, and those that disagree are labeled as part of the conspiracy to hide the truth.
So rather than dispute his work, it’s much more satisfying to be entertained by it.
@ST, Victor
It pains me to see the term WSPR on this site again. You might find hints of my earlier invention of the technique in US3991418, where the details were somewhat obscured for obvious reasons at the time. I can state based on experience that the RCS of a large aircraft at HF is insufficient, by many orders of magnitude, to be detected using low power systems.
This reality seems to be easily ignored by use of a magical thinking process which was expressed at the time as FCP
(fortuitous connection of points). Whatever…
@Sid Bennett: Yes, whatever…
@VictorI:
Thanks for the mention of the Java candidate as a very specific site to search based on physical principles. I believe it’s the only candidate site that is a good match for all the hard evidence.
Besides Ocean Infinity, the Deep Sea Vision team (working on finding Amelia Earhart’s plane) has talked about doing an MH370 search. My tiny site in tropical waters would be a perfect match for their single AUV. I hope it inspires some competition to see which team will be first to take a day or two searching the Java site on the 7th Arc before taking on months expanding previous search areas.
@ventus45:
Be sure to read my review of the Nature article by Kadri on MH370 acoustics.
The bearing calculations in that report are all wrong, and he proposes physically impossible signal propagation. That leads him to false conclusions about the usefulness of the hydrophones for locating MH370. He again calls for more detailed analysis, which I’ve done.
@ST:
I’m flabbergasted that Maskell was in attendance at the OI presentation to promote WSPR for finding MH370.
RG’s methods and conclusions are pure bunk, for so many reasons. He applies the signal strength of recorded short skips but assuming long skips. Long skips beyond the antipode are extremely rare, but a long path around the globe is assumed to be available for every short path WSPR log entry.
What WSPR proponents ignore is that the D-layer of the ionosphere absorbs HF signals on the daytime side of the planet.(I became a night owl as a teen ham working DX=distant stations in the wee hours.) Only short skips are likely during the day, as signals can go up through the D-layer and reflect back from the higher E and F layers. Lower angle signals are mostly absorbed by the longer D-layer path. Maskell et al depend on most of the HF signals going the long way around the globe, which means they must be doing long skips across the daylight side. In reality, that’s extremely rare.
When I have time, I’ll try generating some stats on the frequency of WSPR spots vs the percentage of daylight along their paths.
I’ve previously suggested that it might be possible to train an AI on decades of WSPR+ADSB dataset, but the result would still be down to short skips during the MH370 timeframe. Maybe that’s what Maskell is trying.
Barnacles!:
If @TBill is listing the available evidence, don’t exclude the barnacles. It may not fit the scenario of a mastermind flying to oblivion, but like the acoustics, it’s hard evidence.
Somehow, a summary of an NYMAG article on MH370 barnacles by Jeff Wise made it to the front page of HackAday. It’s more about a “journalistic hack”. I left a detailed review on the site. There’s more in the barnacles report on my website.
The new photo evidence is that barnacle growth on the flaperon didn’t start until AFTER it beached. This is disruptive to previous drift studies, as it moves the arrival date up by 1-2 months. It is consistent with the temperature growth analysis of the barnacle shells showing that growth began at the warm end of their reproductive range, about 25C, matching with water temps at Reunion.
I followed the 87,000 CSIRO drift model particles through daily HYCOM Sea Surface Temps to see which never entered waters cooler than 25C. The southern boundary of a 7th Arc crash site would be about latitude 22S.
@Andrew
Regarding the contention advanced by Forbes that,
“The plane couldn’t land on the ground because it had too much fuel so the next option per the flight manual is to land in the water.”
you wouldn’t happen to have a copy of that particular flight manual, would you?
@ Mick G.
Can you please repeat that ?
“The plane couldn’t land on the ground because it had too much fuel so the next option per the flight manual is to land in the water.”
? That surely wasn’t about 9M-MRO after 7 or so hours in flight was it ?
An aside: “Landing” in, or on, the water has always seemed incongruous to me.
One situation where the word “alight” comes into play.
@Mick Gilbert
As a matter of fact I do. The runway at Penang has 3,354m or 11,0003 ft of landing distance available. The FPPM shows that a 772ER could land at Penang at MTOW in an emergency, assuming there are no abnormalities that affect the landing distance (a possibility that Forbes doesn’t seem to consider). On that basis, his claim the aircraft had to land in the water is horses@#t.
@George G
G’day George, that is a quote from the Ashton Forbes’ Twitter (currently known as X) post that Victor posted a link to on 9 May.
It is just one of the utterly nonsensical things raised in the post. We do need to bear in mind though that this is the portal to another dimension guy, so treat accordingly.
@370Location,
Ed, above in Comment #36482 of May 10, 2024 at 3:36 am,
you write “@ventus45: Be sure to read my review of the Nature article by Kadri on MH370 acoustics.”
I have read your review, but (quickly) forgot where.
Your link you provided in your comment does not seem to work (at least for me).
Ventus45 probably can’t use it either.
@370Location
@George G
Correct, the link does not work for me either.
@ Mick Gilbert, @Andrew
After reading Victor’s use of “Entertaining” and Andrews reply I did not bother reading further.
But after sending off my previous comment I thought I’d better see what you were all on about. You two describe it appropriately.
@Mick Gilbert: Ashton also believes that the “missing” BTO data for the satellite calls at 18:40z and 23:14z is further proof of intervention by the orbs. I unsuccessfully tried to get him to understand why no BTO data can be generated with satellite calls. He believes that because his day job is working with Excel spreadsheets, he is better able to discern anomalies in the Inmarsat data than those of us that have studied the data closely. He doesn’t have the self-awareness to see how comical his statements are, and how clueless his understanding of the Inmarsat data logs.
Unfortunately, the failed surface and subsea searches in the SIO provide ample fodder for all kinds of theories, some reasonable, and some not so reasonable. Ashton’s theory is particularly noteworthy because just about every one of his many claims is false. Again, pure entertainment.
@GeorgeG, @Ventus45:
I may have botched the link to my review of the Nature article on MH370 acoustics by Kadri. Or, shadow-banning or AdBlock happening, as I can’t see it in all browsers. Follow the original link in Ventus45’s post above, then look in the comments section at the bottom.
https://www.nature.com/articles/s41598-024-60529-1#article-comments
If it’s not visible there, then try this reference to my reviews on the Discus site:
https://disqus.com/by/370locationorg/
@370Location
Re: Barnacles
Ed, my opinion right now is that the barnacles are only secondary evidence mainly substantiate a crash in the SIO, which of course we already know.
For example, there is no support for the minority opinion that the debris was first afloat in the South China Sea, and then relocated to SIO. There is no support for planting of evidence from a Russian plot to steal the aircraft. There is no support for a specific crash location within the SIO.
Debris drift models are probably more important, but we here as a group are assigning considerable uncertainly to those models too, allowing for as far south as 38-40s as possibly consistent with debris finds.
Just to put the fuel load in perspective:
According my references for Boeing 772ER
MH370 Fuel Tanks Max Capacity: 45,200-gal = 137 tonnes (6.7-lbs/gal assumed)
MH370 As loaded at KLIA = 49.2 tonnes (tanks ~35% full)
MH370 at Penang = approx 38 tonnes (Boeing est adjusted) (tanks ~28% full)
It is perhaps surprising to realize the tanks are not very full at Penang and thus per Andrew above, close to landing allowance weight.
@TBill
The point I was trying to make in my previous comment is that in an emergency the aircraft could land at Penang even if the weight were as high as the maximum take-off weight (MTOW = 286,897 kg). The weight of 9M-MRO as it passed near Penang was clearly much less than the MTOW, so the notion that without fuel jettison it could not land and had to ditch is pure bunkum.
By my reckoning, the weight of 9M-MRO as it passed Penang was around 212,000 kg. That weight is about 3,000 kg more than the certified maximum landing weight; however, that’s somewhat irrelevant in an emergency if the aircraft needs to land ASAP. In such a situation, the pilot can land anywhere that has a runway long enough to satisfy the landing distance requirements. Penang most certainly met those requirements.
The following article explains Boeing’s view on overweight landings:
Overweight Landing? Fuel Jettison? What to Consider
@Mick Gilbert
You asked if I had the manual with Forbes’ claim “The plane couldn’t land on the ground because it had too much fuel so the next option per the flight manual is to land in the water.”
To be clear, the FCOM says no such thing. What it does say is the following:
If a smoke, fire or fumes situation becomes uncontrollable, the flight crew should consider an immediate landing. Immediate landing implies immediate diversion to a runway. However, in a severe situation, the flight crew should consider an overweight landing, a tailwind landing, an off-airport landing, or a ditching.
Note that ditching is the LAST resort, for obvious reasons.
@370Location, @GeorgeG:
I got it from the comments link, copied it, and made a pdf.
https://drive.google.com/file/d/1CwZCAJU3iW4Gddn5lY7QXXUORI-3F4-Y/view?usp=sharing
@Andrew
Thanks for that clarification.
Relatedly, I recall many years back we discussed whether Kota Bharu might have presented itself as a possibility for an emergency landing.
My recollection is that even though the aircraft would have been about 6 tonne over MLW, the runway length would have been just enough but that there would have been little margin for error. There was also the consideration of whether the KB tower was manned, and whether the ILS would have been active.
@Peter Norton: The images from previous posts should now show the correct longitude.
@Victor
Re: China Eastern 5735
…nice YouTube analysis of China Eastern 5735 situation by Chinese aviation commentator (ala Juan Browne) with perfect English. Many parallels to MH370, but interestingly, the commentator completely avoids that comparison.
Among other issues, he mentions that the USA freedom of info act does not allow disclosure of what USA knows, and international aviation rules give much power to the responsible country to control the flow of information (as in, zero flow of information).
https://www.youtube.com/watch?v=1laOmrAnaf4
@All –
With the new search in November 2024 being likely, would appreciate inputs and comments from this group on Prof Martin (Viking’s) recommended search location.
The updates to his paper are an interesting read particularly for folks like Ed who have expertise on acoustics data and contrails as well as for TBill and Victor to review from a flight simulation standpoint. There are others like Andrew and Mike who can also add valuable comments.
For sure Bobby and Victor and others might also have comments on the mathematical portion of the paper as well. I did not see any comments here as a reader and so bringing it back to the attention of the group as the recommended search location is close to Christmas Island with a very small area to focus.
“Viking says:
April 22, 2024 at 7:36 am
@All
I have now written up the new discoveries as an update to my original paper. Please choose the new version IV:
http://arxiv.org/abs/1811.09315
For those who do not have time to read a 30-page scientific paper, I recommend reading the appendix on ‘The Back Swan’ first. It contains most of the important novelty, including a completely new discovery, I have not discussed in this forum”
@TBill:
That flight 5735 analysis vid is very fishy. The perfect english may be an AI deepfake from text. Look at his identically flipped-up T-shirt left sleeve on all the videos from 6 months ago to current. The gestures are all out of sync with the dialog. I’m not familiar enough with Chinese politics to know whether the other vids are propaganda, but this one has a lot of red flags. I’d not be surprised if many of the gestures and mouth movements are repeated synthesis.
Also, I’m not surprised that you are dismissive of new barnacle and drift evidence, as it doesn’t support your scenario of a disturbed mad genius who outwits us all at every turn with his evil plan to murder the passengers and hide the plane in a deep crevasse.
@ST
Many thanks for noticing my work. Please tell me more about the expressions of interest you receive.
@Ed
re:video- could be fake I suppose, however I heard no factual/content issues, seemed to be valid points.
Re: Barnacles I am waiting for you to state what you think the barnacles show. You reacted strongly. I am not opposed to barnacle evidence, if there is anything we can conclude. I am not hearing any scientific consensus on the barnacles.
@370Location
RE: “That flight 5735 analysis vid is very fishy…
Wang Zhian is a journalist from mainland China, formerly with CCTV. He was banned by China’s government for exposing corruption and subsequently fled to Japan, where he now lives and continues reporting ‘uncensored’ news and analysis of events in China.
The “About” section on his YouTube channel states:
The commentary program on this channel is accomplished through AI technology conversion, followed by proofreading and polishing by an English editor, and then completed through voice conversion. We are grateful for the technological advancements that allow us to break through language barriers for communication and exchange.
His analysis of MU5735 fits with the known facts and previous reporting of the accident, including the details about the crew. Pilot suicide is a very touchy subject in China and public perceptions of airline safety in the country would be seriously affected if that were found to be the cause of the accident. That certainly fits with the government’s near total silence on the matter apart from two interim reports that offered no new insights.
@Mick Gilbert
Yes, as I recall Kota Bahru was doable, but KL or Penang were better options for a diversion for a bunch of reasons, including ATC, RFF and navaid availability. Kota Bahru closes at 1530 UTC (2330 MYT), with no ATC or RFF available after that time.
@ST
@Viking
Re: Critique of Viking
I always need to see one thing for a flight path, which is a summary Table with each Arc, Arcs 2-7, BTO, BFO (predicted/observed), speed, heading, altitude, location (latitude/longitude) BTO/BFO calculated by the normal methods. Helps to add the two SATPHONE calls for BFO. I usually assume I am given a path with decent BTO match, so BFO is often my focus. Secondarily, I look to see how the path handles the “apparent” offset maneuver at 1828. Also flight mode, how is this done and why.
In any case, Vikings path is very unorthodox for me, so it is hard for me to embrace. I have become in general agreement with Mike Exner’s first rule: somewhere south 30-36s. This is qualitative, but the the apparent straightness of the flight path does not necessarily mean no maneuvers (for me), but it indicates the pilot did do something to address the high winds below 22s: either LNAV/True Track or descent/slow down. I feel the pilot was active and probably did not want to violate Indonesian FIR airspace (until later say ~ISBIX where nobody would know). Seems to me an initial turn south at ANOKO off vector B466 fits the 1840 phone call BFO and FIR “fly-around” well, and up until then, the aircraft continued northwest (perhaps with an rt offset or jog left).
There was a time I was 30% on Xmas/Java but now I think much less.
Hello Victor Iannello,
I am very impressed with your work. Very nice work. I was thinking about the Malaysian government’s decision regarding the new search. Not many people familiar about WSPR, so how about proposing three locations (WSPR, Captain Belly-Jean Luc Marchand, and Independent Group-Victor Iannello) in one search?
There is a possibility that the Malaysian government will accept the proposal, especially when the media is aware of the proposal (three location in one search). Otherwise, they will be questioned by the public. Ocean Infinity has great ships and technologies. Previously, in 2018, they searched for MH370 for six months, covering 120,000km2 with one ship and eight Autonomous Underwater Vehicles (AUV). Everything will get easier with the advancement of technology.
That’s why important of three team (WSPR, Captain Belly-Jean Luc Marchand, and Independent Group-Victor Iannello) need to sit together, discuss with Ocean Infinity, cooperate together, increase possibilities. ATSB and DSTG may need to be involved. Three locations in one search (total 120,000km2), once in lifetime, for history, the world will remember about this search many years to come. Thank you.
@Andrew,
Thank you for your info on Wang Zhian.
@Andrew:
Thanks for confirming my suspicion about that vid. Even if it’s an expat reporter deep-faking himself, he has duped some into believing he is a Chinese aviation expert at the level of Blancolirio.
I’ve seen the rumors and speculation filling the void of withheld MU5735 evidence. We know little more than a leaked quote that, “the plane did what it was told to do by someone in the cockpit”. Disoriented pilots reacting to an emergency have crashed planes in the past. It seems odd to blame the seasoned trainer rather than the trainee without more concrete evidence.
@Viking, @ST:
My apologies for neglecting our discussion of the recent update on the contrails. I’ve rechecked the METEOSAT7 images I have, in various band combos with contrast enhancements. I don’t see obvious contrails. If we can get an unmodified copy of the other geosynchronous weather satellite data, I could apply similar techniques. (I’ve never tried Principal Component Analysis PCA stretching, as done to color enhance mil sat photos for the debris detection report, but it might reveal something.)
I unfortunately don’t see any acoustic evidence for the proposed impact site.
I’m currently examining hydroacoustic waves for evidence of the impact being carried outside the SOFAR channel at different apparent sound speeds and dispersions.
Incidentally, there’s a broadband event seen on seismometers peaking at 00:14 on CISI on Java, and 00:19 on XMIS. If that were MH370, it would have to have been in the water between the 6th and 7th pings. It’s going to take more digging into nearby stations to see if it’s more than coincidental local noise. A seismologist with the right tools would be able to tell more.
@TBill – Thank you for your comments on @Viking’s solution. I don’t see a table of BFO and BTO values in the updated version of the paper and so that is something the author(s) have to provide. There is detailed discussion of what values were considered and the weights given but not an exact table.
The best part of this blog is the lack of bias and focus for so many years on the science to find a solution irrespective of whether it contradicts one’s own theories or alternate science and so hoping like every other discussion here that this new updated paper by @Viking is also reviewed for possibilities or lack thereof to constrain the solution/search area.
@Ed – Thanks for the review of the contrails in METEOSAT7 images. As you do more research and maybe receive more images from the author or links to other satellite images, please keep us posted on your findings. It is interesting to learn about the seismometers peaking at 00.14 at Java and 00.19 at XMIS due to a broad band event. It is precisely the review that is needed to support or rule out the new crash site location of 13.53 South 107.11 E proposed in the paper.
@TBill, @All
The BTO errors are extremely small for my solution near CI, and small for the solution in the southern Indian Ocean. If you are happy about error-values with a small uncertainty on the last decimal, they are easy to calculate from Table 3 in my paper. For the CI-solution, we get the following BTO errors (us=microseconds):
3.0us@21:41 -2.8us@22:41 1.8us@00:11
For the SIO-solution, we get:
6.6us@21:41 -6.6us@22:41 5.3us@00:11
All values are considerably below the digital resolution (±10us), so both BTO fits are good. I guess one should use a sigma-value of roughly 7us for a Gaussian-style analysis based on the digital resolution.
The BFO errors are more complicated. My first analysis gave the Deviations (Hz) listed in Table 3. These values are wrong for two reasons. I omitted part of the systematic correction made by the SDU for the CI solution. Instead, I used the same SDU correction as Inmarsat calculated for the SIO-solution (as discussed several years ago). Secondly, I did not consider an attempted parachute jump at Bandar Aceh (possibly the reason to fly over Aceh?) followed by a temperature drop due to an open door for the CI-solution. From a purely scientific point of view, the main reason to consider a temperature drop is that it is the only way to get a perfect match with the received S/N levels at the satellite. Coincidently, these two effects give almost the same result, but with opposite sign, so the net effect is practically zero. Figure 7 shows a comparison of the two effects. To get the updated result for the CI-solution one must subtract the difference between the two curves from the initial Deviations (Hz) in Table 3 for the CI-solution. The results for the SOI-solution remain unchanged:
-10Hz@21:41 0Hz@22:41 -1Hz@00:11
For the CI-solution, we get the corrected BFO deviations (with all effects included):
-16Hz@21:41 -1Hz@22:41 1Hz@00:11
The large BFO deviation at 21:41 may have one of two reasons. Passage of a tropical thunderstorm, or the door opened 10-15 minutes before an attempted parachute jump instead of immediately before. Both are only relevant for the CI-solution.
In all cases, I have only given the values at the three relevant times where both BTO and BFO values are available, and where one can safely assume constant heading and flying height.
@Viking @ST
These days I look for very close BFO match. One of my essays is “On the Straightness of MH370 BFO vs. Flight Time Trend.” You can see the 38-South path (187s deg heading) is almost perfect BFO match to Arc5, and the 180-South is very close to Arc5. These results convinced me that these trajectories are close to actual flight, versus the initial paths I was tried to work with (to NZPG).
https://docs.google.com/document/d/1d5_ZM6lfcgjVelP6k6RjPIvsZeOgDYtJXELYHDewg3k/edit?usp=sharing
Please note that I suspect there were maneuvers after Arc5, so I am not saying these trajectories to the end, but I think probably how the flight got headed from Arc2. I have always favored 180s start (vs. 188s) despite the apparent superior BFO match (to Arc5) for the 38s options in the essay.
@TBill
I assumed no active maneuvers after 21:00 and gave only little weight to the BFO values in the optimization. In that light I think the BFO errors are acceptable.
@TBill
It looks like the contrail evidence delivers most of the explanation for the BFO error at 21:41. The airplane returned more slowly to full flying height after 21:00 than I assumed. Crudely read off the satellite pictures this will reduce the BFO error to around -10Hz.
Thanks for making me revisit this issue.
I will come back to the details of the contrails soon.
@370Location
Contrails are generally difficult to see in weather satellite images. Initially, I tried things like contrast-enhancement (as you did), but it did not work except for the Japanese satellite, where data processing had introduced artificial mixing between the reference grid and the original picture. This artificial junk partly filters out with simple image processing. For all other satellite images, contrast-enhancement actually makes contrails more difficult to see.
Instead I found that toggling between pictures taken before/after the airplane was expected to pass (or equivalent) worked best, particularly if I made wind-drift and light-level corrections (with normal clouds as reference – it is important to ignore effects from stratospheric clouds since they are moving differently and receiving more constant sunlight). This procedure is inspired by methods developed by astronomers for finding distant supernovas, variable stars, and nearby stars/planets (since nearby objects move compared to the fix-stars mostly due to the parallax effect). The last two years some astronomers have moved to AI-based methods instead (would be interesting to try on MH370 pictures).
I will mail you some files set up to toggle between relevant pictures. One can improve the results by making smart selections of wavelength range, etc. These tricks are also inspired by astronomy.
Because of the discussion with TBill, I will first mail you the Meteosat images from the area near Bandar Aceh at 22:00. The (manual) toggle is between IR and Water vapor pictures taken at the same time. The IR picture shows the change from white to black contrail left from the airplane diving below the hazy cloud cover and the ring around the fall streak hole. The WV picture exclusively shows contrail at high flying-height because the high humidity near equator prevents observations at low altitudes. To help I have inserted the coastline as a thin blue line in both pictures.
@TBill – Thank you for sharing your current thinking and the essay. I recall reading this essay when you had posted it first. As days and years go by, locating the debris is going to get harder and so hoping this year brings the search to fruition in a meaningful way.
@ST
You write: “As days and years go by, locating the debris is going to get harder”
Would you like to expand on your thoughts on this.
@Viking:
I think visuals are needed to show what you are describing here in the forum. I’ve taken the liberty of converting your .pptx file into an animated .gif image.
The map graphics get in the way a bit, so I’ve redone it from the original data with simple histogram equalization for contrast (and an experimental 4x AI “Upscayle”).
You’re toggling between two bands taken at the same time, and a feature that doesn’t seem to move is what I think you’re perceiving as a contrail heading S of Banda Aceh.
The two different wavelengths are better presented as color channels.
Then the scene can be presented over time, with more than just a toggle between two images. This is a non-AI crop-zoom over the entire time period of the flight. It starts at 1630 UTC with minimal clouds. There is a gap jump between 1900 and 2200 when METEOSAT7 loses solar power, then on every 30 minutes until 0200. So your single 2200Z frame is the first one after the middle jump. Open the .gif in an image editor to examine each frame.
What I see is that leading up to 2200 there is a weather pattern at the same angle coming from the east. After the gap, the pattern is now aligned to look like that contrail.
Treating the bands as time is tricking the brain into seeing a standing pattern among moving clouds, when the pattern is really just a common gradient on both bands.
Perhaps I’m mistaken, and you see some other feature farther south that appears like a contrail. I think this is different from your earlier reports.
I’m a big fan of the types of toggling tricks you’re talking about. It can be very interesting to take two sequential images from an image sequence and present them as a stereo pair, but it takes special viewing techniques. Almost as good as 3D, here are examples of toggling images shot from an aircraft definitely flying past Mt Fuji (not the Nicobar Islands).
@ George G –
The comment is just in line with this blog post. If the debris pieces are smaller as per this article than originally envisioned, a search in 2024 and retrieval of key pieces is best as there is still the motivation for it being a decade since original loss of flight.
There are weather and seasonal adjustments and efforts to do another search and over time just think the forces of nature might make it harder as we cannot gauge or predict these particularly with the kind of depth that needs to be explored.
@sk999
I’m trying to understand your document lstsq.pdf in detail. On page 2 you write:
“However, for hypotheses other than the true hypothesis, there will be a mismatch between the prediction c_i and the true value o_t_i , and thus the quantity o_i – c_i will include an additional contribution beyond just the error eps_i. Thus, a good strategy might be to let (o_i – c_i) be an estimate of the error and then find the hypothesis that minimizes this error by maximizing the joint probability.”
I understand the model can be wrong resulting in a model/prediction error, let’s call beta_i:
o_t_i – c_i = beta_i
And an error in the measurement, as you stated:
o_i – o_t_i = eps_i
If I add the two equations I get:
o_i – c_i = beta_i + eps_i
which seems to match with your statement above.
Now when you write: “Thus, a good strategy might be to let (o_i – c_i) be an estimate of the error and then find the hypothesis that minimizes this error by maximizing the joint probability.”,
what do you exactly mean with “the error” and “this error”; do you refer to eps_i, beta_i or the sum of the two?
Niels,
“the error” is just eps_i.
“this error” should read “this estimate of the error”
@sk999
Thank you. In this minimization process the unknown eps_i’s that we have then lead to unknown beta_i’s (unknown errors in the prediction by the hypothesis / model). Which you already mentioned and demonstrated.
An interesting read.
https://www.bloomberg.com/features/2023-deep-sea-treasure-hunter-hedge-funds/?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTcxNTkxMjU4MSwiZXhwIjoxNzE2NTE3MzgxLCJhcnRpY2xlSWQiOiJTNDc5WFNUMVVNMFcwMSIsImJjb25uZWN0SWQiOiIyRTVEMTgxNUE0MDY0RTBGQjA1QzQxNUNBNUQ0NzFDRiJ9.qAX6tzu_vzux_o7iP11mmzzcRjZRQY1tAK3w-566qh4
@ventus45
Thanks for the good read to accompany the 2 mugs of coffee I had this morning…
@Ventus45,
🙂👍
@ventus45, all
The Bloomberg story referred above was first published last November. It garnered renewed attention this, past week, as the writers were awarded a ‘Deadline Club‘ prize.
Hi all –
Sharing new article that I read on research/proposal by Cardiff University –
https://www.cardiff.ac.uk/news/view/2813809-underwater-signals-generated-by-open-sea-airplane-crashes-could-be-key-to-detecting-final-resting-place-of-mh370
Actual publication referred to is here: https://www.nature.com/articles/s41598-024-60529-1.pdf
The Cardiff News item is similar to an IndiaExpress story that got attention. Neither bother to link to the report published May 2 on the Nature website. Please read my May 5 review there.
I give more details about the acoustics in a much longer summary I made on Reddit when seeing the news story posted there.
I firmly believe acoustics will be the key to finding MH370, but Kadri’s claims are seriously flawed to the point of misinformation that has distracted from that goal.
@370Location
On your link to reddit I see you wrote comments regarding my initial findings report ‘A Feasibility Study of Kinetic and Radiated Seismic Energy Generated from a Hypothetical (MH370) Forward Fuselage Section’s Impact with the Ocean bottom (Rev 0)’.
You commented: “That makes it hard to see the conclusion of the report as anything more than an elaborate way to dismiss new acoustic evidence that doesn’t fit with prior conclusions.”
Rhetorically:
– What new acoustical evidence? My study was based on your work and seismic commentary circa March 2023? If there is newer evidence please elaborate…
– What prior conclusions?
Anyway, this statement on reddit is incorrect and offensive. Additionally, the statement is in contrast with our cordial private emails regarding your review of the draft study and the very language the published study. For clarity I’m indifferent if the study ultimately supports, or refutes your theories, and remain completely open on this subject. For the record I think you develop and offer interesting compelling theories. Independent analysis and verification should be welcomed, it helps sharpen the pencil and refocus our efforts in fruitful directions.
I’m working on a revised version of the report to specifically address your feedback on reviewing the draft of the study and made on Victor’s website. As you know, this specific analysis exercise is very complex and takes time to work out with some level of confidence.
My study was based on my understanding of your seismic energy theory postulated in March 2023. Can you clarify the energy source signals you are currently considering in your studies? Seismic? Hydroacoustic? Both? My study was focused on radiated seismic energy generated by debris impact with the seafloor, if you are more focused on hydroacoustical signals there’s probably less need for an updated study on my part.
@all:
As nobody here could point to sonar images of crash sites other than AF447 and EA990, I asked this question to aviation.stackexchange. Maybe some images will be posted there. (Or you could, if you have any.)
@Victor Iannello:
Since nobody replied to my thoughts about rocky underwater terrain I would like to put this question directly to you, since you have written several blog articles about the possibility that the crash site was missed due to challenging terrain. I assume this is of major concern to you. In the comment above, I posted sonar images of rocky terrain. I would be interested in your thoughts.
@Peter Norton: If the terrain is challenging, it is difficult to discern manmade objects from geologic formations, especially if the objects do not form a debris field, which is what investigators sought.
@Victor Iannello: This is exactly my concern. What if the debris is spread over rocky terrain like in the sonar images I posted above? How can it be found then?
There was much discussion here about the data holidays due to challenging terrain, but if I understand correctly, the OI proposal would close this data gap. I have not read any discussion on this blog though about the scenario I am describing, i.e. debris was missed not due to data holidays but due to nonuniform background (such as rocky terrain). In this case, MH370 will never be found despite a possibly ever increasing search area in the years to come. Do you know how OI plans to deal with this? Or what would be your suggestion for what should be done?
@Peter Norton: The subsea search for MH370 is better described as a search for MH370’s debris field rather than a search for manmade objects that could be part of MH370. If the debris field is not detectable, the current search strategy probably will not work. I yield to the expertise of the people that interpret sonar data for their career, while also raising the flag that the debris field might be difficult to find. This point was discussed in my last blog article.
https://mh370.radiantphysics.com/2023/11/02/man-made-objects-detected-on-seabed-are-possibly-from-mh370/
@Victor Iannello: I just read your last blog article again but didn’t find any considerations about missing the debris (field) if located on rocky/nonuniform seafloor specifically. The debris fields of both AF447 and EA990 are located on relatively uniform seafloor, which makes identification very easy. Have the sonar experts you talk to ever said if they are able to identify debris on rocky terrain? If you are in contact with them, could you ask them?
It would be interesting to know the approximate percentage of nonuniform seafloor area within the search area and what the chances are of identifying debris on such nonuniform seafloor according to sonar experts.
@Peter Norton
I responded above SA295 SIO wreck, several debris fields, was located by deep sonar. I believe I have seen the images but cannot immediately locate.
@Peter Norton,
Re the Image/s you showed/posted in your Comment #36471 of
May 8, 2024 at 4:30 pm with the identifier S5000V2_150m-2-rgb-300dpiV2-900×450 -:
These views seem very clear and distinct in comparison to many seen.
Please, for the un-initiated, would you kindly indicate the scale or area covered for the two views.
Compare the apparent clarity with views of items “seen” and reported in the ATSB Report (AE-2014-054) “operational-search-for-mh370_final_3oct2017”
For Example, refer to Figure 34 (C) of that report. This is a view of a shipwreck. Of particular concern is, or might be, apparent deposition of deposits, or assumed accumulation over a considerable number of years of deposits against the side of the vessel.
One particular concern of mine has been whether or changes over a minimal number of years, now ten, might be more dramatic in, or on, some regions of the ocean floor than others ? This may be more relevant if looking for a debris field in challenging terrain, particularly if that terrain might be subject to any “rapid” change ?
Although an extreme consideration, it becomes just another potential impediment in any search. Perhaps those knowledgeable in the field will laugh at my concerns ?
@Peter, you will, of course, notice that the surroundings of the shipwreck are of the relatively smooth seafloor type in comparison to your nonuniform background examples.
“C:\Users\Georg\Dropbox\For All\EXTRACT from ATSB Report AE-2014-054.pdf”
https://www.dropbox.com/scl/fi/qbpzvlcyu2ibt8h82d9v1/EXTRACT-from-ATSB-Report-AE-2014-054.pdf?rlkey=ojzjpbffg9hpcv19fr1pan1e1&dl=0
@GeorgeG, @PeterN:
There is a 7.4 megapixel sidecan image of the West Ridge wreck on Wikipedia.
@Kenyon:
Perhaps I did react to seeing my research getting short shrift on Reddit, and went on the offensive. It’s how I honestly felt about how your report was being used, and it is Reddit after all.
In the context of a highlighted news report on MH370 hydroacoustics, experts who know about my work avoided mentioning it over 24 posts. Regulars there rolled out dismissive prior conclusions from 2014. Your report about a bottom impact was cited to refute Kadri’s paper, when he never even considered it. AFAIK, I’m the only one who has proposed bottom impact as the source of an MH370 acoustic event.
I appreciate your invitation to discuss the draft of your report, and we did have a cordial exchange. I was therefore surprised to see it published with no acknowledgement of the mass of entrained water, which makes your energy estimate at least 10x too small.
Your paper has a premise that:
“Perhaps others can identify a scientific study or method to provide further insight into predicting Seismic Energy as it relates to the MH370 scenario case(s). Once Seismic Energy level is estimated, the viability of detection by seismometers could be studied.”
And commentary:
“Further study is required (by others) to determine if the study approach and resulting Seismic Energy Impact values are valid and if it would be detectable for a given distance by existing seismometers in the Southern Indian Ocean area in 2014.”
With a final insight:
“Predicting the seismic energy from a scenario’s given energy budget is very complex due to the various scenarios, environments, and events.”
Even though a prominent unique event was detected, you are declining to accept the evidence until “others” can validate that it’s even possible.
To me, this is not overly complex. Standard seismic analysis methods will show the event. The seabed impact scenario you modeled should be about the equivalent of three tank cars derailing at 20 mph.
In addition to the moving water mass, I also explained why you are approaching the problem from the wrong direction. Seismometers (and hydrophones) are exquisitely sensitive. They are limited mostly by the nearby environmental noise. Seismometers have tracked a flying helicopter 40 km away. A single seismometer might be disrupted, but the event appears on 45+ public geophones, which can be stacked for sensitivity. I’ve done that, and the timing arrivals match with a shallow seabed event. Over a century of cataloged events within 25 km of the Java Anomaly, all of them were at a depth of 32-90 km deep in the subduction zone. This not a typical geologic event.
There have been new developments since March, but I use “new evidence” for the public who depend on Wikipedia for facts, and that info is not current. It is also new in that the candidate was fully realized and presented after the last OI search was complete, and has never been covered in the news media.
I’m currently working on cepstral analysis of the acoustics, using both hydrophones and seismometers. For example, here’s a four hour scan using the lower 75Hz bandwidth of the Diego Garcia H08 hydrophone array.
A “kepstrum” reveals reverberation, echos, and periodic patterns. Features to be seen in these cepstrogram above is the precise timing between airgun shots for two surveys at near 8 or 10 second spacing. The strong Java event is at 01:59. A strong ice event is at 00:11. Most of the other visible noise is due to the airgun splatter. What is not seen is the arrival of an M4.1 quake in the Java Trench at 00:37. (A coicidence, about one catalogued there per week).
The “smoking gun” signal from an MH370 surface impact should arrive at around 01:03. Zooming on this plot does show a brief isolated reverb at a period of 6.5 seconds, with one hydrophone appearing weaker in the triad. This might be due to the physics of surface impacts propagating in widely spaced concentric focal zones.
By instead showing the cepstrogram of very low frequencies below 4 Hz, the M4.1 quake is revealed. It has little reverberance, compared to the Java signal which is reflecting off coastal terrain. A small echo on the quake at 7.3 seconds may be indicating the S-P wave interval. That may determine the distance from the quake where some of the energy converted into the SOFAR channel, allowing it to be detected 4,000 km away.
It’s quite clear that the Java Anomaly is completely unique from very low frequency geological events that are barely detectable on distant hydrophones. It was one of the strongest events of the day on H08.
Initial attempts using the classic cepstrum on seismic data didn’t reveal enough similarity between S and P waves to stand out. However, the above hydrophone plots reduce phase noise by using a DCT instead of FFT cross-spectrogram, and using the gradient of log(x+constant) has greatly enhanced the sensitivity.
To wrap up, please don’t take my Reddit comments personally. They are made in the environment of the self-proclaimed World’s Greatest Expert on MH370 declaring to millions of followers almost daily that there was NO detection of MH370 on hydrophones, as additional proof that it could not have landed in the SIO. That of course leaves only his theory for consideration.
I’ll try not to fall into that ego trap.
> TBill says:
> I responded above SA295 SIO wreck, several debris fields, was located by deep
> sonar. I believe I have seen the images but cannot immediately locate.
Thank you. I have also searched extensively but couldn’t find any sonar image of SA295.
> @370Location says:
> There is a 7.4 megapixel sidecan image of the West Ridge wreck on Wikipedia.
Thanks. Your link seems to be missing, but I assume this is it:
https://upload.wikimedia.org/wikipedia/commons/4/4e/A_shipwreck_discovered_in_December_2015.jpg
More details here:
https://museum.wa.gov.au/maritime-archaeology-db/sites/default/files/no-322-mh370-shipwreck-analysis.pdf
@George G:
> Re the Image/s you showed/posted in your Comment #36471 of
> May 8, 2024 at 4:30 pm with the identifier S5000V2_150m-2-rgb-300dpiV2-900×450 -:
> These views seem very clear and distinct in comparison to many seen.
> Please, for the un-initiated, would you kindly indicate the scale or area covered
> for the two views.
Unfortunately I don’t know the scale or area covered. This is the image source:
• https://www.deepoceansearch.com/2016/09/09/dos-assist-mh370-search-with-ms804-data
I am not sure what you mean by “two views”. If you refer to the image being bisected by the data gap along the centerline, this is a naturally occurring limitation of SSS. See:
• https://d2t1xqejof9utc.cloudfront.net/pictures/files/211755/original.jpg
• https://d3i71xaburhd42.cloudfront.net/fd5294690f570cf2e72722b8274cdf5086664487/2-Figure1-1.png
RE: mud
@George G:
> Of particular concern is, or might be, apparent deposition of deposits, or
> assumed accumulation over a considerable number of years of deposits against
> the side of the vessel.
If you are concerned about accumulating mud/deposits covering up (i.e. drawing a veil over) debris from MH370, a concern you voiced before, I also tried to ascertain whether debris covered by mud could be detectable, but apparently this is possible in principle:
“Sonar signals can penetrate seabed sediments so features observed in sidescan or multibeam sonar images may be buried beneath the surface.”
My comment here includes a link to the quotation source and calculations of penetration depth.
I also point to Warren Platt’s response in 2018:
> Warren Platts says:
> December 18, 2018 at 1:23 pm
>
>> @George G says that “As time goes on the likelihood of currents covering the
>> wreckage with maritime debris (mud) increases.”
>
> The rate of mud deposition out in the middle of big oceans is extremely low. Cf.
> the 19th century shipwrecks discovered by the search so far. The only way I can see
> the wreckage buried is if it got caught in a big mud slide that are themselves very
> rare events on human time scales.
further comments about “mud” here:
• https://mh370.radiantphysics.com/2023/11/02/man-made-objects-detected-on-seabed-are-possibly-from-mh370/#comment-36073
• https://mh370.radiantphysics.com/2017/10/02/atsb-releases-final-report-on-mh370/#comment-7503
@Peter
OK…what I was probably remembering, one of the MayDay on YouTube has some images (which may not be real) and verbal description of images by Rennie Van Zyl, South Africa’s head crash investigator.
@370Location said: I was therefore surprised to see it published with no acknowledgement of the mass of entrained water, which makes your energy estimate at least 10x too small.
When @Tom Kenyon shared the draft of the paper with me, I was trying to create limiting cases of the “permeability” of the water casing to determine how much the trapped water would contribute to the energy of the impact event. In retrospect, that might effect the dynamics of the structural deflections and fractures at impact, but it is probably wrong to ignore the kinetic energy of the trapped water, even if the casing that traps the water is very permeable, i.e., open. In fact, we probably also have to add the entrained water in the boundary layer around the hull in addition to the trapped water.
Interesting fact that I didn’t know: The company “Deep Ocean Search” (DOS) was working with Ocean Infinity on the seabed search for the lost submarine ARA San Juan (S-42). DOS also assisted Ocean Infinity during its search for MH370 in 2018, providing “a number of their key personnel to this operation including an offshore manager (Client Representative), a senior sonar analyst (Target Master) and 4 surveyors”.
@TBill: Thanks. Are the images you refer to in #comment-36551 sonar images or underwater photographs ?
RE: detectability of debris on rocky/nonuniform terrain
DOS says:
« The target will be not missed by our sonar, except in the rare case of it being in pieces and hidden in rocks or a crevice. » [source]
Also interesting (same source):
« Question 23: How can you be sure to avoid the “False Certainty“ trap? By this we mean “how can we be certain that any area surveyed and pronounced free of debris is in fact free? In other words, how can we be sure that the area or box coordinates are correctly positioned and are not out by any margin, which may contain the debris looked for?”
Answer: This has been identified by the ATSB as posing the biggest risk for the MH370 search. The answer lies in the knowledge that the sonar position is known at all times and can be proven to be known. This means that any system which relies on approximate sonar positioning (an estimation called ‘lay-back’) which assumes where the sonar is, or any system which does not have real-time information being relayed back from the sonar to the ship and into the ship navigation system is therefore inaccurate and no guarantee can be made as to which area coordinates have really been covered. This positional error gets worse with depth as the sonar fish position gets harder to predict and where ‘through water’ acoustic systems break down due to the range between ship and sonar and consequently the unknowns increase. The position of DOS’s sonar is known at all times, as explained above. DOS may be one of the few companies in the world which can guarantee that they will not fall into the “false certainty” trap. »
Hopefully Fugro also was “one of these few companies in the world” …
@Peter
In one of the MayDay’s for SA295 think there is both underwater debris photos and some rough sonar images, but I do not know the degree to which they are simulated images vs. real. Rennie Van Zyl’s verbal description of the sonar sounds quite similar to AF447. I am wondering if DOS was also involved in SA295 with different name at the time.
@TBill
@Peter Norton
Oceaneering International (Houston, Texas) were the contractors for the SA295 debris field mapping and recovery operation. Nearly two years earlier, under much less difficult conditions, they had performed wreckage recovery after the Challenger disaster.
You can find a couple of their SA295 wreckage photos over on the Aviation Safety Network’s page for the crash (https://aviation-safety.net/asndb/326653)
https://strumpfer.com/Papers/SA295-Search_Planning_R.pdf
http://www.withmaliceandforethought.com/Translation_of_IMT_Afrikaans_pages.pdf
@Peter Norton,
Thank you.
@Peter Norton. Maybe the next generation of AUVs will discriminate.
https://navalinstitute.com.au/discovering-wrecks-in-wa-with-new-drone/?utm_source=mailpoet&utm_medium=email&utm_source_platform=mailpoet&utm_campaign=ani-e-newsletter-
370Location:
Thanks for the feedback and context regarding your Reddit comment, no worries.
You did bring up entrained water effect in your review of the draft study but it is a complex problem that requires more time to consider and calculate the details of the physics. Privately I explained my thoughts as to why it did not appear to be significant (based on my understanding) to include in the first release and I’m working on addressing a more detailed impact of water entrained in fuselage section.
I agree there are some influences of entrained seawater but realize that (to some degree) it is already accounted for in the current analysis since energy cannot be created beyond the Potential Energy of the partial fuselage starting at the surface of the sea. In the current analysis an ‘increase’ in Kinetic Energy due to entrained water comes at the expense of terminal velocity of the fuselage. A plastic cup (with the volume capacity of three tank cars), full of warm seawater, will not be a seismic event when the reaches the bottom of the cold sea. The answer lies somewhere between your current physics concept and mine.
You state above:
“Even though a prominent unique event was detected, you are declining to accept the evidence until “others” can validate that it’s even possible.”
I agree with your statement, but for clarity (as addressed in the study) I remain open to hydro-acoustical evidence as a potential viable solution to MH370’s whereabouts. In fact, should your postulation of entrained water produce a 10x kinetic energy effect then Radiated Seismic Energy would also come into play. However, my study is focused on Radiated Seismic Energy generated from a feasible Kinetic Energy budget impact (at seabed) that is sufficient to produce a Seismic event (detected by seismometers) and measured in Earthquake Magnitudes above background geological noise. “Others” are needed to estimate the residual Seismic Energy Radiated and consider the signal travel complexities to seismometers underwater, way beyond my skill set.
If your prominent event is based on detection of hydro-acoustic signals I can set the study aside as an exercise. If the event is detection of seismic wave energy then I think we should evaluate the entrained water effect and determine of feasibility. Is your prominent event a seismic event?
Not a physics expert, but the volume and mass of the entrained water in the forward fuselage would be considerable, and if travelling at 10 m/sec, would have considerable momentum, and would likely tend to burst the probably damaged fuselage shell on impacting the sea bed, sort of like a water filled balloon bursting when dropped on concrete, would it not ?
@Kenyon:
Allow me to help with the basics of your energy audit. Potential energy doesn’t factor into the formula for terminal velocity. It doesn’t matter where the object starts falling. Kinetic energy is 1/2MV^2.
The mass in figuring terminal velocity doesn’t include the entrained water since it’s neutrally buoyant and unaffected by the pull of gravity. Just the mass of the fuselage minus its water displacement matters there.
The final kinetic energy result does include the mass of the water, which is why it is over 10x larger. I conservatively estimated using just the interior volume. I didn’t include the water following behind the fuselage, or as Victor also noted, viscously surrounding it.
Your plastic cup analogy doesn’t fit because it is near neutrally buoyant with almost no mass. It would act more like a drogue/parachute/jellyfish.
As I mentioned before, earthquake magnitudes are related to the mass volume of crust that slipped in a geological event. A mine explosion can be related to magnitude, but it’s isotropic with a relatively tiny volume, and near the surface. Such shockwave events are louder nearby and have higher frequency energy up to KHz. Earthquake catalogs often filter out frequencies above 1 HZ to remove anthropogenic noise, since quakes may have a peak energy around 0.1 Hz. Those low frequencies also travel farther to more seismometers.
It sounds like you’re negotiating that I drop pursuit of the seismics to avoid the need for determining whether a bottom impact could be detectable. Both hydroacoustics and seismology evidence are important.
A loud event was detected on all the known active hydrophones. It’s an anomaly because it doesn’t match a quake. Curtin in 2014 didn’t detect the Java Trench M4.1 Quake I pointed out on the Diego Garcia H08 array. They did report the detection of the Java event on hydrophones but dismissed it as a minor quake, even though it was below the roughly M2.5 threshold for getting cataloged, or got filtered out.
The seismology of the event is important because it pinpoints the location. I only used the nearest 8 stations to determine the epicenter, of 45 seismometers that registered the event. Rather than dropping that investigation, I’m still exploring ways to calibrate and improve the epicenter accuracy.
I hope you do pursue validating the estimated kinetic energy for a section of MH370 hitting the seabed, taking all factors into account. It may provide new insights into other surface wave propagation (Rayleigh, Love waves, even AGW) that might be expected.
There are hints of a Rayleigh wave from an impact that is strongest toward Christmas Island. Another coincidence is that the late arriving Curtin event arriving H01 Cape Leeuwin from bearing 301 is a near match (within a few minutes) for the impact event being symmetrically reflected off of the 90 East Ridge. It’s just below SOFAR depth, but there is deep ocean along the rest of the reflected path that would propagate waves at a steeper angle. It would also be consistent with an impact on a SW heading. Navy pilots know to ditch along the crest of a wave, perpendicular to the wind. (My candidate site is not dependent on a piloted ditching, but I keep testing for evidence of a surface impact.)
@All,
Assuming an aircraft is intact and floating on the water, the weight of the displaced water will include the tare weight of the aircraft plus the weight of fuel and cargo, including passengers, etc., per Archimedes’s Principle of Buoyancy. Remember that the contents of an aircraft are homogeneous, and should the intactness of the aircraft be breached, the question of permeabilty comes into play. There is no instantaneous ingress of water; it all takes time, during which buoyancy prevails. In the case of 9M-MRO, the assumption is that the aircraft entered the water at either a high rate of descent, or was being piloted to a ditching that went wrong.
On impact with the water, permeability/time is the principal component, which results in an initial high rate of deceleration, decreasing as water permeates, displacing entrapped air and compressable contents, with increasing water pressure and drag.
All I am effectively proposing, is that there is no “speeding bullet” to the bottom. Also, that the attitude of the principal remnant of the aircraft as it touches the bottom will have a profound influence on any measurable impact forces generated. The initial surface impact when transitioning from air to water (1:864) is likely to create the largest measurable force/sound capable of entering the SOFAR channel.
I seem to remember that a few years ago I equated this problem to, “How long is a ball of string?”.
@ 370Location:
The scenario’s maximum energy budget is the Potential Energy of the fuselage system near the sea surface.
Mobilizing 3 tanks trucks of warm water at the surface down to the cold seafloor requires energy. The force of gravity (density difference between seawater and fuselage system) is acting on the mass of the fuselage system. The fuselage system’s mass is virtually constant. All energy expended to mobilize the entrained seawater, displace surrounding seawater, and mobilize all trailing seawater, and overcome all drag forces, comes at the expense of velocity of the sinking fuselage system’s mass.
Based on your feedback on the study above, “It may provide insights…”, it seems that refined and revised study on Kinetic Energy (KE) might be useful to add to the arsenal of information we are all trying to extract from tidbits of data and potential evidence.
Yes perhaps I was ‘negotiating’ regarding seismology or hydroacoustics , but was generally hoping you had moved away from a fuselage generated seismic event and over to hydroacustical signal detection. My reasoning was that (unless it’s a huge difference in KE) it still wouldn’t provide a definitive value for the Radiated Seismic Energy due to the science needed to calculate seismic signal magnitude from the impact KE budget. The unknowns and complexity of the impact event (fuselage deformation, heat, sound, soil conditions, angle of impact etc.) also comes into play. Keep in mind this is a feasibility grade study, the best I can offer is to attempt to revise the study and fine tune the estimated KE at impact.
@Barry Carlson:
I agree, once the crashed fuselage becomes permeated it will begin to sink. I expect that much of the air will be compressed down in the first few hundred meters. From comparisons with other sinking objects it might take 10-15 minutes to reach 3,400m depth. Depending on impact time, that means at most 45 minutes floating. Nobody has ever suggested a “speeding bullet” (though aircraft like bullets are designed for minimal drag).
Yes, the surface impact would have had more energy, but not into the SOFAR channel. Dozens of known lightning megastrikes over deep water are undetectable on the hydrophones, even within 100km of the hydrophones. Sound from a surface impact propagates in concentric focal zones, spaced 20-80km apart. So, it may be a matter of luck having a hydrophone in the right place to catch it. Sound from a bottom impact doesn’t propagate the same way. The Java candidate appears to be close enough to the coastal slope that the noise was converted upslope into a SOFAR T-Wave, reflecting off the coastline to be heard on all available unobstructed hydrophones.
Posing a feasibility study without knowing the basic variables is bound to end up with “who knows” for an answer. Speculating the conditions could end up like counting the number of angels dancing on the head of a pin.
Still, a distinct and anomalous event was detected, right on the 7th Arc as MH370 would have been sinking. It’s very different when compared to geological events in that area.
It doesn’t make sense to just shrug it off as having some ineffable origin, when it is a good fit for all the factual evidence. The site is very specific, and can be verified using standard tools that depend on known physics.
@Kenyon:
I can’t imagine your analysis will find a way to incorporate potential energy into the terminal velocity formula and thus the kinetic energy result.
The effect of entrained seawater in the terminal velocity formula should all be within the drag coefficient for the shape of the sinking component. More entrained mass or turbulent flow outside the hull would slow it down (V squared kinetic reduction), but also increase the kinetic mass. I only did the 10x estimate based on the additional water mass inside the hull, which does not affect drag. It would accelerate more slowly toward terminal velocity, but it is essentially full of water as soon as sinking has begun.
Moving on to new findings, I mentioned the possibility of the Curtin Event being a symmetric reflection off the 90E Ridge. For an intersect at 90E with depth 1,400m on H01 array bearing 301.6, I ran the numbers on propagation time triangulation with the RCS Perth hydrophone. The actual vs expected arrival time difference error is less than 4 seconds over 4,513 seconds total propagation time (00:19:37..01:34:50Z).
That small error might be accounted for by the Perth Canyon reflection path being slightly north of the H01 path, or 2014 variations from the averaged March temps over a decade, or even the method of estimating arrival time. It’s essentially a precise triangulation.
“The NTSB has opened an investigation into an inflight oscillation event on a Southwest Airlines B-737-MAX 8.”
https://x.com/NTSB_Newsroom/status/1801671189166748119
The crew attributed the oscillation to Dutch Roll. It would seem the next step is to investigate why it was not controlled by the yaw damper.
A sobering investigation report of a serious incident that came very close to a CFIT due to an incorrect altimeter setting:
“After reaching the indicated altitude corresponding to the chosen minima, the crew carried out a go-around because they had not acquired the visual references needed to continue the landing. During the manoeuvre, the minimum recorded and corrected radio-altimeter height was 6 ft, i.e. about 2 m, when the aeroplane was about 0.9 NM from the runway threshold, outside the limits of Paris-Charles de Gaulle airport. As per design, there was no on-board ground proximity alert (TAWS) during the event. In their statements, the crew indicated that they had not been aware of this proximity with the ground.”
BEA Safety Investigation Report – Serious incident to the AIRBUS A320 registered 9H-EMU and operated by Airhub Airlines on Monday 23 May 2022 on approach to Paris – Charles de Gaulle airport
@Andrew
Wow!
@Andrew: Yes, very scary, especially since the incorrect altimeter setting was still incorrect for the second approach.
I was surprised by the statistics of how few airliners are equipped with WAAS for LPV approaches, which should provide vertical guidance about equivalent to ILS.
Scary.
If we still used QFE, a cross check between QFE and QNH against the published Aerodrome Elevation would/should trap such errors.
@ventus: On the ground, it is easy to make sure that the indicated altitude matches the airport elevation, ensuring the altimeter setting was correctly entered. However, once in the air, how would that check work?
@Victor
Yes, the airlines have been slow to install the equipment needed for LPV approaches. Many business jets and even smaller aircraft are far more advanced.
I think it comes down to economics and a lack of incentive. Cat I ILS approaches are still available at most large airports, so the airlines don’t see a need to spend money on new equipment to achieve much the same thing as existing ILS equipment.
Cat I ILS approaches are being phased out at some point, but so far I think only Europe has mandated a cut-off date, together with a requirement for LPV capability from Jun 2030. From that date, Cat II/III ILS approaches will be available in low visibility conditions, but Cat I will only be available for contingencies such as a GNSS outage.
@ventus45
I think QFE could potentially make things worse by adding another set of numbers that a crew or controller could get wrong. The crew would also need to determine the difference between the QFE and QNH, convert that to a height and then compare it to the airfield elevation, introducing more potential sources of error.
In my view, one of the big problems here was the crew’s failure to crosscheck the controller-provided QNH against another source of information, such as the ATIS or a METAR. They already had the ATIS with the correct QNH, but didn’t compare that with the QNH provided by the controller. If they had done so, the 10 hPa difference should have been a red flag and a prompt to confirm the correct QNH. At the airline where I worked, it was SOP to perform such a check, but at this airline there was no such requirement.
@Andrew: A discrepancy of 10 hPa results in height error of around 300 ft. A non-WAAS GPS should accuracy on the order of tens of feet. So, even if an aircraft has no LPV capability, it seems there should be a way to detect and warn against incorrect altimeter settings of this magnitude at low heights.
@Victor
Absolutely; such systems have been available for over 10 years, but to date they have not been widely adopted. The BEA report discusses the Honeywell CAM-BTA system and the Airbus ALTSM Step 1 and Step 2 systems at para. 1.18.3.
The report doesn’t specifically recommend the installation of such systems by airline operators. Instead, it recommends that ICAO reassess the measures available to mitigate the risk of an incorrect altimeter setting, which might include updates to ICAO’s standards and recommended practices. That sounds very much like bureaucratic buck-passing!
Knowing your correct altitude close to the ground is obviously important. But for glider pilots, having an accurate altimeter setting at 18,000 ft is also important. We routinely fly at altitudes up to the edge of Class A.
We had an incident here in Colorado a few years ago when several KBDU based pilots flying on the western side of the continental divide were observed by Denver Center at ~18500 feet. Turned out there was an unusually strong pressure gradient from the eastern side of the divide to the westen side. Even though they were only 40-50 miles from KBDU, their altimeters were all off by >500 ft. Everybody uses the Granby altimeter setting now as soon as we go across the divide.
@airlandseaman: Since you are mostly concerned about staying out of Class A airspace, couldn’t you just use the standard altimeter setting (29.92 in Hg) and stay below 18,000 ft?
Victor: Yes, that sould work as long as you remember to reset it once you are below, say, 17,000. I need to review the thread on this subject. Your idea sounds familiar. Since that incident I just try to stay below 17,500.
@Victor
@ALSM
If you were to set STD on the altimeter, you would need to allow a buffer if the QNH on either side of the divide were higher than 29.92 in. Otherwise, you could bust the lower limit of the Class A airspace, defined as 18,000 ft AMSL.
Perhaps it’s better to set the highest QNH in the area or just stay below 17,500 ft as Mike indicated?
Andrew: Couldn’t we use WASS GPS altitude to stay below 18,000 MSL?
@ALSM
Good question! I suppose the WAAS altitude should theoretically be quite accurate, but I’m not sure of the ‘legalities’ if there were a problem. ATC would certainly be relying on QNH to determine the lowest flight level they could assign to IFR aircraft inside the Class A airspace.
@ALSM
Thinking about that some more, the WAAS altitude should be accurate close to the ground, but that might not be the case at higher altitudes. In a non-standard atmosphere there could be a significant difference between the WAAS altitude and the barometric altitude. It might be better to rely on the altimeter!
@ALSM, @Andrew: How complicated simple things become!
I think the answer may depend on the goal.
1. If the goal is to maintain vertical separation from IFR traffic at FL180, then the altimeter reading should be used as referenced to a setting of 29.92 in Hg.
2. If the goal is to not receive a “possible pilot deviation” notification from ATC by entering Class A airspace, then the altimeter setting should be whatever ATC is advising to use in that area based on the setting from a nearby airport. As @Andrew says, using that altimeter setting, the indicated altitude at 18,000 ft may be substantially different from the true altitude, as the two will in general match only near the ground.
3. As an aside, the GPS altitude (WAAS or non-WAAS) is not usually indicated by installed instruments, at least not in planes I’ve flown. However, the GPS altitude is shown in EFB devices using either the device’s internal GPS, or by interfacing with an external device with GPS. For instance, in my plane, I can see the WAAS GPS altitude on my Ipad using Foreflight, which interfaces with my Sentry ADS-B receiver.
Probably the safest option is to do what Mike practices, which is to use the altimeter setting advised by ATC in that area and maintain a large vertical buffer, i.e., greater than 500 ft.
@Victor
I check in a follow the conversations from time to time on this blog and have contributed a few messages over the years. But the blog seemed to cease after your June 14 message and begin again July 30, except for one message from Andrew on July 11.
I thought maybe it ended – too bad. But it seems alive again. Good
@Hank: Thanks for checking in.
When there are significant developments, either from new evidence or new insights, it gets discussed here. Lately, there has been little of material value that has surfaced, so the activity has been light.
@Victor
@ALSM
RE: “I think the answer may depend on the goal”.
Yes, I assume the goal here is to ensure the gliders remain outside Class A airspace and avoid an FAA violation.
I think the problem arises because the regulations only say the altimeter setting must be that of a station within 100 NM of the aircraft’s position. That’s obviously an issue in cases where there is a strong pressure gradient over a shorter distance, as Mike mentioned. A glider that departed from the Boulder area using Boulder QNH would be ‘legal’ to continue using that QNH over the divide, but its altitude would be higher than indicated if the QNH over the divide is higher than the Boulder QNH.
As Victor said, the altimeter setting should be whatever ATC is advising to use in that area. Mike, if you’re not in contact with ATC when flying over the divide, do you have any contacts in Denver ATC that could tell you what QNH source they use for aircraft flying in that area? If it’s Granby, then you guys seem to be doing the right thing by setting Granby QNH.
Andrew: GPS MSL altitude is displayed on our Dynon PFD. It is accurate to about 10-20 feet at any altitude. So, if the bottom of Class A is defined as 18,000 ft msl, GPS should be the best altitude to use. But I’m not sure actually means to define it in terms of the true altitude. More likely, they mean 18,000 ft as read on an accurate altimeter set to 29.92 in.
The 100 mi rule clearly falls apart around the divide. There is almost always a few hundred feet difference crossing the divide at 18k. And the difference between the pressure sensed altitude and GPS can be up to 700-800 ft depending on the altimeter setting.
We have several AWAS stations on the western slope, including Granby. So it’s easy to adjust the setting as we fly around western CO. But staying below 17500 makes it easy. Serious cross-country pilots typically stay between 13 to 17k anyway.
@ALSM
Thanks – I worded that comment poorly when I said the WAAS altitude might not be accurate at higher altitudes. It’s the barometric altimeter that’s subject to error in a non-standard atmosphere. I meant to say the WAAS GPS altitude and the barometric altitude should be very close near the ground, but at higher altitudes there could be a significant difference due to errors in the barometric altitude.
If ATC are monitoring your altitude (presumably from a transponder in the glider?), I assume they’d be looking at the barometric altitude – ie the pressure altitude from the transponder, corrected for the local QNH by their equipment. That’s why I think it would be better to use the altimeter with an accurate local QNH, rather than the GPS altitude.
I’ll ask the guys at Denver Center. They see the modeS altitude on their display.
@airlandseaman: Here in Roanoke, the displayed altitude on the radar screens is the Mode C pressure altitude from the transponder with a correction for the altimeter setting. This means that if the aircraft has correctly set the altimeter according to notifications from ATC, the altitude appearing on the screen is identical to the altitude on the cockpit display. Also appearing on the radar screen is the (Mode A) squawk code and the call sign, which is entered into the ATC system when the squawk code is assigned. So any uncleared incursions into Class A, B, or C airspace would be determined by the Mode C altitude from the transponder as corrected by the ATC notification for altimeter setting.
If no squawk code is assigned, e.g., for VFR traffic squawking 1200, the controller can still see the (Mode C) altitude, but this would be “unconfirmed” because the altitude would not be verified as it would be typically confirmed upon initial “radar contact”. The controller can optionally use the ADS-B out information from the airplane to display the call sign and altitude, although normally, the ADS-B information is not displayed. I was also told that various ADS-B receivers are used to “fill-in” the radar data due to spotty coverage from terrain.
I imagine different ATC locations may do things differently. Roanoke approach is for a Class C airport, and its radar and display equipment might be different (more outdated) than Denver Center.
I think that the issue here is one of ‘thickness between pressure levels’ which depends on the temperature profile with altitude.
An altimeter can only measure the pressure where it is against a datum (QFE, QNH, QNN) but it can not measure the temperature of the air between the datum pressure level and the present pressure level.
See section 6 and figure 7 of https://atmos.uw.edu/~houze/301/Miscellaneous/Skew-T.pdf
And section T figure 22 (last page) of https://atmos.uw.edu/~houze/301/Miscellaneous/Skew-T.pdf
@ventus45
Yes. In somewhat more simple terms, the altimeter measures pressure and displays that pressure as an altitude, calibrated for a ‘standard’ atmosphere. If the actual temperature profile is colder than standard, the pressure levels are closer together and the aircraft’s true altitude is lower than indicated. Conversely, if the temperature profile is warmer than standard, the pressure levels are further apart and the aircraft’s true altitude is higher than indicated.
Juan Brown (@blancolirio) has a good YouTube video on the tragic crash of the ATR-72 in Sao Paolo, Brazil, that ended in a flat spin, probably due to accumulation of ice.
https://www.youtube.com/watch?v=ozinSmylTmQ
@Victor
Tragic. Similar types of aircraft have been involved in a number loss of control accidents/incidents over the years, mostly attributed to airframe icing. Flat spins are extremely dangerous in any aircraft, and recovery is sometimes impossible.
Andrew, I have a very vague recollection, of a report (I think it was in an old Flight Safety Australia Magazine back in the 70’s, perhaps early 80’s) of an F27 Focker Friendship getting into a similar state over central western NSW, which was initially unrecoverable by normal means, but was eventually recovered when the pilot (after having initially reduced power on both engines to idle), deliberately applied what was said (as I recall) to be a surprisingly small amount of torque to the engine on the inside, which broke the spin, and enabled recovery. Do you remember reading or hearing about that incident ?
@ventus45
Is this the incident? See the article “MAYDAY” on p.4:
Aviation Safety Digest Number 100, 1977
Scanned copies of all issues of the Aviation Safety Digest from 1953 to 1991 are available on the ATSB website:
Aviation Safety Digest
Former NTSB Greg Feith has a very good YouTube video re: Brazil ATR72, what will be looked into in addition to icing, and how to recover from various types of stalls. Flat spin being difficult but very rare situation for commercial flights. France would probably be invited to participate as is French-built, USA might be invited as user of the ATR72 aircraft and investigative resources.
https://www.youtube.com/watch?v=c-Lgg7ItRZw
@Andrew
Yes!
The ‘gust lock’ incident.
The memory was clouded somewhat.
Not a spin, more like an incipient spiral dive.
Thanks for finding that article / magazine (and the index page to the others). I will download them all.
@Andrew:
I think the working theory on this blog is that the SDU stopped working because the L Main AC bus was isolated.
May I ask you how plausible it would be for either one of the two pilots to have known about this method to disable the SDU from either training, logical deduction, conversations with colleagues from maintenance, online research or other ? Is it safe to assume that a pilot willing to find this information will fairly easily be able to find it ?
(This may be a very subjective question, but I would be interested in your opinion.
If you can share why it would be easy to find out, that would be helpful, too.)
Jeff Wise said in his podcast (S1E30):
« When you start pulling circuit breakers or even isolating whole busses, you don’t really know what’s going to happen. You can do a lot of research, I mean … if we are even gonna contemplate that somebody did this, they would have had to do a ton of research, because they must have known things that an ordinary triple-7 pilot wouldn’t know. I mean, I remember, right when all of this was happening back in early 2014, I asked triple-7 pilots, I said: “How would you depower the SDU?”. And they were like: “What?? What is the SDU?” »
If that answer is in any way representative, it would seem that pilots usually don’t even know the SDU exists, let alone how to disable it.
My second question based on this:
In case the perpetrator was unlikely to know about the SDU’s existence and/or ways to depower it, let’s assume he didn’t know. In this case there was no reason to isolate the L Main AC bus. Would it be a realistic scenario that he shut down the entire electrical system instead? One reason for doing so would be to make absolutely sure that no communication could occur. Or maybe the perpetrator knew he had to cut off the SATCOM link but also knew that the standard way of deselecting SATCOM via the CDU would send a logoff message, which he wanted to avoid, but he didn’t know for sure how the SATCOM is powered, so to be sure, he shut down all the power. Either way, is it possible and realistic to fly a B777 without any electrical power (a) generally and (b) specifically in the way MH370 was flown from IGARI to the moment when the SDU was repowered ?
The reason I am asking is that I am not aware of anybody having found a good explanation for why the SDU was repowered. I was thinking that if the perpetrator only isolated the L Main AC bus, he wouldn’t lose the autopilot, but by shutting down all electrical power, he would lose the A/P, right? So there we would have the explanation right there for why the SDU was repowered: the perpetrator needed the autopilot for the long flight into the SIO.
What are your thoughts about this scenario from a practical point of view ?
To me it sounds pretty far-fetched that someone would shut down the L Main AC bus (let alone the entire electrical system), because he wouldn’t have had any idea what the consequences would be, what systems would be lost, etc. no?
The B777 pilot interviewed by Jeff Wise in S1E30 said exactly that:
Juan Browne (B777 pilot): « You do not under any circumstances go messing about with these switches until you are directed to by an emergency procedure and a checklist will pop up on the ICAS screen and it will tell you exactly what to do and in what order to do ’em. And that’s the only time, you would ever mess with any of those switches. »
Jeff Wise: « If you were to do something like, say, isolate the entire left AC bus, that would cause … I mean, I don’t know if you even know what it would do? »
Juan Browne (B777 pilot): « Right, you don’t even know, what it causes. See, what was the old saying: “Somebody knows, where all the wires are. But nobody knows, where all the wires go to.” In other words, when you start messing with these systems, you’re gonna cause downstream effects, you have no idea of what those outcomes are gonna be. The [road?] of unintended consequences are huge, when you start messing with these different electrical systems. So, again, by no means do you ever mess with these electrical systems, unless you are directly told to by a checklist procedure. »
But on the other hand, from the BFO values associated with the SDU’s 2nd logon, we know that the SDU had no power beforehand. So we know that at least the L Main AC bus was disabled if not the entire electrical system, right ?
It still seems like a mystery why the SDU lost and regained power.
Or what is your best guess based on all we know ?
sorry, typo: EICAS screen
@Peter Norton: I look forward to hearing Andrew’s perspective as a retired 777 captain, but I’ll offer some points that we’ve discussed here before:
1. Reason for the shutdown of the SATCOM: Possibly to prevent the use of the SATCOM phone in the passenger cabin before passengers were incapacitated.
2. Reason for the re-start of the SATCOM: To restore the electrical system to the preferred, fully redundant state.
3. Why 777 pilots are not aware of the SDU: The SDU is a component of the SATCOM, which is well-known to the pilots. For instance, EICAS messages refer to the SATCOM, not to the SDU. If Jeff had asked pilots about the SATCOM, he would have received much different responses.
4. Pilots’ knowledge about how to disable the SATCOM: An airline pilot with access to the training manual would have the schematic showing that the SATCOM is powered by the left AC bus, and would therefore know how to disable it if researched before the flight. Also, when the left bus is isolated, messages appear on the EICAS referring to the disabled SATCOM. The PMDG 777 model for FSX, which the captain had installed on his simulator, would show these messages. In short, the capability of an airline pilot to gain information about how to disable the SATCOM and to understand the consequences of isolating the left bus is not as difficult as Jeff claims.
@Peter Norton
Question 1
Before MH370, I think it’s fair to say that most B777 pilots would not have been aware of the SDU or its power source. The FCOM does not contain any detail on the components of the SATCOM system or their power sources, and that information is not covered during training.
However, a pilot who wanted to find more detailed information on the system could easily do so by referring to the maintenance manual or maintenance training manual, or by asking a technical training instructor or engineer. The pilot would only need to determine the SDU to be the main component of the system, along with its power source. Armed with that information and existing knowledge of the electrical system, the pilot could easily isolate the appropriate bus to de-power the SDU.
To answer your question: Yes, I think it’s safe to assume that a “willing” pilot could fairly easily find the necessary information to de-power the SDU.
I expect that a pilot contemplating such an act would research the other systems that would be affected by isolating the bus. Again, a “willing” pilot could easily find that information. Remember too, that Zaharie was an experienced B777 Type Rating Examiner. He probably had far more knowledge of the aircraft systems than a regular line pilot, and he would also have had access to a full flight simulator where he could test system effects, in addition to the simulator he had set up at home.
Question 2
When you ask “would it be a realistic scenario that he shut down the entire electrical system instead?”, I assume you mean the entire AC electrical system. Such a scenario is possible, but in my view not realistic, given the number of major systems that would be lost, including the autopilot. I think that’s especially the case for a pilot operating the aircraft alone, well after midnight, though I know others (eg CaptioN) disagree. De-powering a single Main AC bus is nowhere near as big an issue, because all the essential systems would remain fully powered. A few peripheral systems would be lost, but anything important would be duplicated and powered by a different bus.
Regarding Juan Browne’s comments, I agree that a pilot would not normally “go messing about with [the overhead panel] switches” unless directed by a checklist procedure. However, I think we can all agree that MH370 was anything but normal. As I said earlier, Zaharie had tools at his disposal to research which systems would be affected if the Left Main AC bus were de-powered, so I don’t think Juan’s comments on that aspect are valid.
At this point, all we can do is make reasonable assumptions about what occurred with the SDU power. I agree with Victor’s earlier comments that the SDU was likely de-powered to prevent use of the SATCOM by cabin crew or passengers before they were incapacitated, and that power was subsequently restored to put the system back into a “normal” state.
@Peter Norton
In addition I would say-
>I assume a savvy pilot planned to make the aircraft disappear, and that what we witnessed was a highly technical plan by a master B777 pilot. Jeff Wise has no idea what the plan was. We do not have Boeing and FBI advising us on the possible nefarious elements of the flight, and there is reluctance to have that speculation in forums like this. This reluctance in the online discussion gives Jeff Wise a point of departure to say the MH370 pilot did not follow Jeff Wise’s logic. That is meaningless justification on Jeff’s part to pursue a conspiracy theory. So far I see no mistakes by the pilot except maybe Arc7 reboot. Pilot may have wanted Right Bus IDG off, and needed Left Bus back on. Also pilot was allowing incoming Sat phone calls, which tells me he may have wanted to monitor phone activity on Left Bus SATCOM, and if the pilot was as smart as I think he probably was, pilot may have assumed phone calls might reveal direction of flight. I assume SATCOMs were turned off at Arc7, for that reason, also SATCOM=OFF at Arc7 also prevents incoming SAT calls which could have revealed that the aircraft still flying after Arc7 (in the event of an incoming call).
>We do not know what power settings were chosen at IGARI. Yes it is true that a nefarious pilot may have chosen a more complete power outage. We only know that Left Bus off is the minimum requirement. Once you start talking about more than the absolute minimum nefarious actions to explain what we saw, that gets into the realm of nefarious active pilot to the end, which is a taboo topic avoided by many. Those willing to consider the nefarious case do not have full modeling/tech support in the volunteer online blogosphere needed to properly address that case.
@TBill said: Those willing to consider the nefarious case do not have full modeling/tech support in the volunteer online blogosphere needed to properly address that case.
There is no reluctance to consider the case in which there were multiple pilot inputs to the end. However, it becomes impossible to develop a hot spot for the search for these cases without making assertions that are just guesses. So, the approach that many of us recommend is to first search the hot spot(s), and if unsuccessful, expand the search, recognizing that the assumptions required for the hot spot might not be true.
@Peter Norton,
As someone who believes that the case for the accident scenario has yet to be ruled out, I believe this sequence of events can fit with the SDU power loss and 18:25 power up.
Damage to the left side of the avionics bay by rupturing O2 bottles causes ELMS to loadshed all loads off the L AC BUS(maintenance manual says it will do this for a ‘loss of load level data or no power for 2.5 seconds’).
The power up at 18:25z may have been when ELMS control over the L AC BUS failed, due to the main battery going flat(the battery is one source of power to ELMS-the other power sources already damaged). So the load shed is now removed and power is restored to the L AC BUS and SDU.
Although complicated, we also have to consider a damaged electrical system might not behave in a predictable way.
@Tim: Why would the battery fully discharge if the right bus is working?
@Victor,
The battery could discharge because the battery charger is damaged, or cables to/from it have been severed. The battery charger is in the E3 rack near the O2 bottles, left side of avionics bay.
@Tim: Thanks for your comment. How do you explain
– the transponder being switched off manually,
– the flight path, particularly past Penang,
– and the SDU regaining power ?
@Tim:
And how do you explain that ACARS was manually turned off ?
In a catastrophic accident scenario, that’s the last thing a pilot would do and nowhere on any checklist.
Mike Exner wrote: « When the AES reconnected with the Perth GES at 18:25, there was no ACARS log on. That means that the PF turned of all ACARS from the CDU before repowering the L Main Bus. The PF had no idea that the AES would still leave some “bread crumbs” after the ACARS was turned off. »
@Andrew + @Victor:
thank you both for sharing your knowledge.
Your comments were very informative.
Jeff Wise says:
« I don’t consider Victor’s explanation a good one because you can turn off the sat phone from the cockpit without having to resort to such extreme shenanigans as isolating an entire bus. »
Is this true ?
If so, doesn’t he have a point there ?
Why go for such an extreme, rather complicated measure when the objective can be reached much easier ?
Could you also please clarify another aspect:
Mike says the AES (including the SDU) can be depowered from the cockpit:
« The SDU is one of several subsystems associated with the AES (Aeronautical Earth Station, or “Satcom”). The AES can be easily depowered from the cockpit. The SDU cannot. IOW, only the entire AES (SDU, Receiver, Transmitter, antenna control unit, etc.) can be depowered from the cockpit. »
This is news to me.
My understanding always was that the SATCOM can NOT be (individually/selectively) depowered from the cockpit, but only
– deselected as a source for ACARS
– or depowered as part of isolating the L main AC bus.
How can the AES “be easily depowered from the cockpit” ?
In case this is true … again, same question:
Why isolate an entire bus if the objective can be reached much easier ?
@Victor:
Mike says:
« we know from the KB PSR data that when 9M-MRO turned around at IGARI, it climbed to a higher altitude. Between 17:30 and 17:37 (inbound to KB) it made three discreet altitude reductions, leveling off at ~40,000 ft and 530 kts (above the original cruse altitude of FL350). These maneuvers are consistent with the theory that the cabin was depressurized to neutralize the passengers and cabin crew. »
Is this true?
I thought the altitude data was considered unreliable and the newspaper stories about the perpetrator climbing to a higher altitude to kill the passengers always were nothing but mere speculation.
So is this a proven fact now ?
@Andrew + @Victor:
I forgot to add, this is how Jeff says the satphone can be disabled from the cockpit:
« it’s pretty widely accepted that the pilot can deselect the Infllight Entertainment System, which the satphone is a part of, from the panel. »
@Andrew @Victor @Tim:
Here a retired B777 captain explains how a “decompression event would account for the SAT Com cycling OFF/ON”:
https://www.deepdivemh370.com/p/episode-6-reboot-redux/comment/53280191 (+subsequent comments below)
@Tim said: Although complicated, we also have to consider a damaged electrical system might not behave in a predictable way.
You are proposing a selective failure and restoration of systems with a caveat that the system might not behave in a predictable way. There is no way to refute this, other than to say it would be extremely low probability, and there are much simpler explanations with much higher probabilities that don’t require systems to behave in unpredictable ways.
@Peter Norton: The IFE switch does not disable the satphone in the cabin. On 9M-MRO at the time of the disappearance, the switch could disable cabin equipment such as seat displays, seat power outlets, and bulkhead displays. If there is documentation to the contrary, many of us would like to see it.
[Correction: The IFE switch CAN disable the satphone in the cabin. See comment of mine below.]
As for the SATCOM overheating, if I understand the scenario correctly, “Gordon Jack” is proposing that the EE bay was overheating from reduced cooling air after depressurization which caused the SATCOM to shutdown. However, the SATCOM is located in the cabin and not the EE bay.
@Tim
RE: “The power up at 18:25z may have been when ELMS control over the L AC BUS failed, due to the main battery going flat(the battery is one source of power to ELMS-the other power sources already damaged). So the load shed is now removed and power is restored to the L AC BUS and SDU.”
ELMS controls loads through the operation of relays. My understanding is that a relay must be powered by ELMS to connect the associated load to the system. If that power is removed, the relay automatically goes to its relaxed state and the load is isolated or “shed”. If ELMS loses power, it presumably loses the ability to connect loads to the system.
In your scenario, how would the L Main AC bus be restored?
RE: ”Mike says the AES (including the SDU) can be depowered from the cockpit…”.
I haven’t listened to the podcast, so I don’t know the context of Mike’s comment. All the AES (SATCOM) components are powered by the L AC SEC 2 bus. Removing power from that bus de-powers those components. He may mean the entire SATCOM system can be de-powered, but not the separate components. There is no other way of de-powering the system from the cockpit other than removing power from the bus.
@Peter Norton: I stand corrected. The IFE switch can disable the wall-mounted satphone in the cabin, although it is unknown whether or not the captain was aware of that. Nonetheless, it does weaken the argument that the left bus was isolated to disable the cabin phone.
@Victor.
I think Zahari wanted to go TOTALLY dark at IGARI, specifically to make ATC think he had crashed at IGARI.
If Vietnamese ATC had been ‘on the ball’ and followed up with Lumpur Radar within a few minutes, (instead of the very long time that they took), and if Lumpur Radar had then quickly asked MAS to call MH370 by Satphone, it is possible, if everyone had reacted quickly, that the first call attempt could have been made about the time he was half way across the peninsula.
The last thing Zahari wanted was for the call to get through, because that would prove he hadn’t crashed (yet).
He wanted ATC to believe he had crashed at IGARI, so that they would be totally focused on getting a SAR operation going out there, and not look for him anywhere else. He wanted to totally distract everyone, which would help him escape PSR coverage unnoticed, way before anyone started ‘scratching their heads’, so to speak.
Once he was out of radar coverage, an hour later, it no longer mattered.
@ventus45: What you described could be true. It also might be that he feared an interception by the Malaysian Air Force. When that no longer was a possibility, he restored systems.
We can’t know what the captain’s intentions were, nor do we know his technical knowledge.
Jeff Wise quote (to Mike Exner’s explanation)
“I should also clarify to readers that when I say that no one has come up with a good explanation, what I mean is that IN MY OPINION no one has been able to come up with an explanation that isn’t problematic for some reason or another.”
This is exactly what I am also trying to say. Jeff’s opinion (literary gimmick) he rejects all explanations other than hijack by Russia. Except he concedes pilot-intent also may explain. But none of us knows exact plans of a nefarious pilot and why. Cutting off data recorders could easily have been part of power strategy. My understanding CVR is on LEFT XFER Bus, DFDR is Right XFER Bus. Fuel savings could also be part of it.
The fact a nefarious pilot did not share his plans with Jeff ahead of time, so that Jeff could understand it, is not a justification for conspiracy theories.
Jeff’s next literary gimmick is to say fiddling with the aircraft power settings is risky and nobody knows how to do it. A few months ago on this blog we thoroughly discussed it is easy to turn off LEFT BUS and minimal impacts on aircraft. This is easy to see on PMDG777 MS flight sim. Meanwhile the aircraft is designed to fly with much more radical power-off modes.
Jeff’s next literary gimmick is to say MH370 must be in Russia somewhere, because it is not in the SIO, because the scientists said we know exactly where it is in the SIO, and it is not there. Not true either. We do not really know exactly where in the SIO the aircraft may have crashed. If someone wants to observe that there has been some over-optimism in the crash location educated-guess-work/assumptions, guilty as charged. But the evidence definitively says it is in the vast SIO, and we just missed it in the searches to date.
@TBill: Are you aware of any new data or insights that can help us find the plane? Mostly I see rehashes of old theories and analyses without much progress.
@Victor
If we say active pilot who knew what he was doing, I only give us a C- grade on that. Even those willing to say active pilot to end, have tended to get bogged down in the popular passive flight logic.
Looks to me like the best place to start is to assume “straight” to Arc5 instead of Arc6. Philosophically then Arc5 to Arc6/7 becomes more like Arc1 to Arc2: trial and error to see what fits. But I believe what fits is descent/slowdown which allows curvature over to Arc7 into the 28-32s region. With the Arc curvature, it becomes a short cut to Arc7 which saves fuel, and in addition, yet more fuel is saved by descent (if properly flown).
Implication is aircraft could be further from Arc7 (further potential distance from Arc7 in that region of the Arcs) and implies the best we can do is approximate Arc7 crossing point. Seems to me search approach is less hotspot, and more Bob Ballard approach (as I understand it) which is to try to pick up downstream debris drift trail and work upstream to source. Could have to search pretty far East though to find debris drift beyond how far MH370 itself could have conceivably crashed in the worse case scenario distance-wise.
@TBill: If you remove constraints after the 2nd arc, you have to replace that constraint with new data or insights, or you’ve simply spread out the hot spot. I don’t see anything new data in your comment that would help us to define a new hot spot.
I was hoping that somebody was doing something new. You are more active than me on other platforms like Facebook, so I was hoping you were aware of something interesting that many of us are not.
The way I see it, we have the glide area surrounding the UGIB hot spot, the glide area surrounding the Paul Smithson hot spot around 40 S, and we have the small area defined by the acoustic anomaly of Ed Anderson near Christmas Island. We also have the pseudoscience presented by others like Kadri and Godfrey, which offer nothing of any help other than keeping MH370 in the news.
So basically, nothing new.
@Andrew,
Andrew said ‘ELMS controls loads through the operation of relays. My understanding is that a relay must be powered by ELMS to connect the associated load to the system. If that power is removed, the relay automatically goes to its relaxed state and the load is isolated or “shed”. If ELMS loses power, it presumably loses the ability to connect loads to the system.
In your scenario, how would the L Main AC bus be restored?’
Thanks, if that is the case then it is working in the opposite way I was thinking. The details of the system is way above my pay grade. However, I can always say that a damaged system will not necessarily work as advertised
@Tim says: However, I can always say that a damaged system will not necessarily work as advertised.
As I said previously:
You are proposing a selective failure and restoration of systems with a caveat that the system might not behave in a predictable way. There is no way to refute this, other than to say it would be extremely low probability, and there are much simpler explanations with much higher probabilities that don’t require systems to behave in unpredictable ways.
@Peter Norton,
Peter said, ‘How do you explain
– the transponder being switched off manually,
– the flight path, particularly past Penang,
– and the SDU regaining power’
-I don’t think you can say the transponder was turned off. I think the transponder suffered a power failure and/or damage. The L transponder is on the L side of the avionics bay.
– I don’t think we have enough evidence just how and where up the Straits the aircraft flew after 18:01. We don’t have the primary radar returns to determine even if the autopilot is on or off.
– the power up of the SDU is difficult to model, a damaged electrical system will not work in a predictable way.
@Tim:
(1) re transponder: Victor Iannello wrote: « the altitude is missing for the last two [ADS-B] messages, spanning a time of less than one-half of a second. This could mean that the intermediate switch position ALT OFF was captured as the selector switch was rotated to the standby position. »
(2) re flight path: I meant the 2nd and 3rd major turn. They don’t appear random. How do you explain them in an accident scenario?
(3) Do you think the AES regained power by itself ?
(4) How do you explain that ACARS was manually turned off ? This fact is one of the hardest one to overcome for any accident scenario.
@Victor Iannello wrote:
« The time interval between the last radio transmission from the crew and the first message with no altitude reporting is 64 seconds. If the diversion from the flight path was caused by a third party forcing their way into the cockpit and taking control, those events would have to have occurred in 64 seconds or less. It is very unlikely that this could have been achieved by a third party in such a short amount of time. »
In case of a hijacking, at the moment of the last radio transmission the cockpit could have already been under control of hijackers, holding the pilot(s) at gunpoint or otherwise physically threatening them.
@Victor Iannello:
May I please ask you:
(1) Is it true, that MH370 climbed at IGARI ?
(2) What could be reasons for descending between 17:30 – 17:37 ?
(3) Do you have any news about the proposed search by OI since you posted this article in March ?
@Victor Iannello wrote:
« As for the SATCOM overheating, if I understand the scenario correctly, “Gordon Jack” is proposing that the EE bay was overheating from reduced cooling air after depressurization which caused the SATCOM to shutdown. However, the SATCOM is located in the cabin and not the EE bay. »
True, but wouldn’t the system responsible for monitoring the temperature and shutting down equipment (including the SATCOM) still be in the E/E bay ?
Or is this a fail-safe mechanism built-in in the equipment itself (in this case in the AES located in the cabin) ?
For reference, here is the scenario proposed by Gordon Jack:
« If the aircraft was depressurised. The Equipment cooling would be compromised.
In the Normal state, differential pressure, between the pressurised cabin and the outside air woud route a flow of cabin air through the Electrical racks providing cooling to the electrical systems before being exhausted overboard.
When this cooling flow goes to Alternate Mode.reduced cooling is provided by an Equipment Cooling Fan. This fan does not provide sufficient air to cool all components and some automatically shut down to protect them from overheat.
IF MH370 was depressurised it is conceivable the the reason the SatCom switched off and then back on is associated with the unit being Automatically shut down, then when they temperature in the Equipment Bay reduced sufficiently, automatically restarted. […] I have been in the El/E Bay of a 777 in flight. The heat generated is significant. »
to clarify: in the comment above, I meant “locally built-in”
> @Andrew wrote:
> « There is no other way of de-powering the [SATCOM] system from the
> cockpit other than removing power from the bus. »
Many thanks for your confirmation!
> @ventus45 wrote:
> « I think Zahari wanted to go TOTALLY dark at IGARI »
I share your opinion that whoever took the plane apparently wanted to make absolutely certain that no communication could occur. This could explain the radical measure of isolating the L main AC bus.
> @TBill wrote:
> « Cutting off data recorders could easily have been part of power strategy.
> My understanding CVR is on LEFT XFER Bus, DFDR is Right XFER Bus. »
DFDR is a good point. Does the fact that the SDU had lost power give us any insights into whether or not the DFDR also lost power ?
@Peter Norton
RE: SATCOM Cooling
“Gordon Jack’s” description of the B777 forward equipment cooling system is wrong in many ways. He says that in the “normal” state, cooling is provided by the flow of air caused by differential pressure between the pressurised cabin and the outside, and that in “alternate” mode reduced cooling is provided by a fan. That description is the reverse of how the system actually works. The “normal” mode (properly called AUTO mode) relies on fans that direct cooling air drawn from the cabin into the equipment. The warm exhaust air is then ventilated through the forward outflow valve or into the forward cargo compartment. The “alternate” mode (properly called OVERRIDE mode) relies on differential pressure to draw cooling air over the equipment.
Equipment cooling in the MEC could be an issue in the event of a depressurisation combined with the use of override mode (eg supply fan failure, smoke detection, forward cargo fire), but that is not the scenario “Gordon Jack” presented. In any case, and as Victor mentioned, the SATCOM equipment is not located in the MEC. It is mounted on the E11 rack above the ceiling in the aft cabin. That equipment is cooled by the aft equipment cooling system, which is entirely separate from the forward equipment cooling system.
The aft equipment cooling system has two fans that operate one at a time to continuously draw air over any equipment in the aft part of the aircraft that needs active cooling, including the SATCOM components. The warm exhaust air is directed through the rear outflow valve. If both of those fans fail to operate, the SATCOM system has its own backup fan that operates automatically to cool the SATCOM equipment.
@Peter Norton
RE: “True, but wouldn’t the system responsible for monitoring the temperature and shutting down equipment (including the SATCOM) still be in the E/E bay ?
Or is this a fail-safe mechanism built-in in the equipment itself (in this case in the AES located in the cabin)?”
As far as I’m aware, there is no separate system that’s responsible for shutting down overheated equipment on the aircraft. Most electronic devices shut down automatically if they get too hot. As I said above, the SATCOM equipment has its own backup fan that operates automatically in the event the aft cooling system fans fail to operate.
@Peter Norton,
You Asked-
1) ‘re transponder: Victor Iannello wrote: « the altitude is missing for the last two [ADS-B] messages, spanning a time of less than one-half of a second. This could mean that the intermediate switch position ALT OFF was captured as the selector switch was rotated to the standby position. »
(2) re flight path: I meant the 2nd and 3rd major turn. They don’t appear random. How do you explain them in an accident scenario?
(3) Do you think the AES regained power by itself ?
(4) How do you explain that ACARS was manually turned off ? This fact is one of the hardest one to overcome for any accident scenario.’
1/ I think the jury’s still out how passing the ‘alt off’ affects ADS-B. Could be also be evidence of a failure of the altitude feed just before power failure.
2/ 1st turn seems to be autopilot off manually flown. The Penang turn shows a lazy non normal turn, so autopilot possibly still off. Any subsequent turns are unknown as not captured by radar. Who knows how many other lazy turns there were as the aircraft progressed south. Hypoxic ghost flight with autopilot off.
3/ Yes, I think the electrical system reconfigured itself for some reason.
4/ Any damage in the left side of the avionics bay could cut the link between L Aims and the SDU. ACARS would not send any data.
@Andrew: thanks again for sharing your knowledge, which I must say is truly impressive! It’s like you had devised the aircraft yourself.
@Tim: I agree with Victor, that it’s hard to refute your argument that damage may have unpredictable consequences. My 2nd question to you may have been unclear: What I meant was that the entire flight path really doesn’t seem random: Turnback immediately after IGARI, straddling country borders on the way back, turning at the right moment at Penang to align with an airway so that the flight doesn’t look suspicious, SDU repowering shortly after leaving radar coverage, etc. I don’t think you can argue that this is all coincidence and it also doesn’t look like the erratic actions of a hypoxic pilot, but neatly fits the scenario of someone wanting to disappear.
@Peter Norton,
I see MH370 as a classic case of an accidental hypoxic flight, where the aircraft continues on to fuel exhaustion with no one at the controls. The only difference this time is that the autopilot is out so the aircraft does not follow a predictable path.
Backed up with the few electrical issues we know happened, it points to damage in the L side of the avionics bay. Just where the Oxygen bottles are mounted.
Until I see good evidence of a something nefarious without any unnecessary speculation I will continue to pursue the accident scenario.
@Peter Norton
My understanding from past discussions (caution: I am slightly rusty)-
To Turn off DFDR from Cockpit-
RIGHT BUS OFF
RIGHT XFER BUS OFF (DFDR is on RIGHT XFER BUS)
RIGHT TIE OFF
(SATCOM is not impacted)
To Turn off CVR from Cockpit-
LEFT BUS OFF
LEFT TIE OFF (This cuts off SATCOM)
LEFT XFER BUS OFF (This cuts off CVR)
This implies a more complete power outage at IGARI would be needed to cut off both DFDR and SATCOM. If you opt for full power outage (add CVR outage), cut off LEFT XFER BUS and RIGHT XFER BUS, the RAT deploys, which is extreme but suggested by some groups such as CAPTIO/CAPTION/Blelly-Marchand and recently Mentour Pilot adopted their view.
The fact that SATCOM is OFF is easy to see (eg; on MS Flight sim PMDG777) due to EICAS screen messages. The CVR/DFDR outages are not announced on EICAS screen (using this method), so that would have to be known to the pilot.
Andrew has previously pointed out DFDR outage by circuit breaker in MEC Bay does give an EICAS screen message. Greg Feith (former NTSB) has pointed out DFDR is 25-hr loop, so even if DFDR was turned off, there would be prior flights on there, and we could see when power was cut off to the DFDR. So even if the DFDR is empty, it could still be very important evidence.
@Victor
This is a very quiet MH370 period since March actually, everyone waiting to hear OI search status. There has been near complete secrecy and silence from all “players” in that since March – and the player list is presumably long (Malaysia, OI, NoK, anyone who has a hotspot below 25s). Kadri paper was a topic but quiet now, does not seem to hold up.
Michael Glynn book is just out on Amazon/Kindle, but no discussions about it yet.
I think it was probably active pilot to end. Implies we are searching too far South and West, because flying without maneuvers until fuel exhaustion at FL350+ is not a logical strategy to end an active pilot flight, which presumably had a certain end-game in mind.
@Andrew:
Here is Jeff Wise’s answer to your comment about SATCom cooling:
« Ron Rogers made the point in our interview that fans are used on the ground, but not when the plane is flying at altitude; in that case, circulation is provided by the differential between cabin pressure and the outside environment. He also pointed out that if most of the electrical system had been shut off then there would be commensurately little heat to disperse. »
As far as I understand, this claim stands in contradiction to your explanation above and is therefore erroneous ?
@TBill:
Many thanks for the information. So the upshot seems to be that the SDU having lost power doesn’t tell us anything about whether or not the DFDR was depowered, correct ?
@Tim:
I actually want to believe in an accident scenario because the pilots seem like nice guys. But if “no one is at the controls” and “the autopilot is out” who makes all those turns ??
@Peter Norton said: @Tim: I agree with Victor, that it’s hard to refute your argument that damage may have unpredictable consequences.
However, I believe there is very low probability that the systems we are describing behaved in unpredictable ways, especially since there are relatively simple scenarios in which systems behave as they should. That said, it’s simply impossible to prove with 100% certainty that they behaved predictably without observational data.
@Peter Norton
The forward equipment cooling system does not work the way it was described in Jeff Wise’s comment.
The system has two operating modes, AUTO and OVERRIDE, manually selectable by a switch on the overhead panel. The system reconfigures automatically to OVERRIDE mode in certain situations, such as supply fan failure, smoke in the equipment cooling system or if the FWD CARGO FIRE ARM switch is armed. The system is normally operated in AUTO mode at all times.
The FCOM states:
The forward systems use internal fans and valves to direct air drawn from the cabin to the equipment and ventilates the warm exhaust air into the forward outflow valve or the forward cargo compartment, if the compartment requires additional heat. There are two cooling system supply fans, a primary and a backup. If the primary supply fan fails, the backup supply fan operates automatically.
The forward system reconfigures automatically to an override mode when:
• in flight, both supply fans fail, or
• in flight, low airflow is detected, or
• smoke is detected in the forward equipment cooling system or the forward equipment ventilation system, or
• the FWD CARGO FIRE ARM switch is ARMED, or
• the EQUIP COOLING switch is in OVRD
In the override mode…cabin differential pressure draws air through the flight deck panels and into the E & E equipment compartment to create a reverse flow of air across the equipment, then through the override valve to an overboard vent.
The following B777 CBT presentation has a good description of the system, starting at 11:27. A description of the OVERRIDE mode starts at 24:17.
Boeing 777 Cbt #6 Heating And Air Conditioning 63
Forward equipment cooling could become an issue if the aircraft were depressurised and the aircraft operating on standby power only, as promoted by the MH370 CaptioN team. In that scenario, all the generators are selected off and the APU autostart terminated, forcing the electrical system to operate on standby power, provided by the RAT. In that condition, the forward equipment cooling system supply fans (powered by the L & R Main AC buses) would stop and the system would automatically reconfigure to override mode. With the aircraft depressurised there would be no differential pressure and no airflow to cool the equipment. Perhaps that’s the scenario that Jeff alluded to in his comment?
In any case, the description of the forward equipment cooling system is wrong and the SATCOM equipment is cooled by a separate system, with a backup, as previously described. If Jeff is referring to a scenario where the aircraft is operating on standby power, then the SATCOM cooling issue is irrelevant because it would not be powered.
@Andrew: Thanks again for your effort, a very interesting read! I’ll point readers from the other blog to your comment here.
@All: I read Mike Glynn’s book on MH370. I think it is one of the better books written on the subject. I have some minor disagreements (such as the altitude across the Malay peninsula after the turnback), but overall, I think that Mike hits on the key points and the book is worth reading.
Here is a passage from the book referring to when MH370 was holding short of the runway before getting takeoff clearance:
Strangely, Zaharie does not call “Ready” to signify the aircraft is prepared for departure. Even more strangely, Zaharie answers a WeChat call on his telephone, and talks briefly to the female caller. This, if nothing else, is a poor example of flight discipline being set for Fariq.
I know that WeChat activity was reported in the RMP report, but I do not recall that that activity was identified as an incoming phone call from a female caller that was answered by Zaharie. Does anybody know the source of this?
@Peter
Correct, selecting LEFT BUS + LEFT TIE = OFF mainly cuts off SATCOM and not too much else of consequence. CVR and DFDR not impacted by that.
@Victor
The WeChat call shortly before take-off certainly set a poor standard for the trainee pilot. However,I respectfully disagree with Mike’s comment that Zaharie’s failure to call “Ready” was “strange”.
The aircraft was parked at gate C1 at Satellite Terminal A, close to the threshold of runway 32R. From there, it’s a short taxi to the threshold. The crew’s first contact with Tower occurred less than four minutes after the aircraft commenced taxiing. Tower instructed the aircraft to line-up on the runway a little over two minutes after that first call.
My former employer used the same gate area at Kuala Lumpur airport. The entire taxi time from those gates to the threshold of 32R is normally spent completing the before take-off procedures and checklist; the biggest hold-up is often the cabin ready signal from the cabin crew. In my view, it’s likely the crew were only just completing their before take-off checks when they were issued the line-up instruction.
In my experience, it’s not unusual for ATC to issue line-up instructions without a ready call from the aircraft. At busy airports the controllers are trying to keep the traffic flowing and they expect aircraft to be ready when they reach the holding point. The ATC transcript in the SIR shows there were 11 aircraft on the Tower frequency around the time MH370 departed. Of those aircraft, only 3 reported “Ready”, including one that reported “Ready” after it had received line-up instructions.
@Andrew: I would interpret a “Ready” call from an aircraft after a “line up and wait” as a gentle reminder to Tower that the aircraft believes the conditions causing the wait (e.g., traffic vacating the runway or activity on a parallel or crossing runway) are resolved and the airplane wants to depart for scheduling but also out of safety concerns over sitting on an active runway (especially at night). I understand your belief that the missing “Ready” call was not strange.
@Victor
Re: .. ‘but also out of safety concerns over sitting on an active runway (especially at night)’.
Given the recent Haneda Dash-8/A350 disaster, a very valid point.
Has there been any proposal to install tail mounted antennas and provide a software update to enable TCAS to detect an aircraft on final approach that might hit you if you are lined up and holding, and / or to enable the landing aircraft in such a situation to get an advisory that there is an aircraft on the runway, and thus generate a go around warning – (different to a ground proximity warning that might be ignored) ?
@ventus45: I don’t know. On my little DA40, my transponder stays in ALT mode whether I’m in the air or on the ground, and the same for other ADSB traffic I see on my display, unless I put the transponder in STANDBY, which I never do. Andrew might know whether TCAS on airliners can detect an aircraft on the ground. I can see that as a problem if planes holding short of the runway are detected as collision targets by landing aircraft.
@ Peter Norton: you wrote
In case of a hijacking, at the moment of the last radio transmission the cockpit could have already been under control of hijackers, holding the pilot(s) at gunpoint or otherwise physically threatening them.
This is supported by an observation that Mentour Pilot (MP) made in his video, MH370:A NEW Trace! The FULL MH370 Story, so Far..
Starting at 19:13, he makes the following observations…
1. the pilot’s voice was normal & relaxed at 01:01:14 “maintaining FL350”
2. the pilot’s voice sounds stressed/busy at 01:07:55, repeating the same message relayed at #1
MP says pilots tend to make the extra call for 2 main reasons:
1. the pilot may have forgotten that he made the call already and makes the call again to be on the safe side
2. the pilot may have been away from the frequency for a while because he has been fiddling with the radios, or turned down the volume, or something else and pilot wants to make sure that ATC hasn’t tried to call the aircraft while they were gone. If ATC responds with “ROGER” or something similar, then pilot knows ATC has not been trying to contact.
There were 7 minutes between the calls meaning something might have happened to take the captain away from the radios between the calls.
MH370 may have been hijacked by someone that the pilot knew and allowed into the cockpit, or the pilot was acting on instructions given by someone, such as an In-flight Security Officer (IFSO), whose authority the pilot recognized.
@Victor Iannello
Victor, you may recall that some time back we discussed the “Zaharie Last WeChat Log In at 00:40:02 8/3/14“ reference.
To recap, Appendix A-16 of the RMP SKMM folder details the Captain’s online activity via his cell phone. There are a number of activities referenced IM WeiXin; these are WeChat. There are two WeChat activities on 8 March; both are identical in that they are the same size; 754 bytes upload and 967 bytes download. There is also a similarly sized 754/967 byte exchange on 7 March.
The last WeChat “log in” was not a call; it was almost certainly just the WeChat app polling the server after the Captain’s phone registered on a different cell tower. All this indicates was that the Captain had not turned off his cell phone.
@Victor
@ventus45
Yes, pilots sometimes make a “ready” call after line-up for the reasons you stated. I only mentioned it to highlight that it was slightly different to the other “ready” calls that occurred while the aircraft concerned were still on the taxiway. In this case, I suspect the “ready” call after line-up was a subtle hint to ATC the crew was still waiting for a take-off clearance. Kuala Lumpur has a widely-spaced parallel runway system, with 32R normally used for departures and 32L for arrivals. The transcript shows there were no aircraft landing on 32R during the period.
On the TCAS question, I don’t know of a proposal like the one mentioned by ventus45. Landing aircraft can “see” aircraft on the ground as TCAS targets, but the resolution isn’t good enough to determine if a target is on the runway, and the system certainly won’t issue an advisory. At this point, pilots still need to rely on the Mark I eyeball to determine if the runway is clear. That’s not always easy when there’s a sea of other lights, as the Haneda accident showed. On departure, the Airbus and Boeing procedure is for pilots to check the TCAS display for signs of other aircraft on approach to land.
I don’t know much about it, but the FAA was evaluating a ground collision avoidance system some time ago, which didn’t go ahead:
A runway collision warning system for pilots stalled at the FAA
Honeywell is currently trialling another system that uses ADS-B:
Honeywell Is Developing and Testing New Software To Prevent Runway Incursions
@Mick Gilbert: Yes, I remember the 754/967-byte exchange. I thought perhaps Mike Glynn had new or contradicting information.
@Victor
@ventus45
Correction to my previous comment: TCAS II does not display aircraft that are on the ground. My apologies.
@Victor
@Andrew
My initial thought was for creating a new TCAS software feature, which would provide a warning for the aircraft holding on the runway from it’s own TCAS, of an aircraft maintaining a steady bearing approaching from the rear at 100+ knots !
If so, it could self generate an RA to the approaching aircraft’s TCAS if that aircraft crossed a minimum range threshold (say 1nm for example which would effectively equate to 30 seconds or so to impending impact), forcing the go-around.
Feasible ?
@ventus45
Not sure if it’s feasible. At the very least it would take a major rethink of the current TCAS logic, which is designed to exclude aircraft on the ground.
I think the SURF-A system that’s under development provides a far more comprehensive ground anti-collision system that can be integrated with other systems that enhance pilots’ situational awareness on the runway. The SURF-A system is based on GPS data and ADS-B, and can provide warnings for a number of different scenarios, including an aircraft lined-up on a runway (your scenario), an aircraft entering or crossing a runway, or an aircraft on a crossing runway. It can also be used to warn aircraft of ground vehicles on the runway.
@Victor
@ventus45
A new approach to airborne collision avoidance has been under development since 2008 and looks set to eventually replace TCAS II. ACAS X is compatible with TCAS II systems and presents information to pilots in the same manner, but the alerting logic is entirely different. ACAS X also uses an array of surveillance sources and has variants that suit different types of aircraft operation, including small unmanned aircraft (ie drones).
ACAS X update: A small, big change
EUROCONTROL – ACAS Guide (p.22, et seq)
@Andrew
Thanks for those references. Heavy reading ahead !
@Victor – I recall that Ed Anderson was working on further research into his acoustics related work using some additional tools that he had mentioned would take some time. It would be interesting to hear any updates on that.
There are no additional research papers that I found on MH370 recently as I try to read up on most of those. I did find it interesting that @Viking and @Ed both have the same longitude recommendation almost with a 3 degree difference in the latitudes they recommend though approaches are quite different.
@ST: Thank you.
@ST
I do not know how far back in time you went concerning ‘additional research papers’, but did you read the UK Nature paper from May and my update a couple of weeks before?
@Ed
I did not hear from you for quite some time. Are you interested in some updates to the accoustic input I sent you in April?
@All
Does anyone know the absolute accuracy of the UTC times recorded by Inmarsat in their satellite log. Are we talking microseconds, milliseconds or seconds? The relative accuracy is clearly in the microsecond-regime.
@Viking – Thank you. Both your paper and the paper from UK were previously discussed in this forum. I was referring to the period after May 2024 to present.
@ST
Fine, I just wanted to be sure you har read these two papers.
Do you happen to know something about the absolute timing issue? One of my Australian friends brought it up this week, and I did not have a good answer.
I tested my TV-set against my computer and found a 6-second delay for the computer. The TV-set is linked to a relaible standard through an arial network with only two intermediate stations while the computer gets its standard from somewhere on the Internet (through 3R regenerators, servers, long fiber cables with group refractive index around 1.45, etc). That naturally gives some delay, but 6s was more than I had expected.
@Viking: If that approach is to use the network timestamps in the satellite logs to calculate the roundtrip time for packets along the path GES/SC/AES/SC/GES, there is no way the timestamps have sufficient resolution or accuracy due to undefined and unrepeatable network latencies. That’s the reason that Inmarsat chose to include the BTO values in the logs. That said, one investigator from Australia (Sergio Cavaiuolo) continues to promote this approach.
@Victor
That was not at all my reeason to ask. Even though the relative accuracy seems to be fine, it is always a poor approach to subtract two large numbers to arrive at the result. It tends to maximize all types of errors.
My deeper reason to ask is that I noticed that there seems to be a small systematic discrepancy between the absolute Inmarsat timestamps and data from other sources (e.g. details recorded by the French weather satellite). If the Inmarsat data was recorded with 10-30 seconds delay things would fit better.
Do you think such a relatively long delay is realistic? I guess the logs were not originally made for delivering precise absolute times, so perhaps Inmarsat did not consider a moderate, constant delay to be a problem.
@ST and All:
Thanks for asking about my recent progress on the MH370 acoustics.
I’ve been working on cepstrograms to visualize echoes and have come up with some big improvements on the standard cepstrum algorithm. (That log method generates noise from nothing, especially near nyquist cutoff). It’s fascinating to see the exact timing between airshots for two simultaneous seismic surveys arriving at the H08 array:
https://drive.google.com/file/d/1529UufMTPV2g5M1_mwGMrkcQSM2tu3as/
Note the two very loud event arrivals. The first at 00:11:30 is coming from a huge 20 km ice floe about 800 km from Antarctica. The bigger spike at 01:59:20 is the Java anomaly that is consistent with a section of sinking MH370 hitting the seabed.
The bigger and more recent breakthrough is a new algorithm that is now picking up known lightning strikes at both H01 and H08 hydrophone arrays. Not all strikes stand out, indicating that the previously described sparse convergence zones need to overlap with terrain for conversion to SOFAR depth. The directional bearings appear to involve reflections off of seamounts and coastlines. Quakes that were barely detectable are now prominent.
Most significant is that both arrays are detecting a unique event that coincides with a surface impact at approximately 7th Arc time. The arrival times are again consistent with the Java candidate site.
I’m still validating against known events. Yet to do is to try the new algorithm on the IMOS hydrophones to see if any caught the impact in their 5 min on, 10 min sleep cycle. It’s also possible that this technique might also work with island or coastal seismometers acting as T-wave sites. I’m still grokking what’s going on, but the sound waves appear to be traveling outside the SOFAR channel, but nearly at SOFAR speeds. At some point I’ll get the details written up.
@Viking:
You’re of course welcome to send new info privately or share it publicly. I’d still like to see the unmodified WX Sat captures besides METEOSAT that you’re using, and I’m sure others would as well.
I wish you’d take a closer look at my report on the barnacle growth. You seem to think that I analyzed them growing from the crash site. Instead, careful examination of the highest resolution flaperon photos shows the barnacles growing on top of abrasion damage that could only have been caused by prolonged beaching. That is, growth started after the flaperon beached at Reunion, when the winter temps got cool enough for growth. The sea surface temp graph for mid-May to July is a good match for the chemical analyses shell growth temp charts. It’s apparent that the flaperon first beached before mid-May. Most other debris didn’t have barnacles because it never got into waters cool enough for growth, which is consistent with a tropical crash site.
On clock accuracy, the best you can get is probably from GPS. With speed of light at about a nanosecond per foot and accuracy within several feet, that’s hard to beat. An app on your smartphone should give GPS time vs cellular time, which is itself good enough to assign TDMA timeslices. I suspect INMARSAT is about as accurate as cellular, but that’s just a guess. GPS sats have atomic clocks, and they are all transmitting their time and location to your GPS receiver.
@Ed: If you can show that an acoustic event just after 00:19z has the same location on the 7th arc as your previously identified Java acoustic anomaly, that would greatly elevate the interest in that site.
@Ed – Thanks for sharing your latest work and updates. Your new work sounds very interesting. I expect there will be a lot of interest in your recommended search location as you are able to validate against known events and can share your results. All the very best!
@370Location
Ed, thanks for the nice summary level update on hydrophone acoustic signals, look forward to your writeup, and as Victor notes above hopefully a good tie with 00:19 UTC and 7th arc.
Most of you are probably aware that the work of Vincent Lyne has been picked up by many news outlets around the world.
https://nypost.com/2024/08/26/world-news/scientist-shares-where-mh370-could-be-perfect-hiding-place/
In addition to recommending an area to search for the debris field, Vincent believes that the 0.7g downward deceleration inferred from the final two BFO values is incorrect. He proposes a new “Declination Model” that he believes properly accounts for the horizontal velocity (groundspeed and track) that is ignored in the “Up-Down” model that has been previously used. This is described in the following paper:
https://www.researchgate.net/publication/366587345_Flawed_Vertical_Dive_Conclusion_from_MH370_Doppler-Shifts
I don’t understand why Vincent believes that horizontal velocity is ignored in the previous models. Nor do I understand why he says that previous models assume the pre-Doppler compensation of the SDU completely compensates for the horizontal velocity. (It does not completely compensate for the horizontal velocity because the pre-compensation model assumed the satellite was geostationary, i.e., no orbital inclination. That’s why the BFO can be used to prioritize paths that satisfy the BTO constraints.)
Maybe I am missing something. Has anybody else tried to make sense of his claims?
@Victor
I do not understand/agree with all of Vincent’s arguments, but I am in philosophical agreement with him: active pilot to end possibly to the Broken Ridge area. I believe the preponderance of the data – BTO/BFO, debris drift, simulator data – supports that Vincent Lyne could be directionally correct. To some extent Vincent uses logic, so for example, he gives extreme importance to pilot avoiding JORN in his flight path.
Several years ago, when I finally proved to myself that the BTO/BFO strongly supported a flight path towards 32s, and that there were probably fuel reserves to keep flying further out, I then saw a MH370 FB post by Vincent, and had to admit Vincent Lyne was already ahead of me on the assertion.
I personally do not see an issue with the Arc7 BFO, but I don’t see why we need that accusation. I envision a series of descent maneuvers after Arc5 almost obvious final deliberate descent at Arc7, followed by loss of SATCOM. Like Vincent, I envision the pilot flew off quite some miles with some fuel remaining after Arc7, but at lower altitude so much less fuel efficiency and less glide potential, as some some of the glide altitude is used up before Arc6.
@TBill: I was only referring to the part of Vincent’s work that we can objectively critique, i.e., his interpretation of the final BFO values. His path reconstruction is pure conjecture, as are your “feelings” about how the plane was flown. You (and Vincent) may be right or wrong.
Victor,
As best I can tell, Lyne seems to think that the SDU compensation is computed using the full 3-D space velocity (not correct), which differs from the horizontal component for an aircraft with a significant downward pitch. Otherwise he seems to be arguing about differences with no distinction.
Lyne has authored close to a dozen articles on his ideas in the past few years. It looks like this one:
https://www.researchgate.net/publication/355242503_Final_Two_Communications_from_MH370_Supports_Controlled_Eastward_Descent_Scenario
is the one that spurred the article in the nypost (and other fine news outlets). He claims it has been accepted for publication in the Journal of Navigation. If so, one hopes that JON will reconsider.
@Victor:
I don’t think I’ve mentioned Lyne much since I reviewed his paper back in 2021 (which incidentally triggered an outside kerfluffle):
https://mh370.radiantphysics.com/2021/03/07/two-new-reports-on-mh370-debris-suggest-a-high-speed-final-descent/#comment-30968
I stand by that review.
Looking for it brought up some related links I’d seen before:
https://www.google.com/search?q=“vincent+lyne”+”ed+anderson”
Along with a new surprise. Lyne put out a paper last week on Aug 20 reviewing Kadri’s claims:
https://www.researchgate.net/publication/383240262_Sounding-Out_MH370_Locations
(n.b. The pdf file properties author is Mister Squiggles.)
He attributes my Java anomaly to dynamite fishing, yet also says there wasn’t an MH370 surface impact detection so it must have been a ditching. Even a ditching would be more than equivalent to 10 kilos of TNT.
Like the declination model (vector vs orthogonal look at the same data?), Lyne also came up with unfathomable new measures of the acoustic arrivals. I can’t make sense of his posed riddles that he solves. I’ve read through all of his papers, but fundamental errors and difficulty deciphering have put me off of even trying to fact check, which might take forever.
To be generous, he appears to use deductive reasoning as many do. He started with a premise of the pilot being mastermind who must have set out to baffle us all because we haven’t found the plane. (Wouldn’t the FBI have also examined browsing history but found nothing suspicious, or was he too smart to leave such evidence?) Shah’s public presence was long videos like helping people to find the optimal refrigerant pressure on their air conditioner. There is no indication that he was deceptive or trying to hide anything. Anyway, Lyne then looks for evidence to support his premise, but neglects to test that his picks actually don’t.
@sk999, @370Location: Thank you for your input.
The preliminary report into the Voepass ATR-72 accident in Brazil was released a couple of days ago:
CENIPA Preliminary Report: PS-VPB, 9 Aug 24
According to the sequence of events in the report, the crew received several speed related warnings in the 2½ minutes before aircraft control was lost. The crew appears to have been pre-occupied with descent preparations at the time, and there is no mention of them acknowledging or taking steps to deal with the warnings, other than a comment about “a lot of icing” and the activation of the de-icing system.
@Andrew: Private pilots are taught to hand-fly the plane when icing conditions exist so that changes in performance can be more readily detected. Perhaps this would have provided further awareness that ice was accumulating to dangerous levels.
@Victor
Possibly. Many years ago, the FAA banned use of the ATR autopilot in icing conditions. That ban was lifted after modifications to improve the aircraft’s handling characteristics in severe icing.
Here’s an interesting article on the history of the ATR’s icing issues. It’s long!
Into the Valley of Death: The crash of American Eagle flight 4184 and the ATR icing story
@Andrew: Thanks for that paper. Admiral Cloudberg put an impressive amount of work into compiling that story.
I’m sure there will be a lot more describing this incident in Atlanta, but a CRJ-900 holding short and perpendicular to the runway was clipped by the wing of an A350 that was passing on the taxiway parallel to the runway, shearing off the CRJ-900’s tail.
https://www.youtube.com/watch?v=xBV8aDIO1bg
@Victor Iannello
Here’s Juan Browne’s piece on that mishap https://youtu.be/U95S6dQSga8?si=izFvh8i5RxXaenH8
@Mick Gilbert: Juan did a nice job, as usual.
@Andrew: Can you provide some perspective on what is typical in a situation like this? Can the CRJ-900 crew be faulted for not pulling up to the hold short line? Can the A350 crew be faulted for proceeding past the CRJ-900 as it was positioned? Should the tower have prevented this mishap?
@Victor
The CRJ crew could be criticised for stopping well short of the holding point, but it’s not clear if they had stopped there before the collision, or if they were still taxiing forward. AvHerald reported the CRJ had stopped short behind another aircraft at the holding point, but there doesn’t seem to be enough room for another aircraft in front. Perhaps they had initially stopped behind another aircraft and were then taxiing forward to the holding point after the aircraft in front entered the runway.
ATC obviously has a role to play in preventing ground collisions, but given the controllers’ perspective and distance from some parts of the airfield, it’s often difficult for them to determine the separation between taxiing aircraft. They’re also extremely busy handling multiple aircraft as well as coordinating with other controllers, so there are plenty of distractions that might take their eyes off a developing situation.
Ultimately, it’s the pilot-in-command’s responsibility to avoid a collision with another aircraft while taxiing. In this case there are some mitigating circumstances because the PIC was seated on the opposite side of the aircraft and would have lost sight of the CRJ as they taxied past. The PIC would then have had to rely on the FO to advise if there was sufficient clearance (or not), but even the FO’s view is compromised because the wingtip is so far behind their seat position. The safety lesson here is that if there’s any doubt the safest thing to do is STOP!
@Victor
Video of the collision. The CRJ appears to be stationary immediately before it was hit by the A350.
Collision between Delta A350 & Endeavor CRJ-900 – Atlanta, 10 Sep 24
The Royal Aeronautical Society (Australian Branch – Canberra) has finally released (after 6 months) video of Peter Foley’s presentation at ADFA on the eve of the 10th Anniversary of the vanishing of MH370.
https://vimeo.com/997685457?&login=true
@ventus45: Thanks for making everybody aware of the presentation, which was a very fair representation of the disappearance, the search, and where we are today.
@Ventus45. Thanks. The transcript helps.
@Victor. Re those OI data holidays (at 50/51 mins) of 3.1%, he describes how Fugro’s like 0.5% resulted from its AUV search of any gaps exceeding 200 X 200 metres in its initial sweep, assumed to be the minimum size of the debris field.
While you have proposed that OI search one large steep gap, the question arises as to whether there are more from OI’s search also warranting attention, based on that 200 X 200 assumption?
Also, re the Lower Probability of Detection Areas comparison, do you know if these were to the same standards and under similar scrutiny?
@Viking in your Aug 23 post you wrote “My deeper reason to ask is that I noticed that there seems to be a small systematic discrepancy between the absolute Inmarsat timestamps and data from other sources (e.g. details recorded by the French weather satellite). If the Inmarsat data was recorded with 10-30 seconds delay things would fit better.” Could you elaborate on how you have determined a timebase discrepancy?
@David: I didn’t mean to imply that OI should only search the large steep gap. Rather, it was the area I deemed to be the highest probability. I would recommend searching all the gaps.
I don’t know the criteria for the areas with lower probability of detection.
@paul smithson
It was mostly a gut feeling based on a couple of small details (not just from the French satellite). I have no direct proof and no exact value yet. In addition, I think I may have found another dominant reason for the most significant contribution, since I wrote the post. I will return with more information on that particular point later. I am still working on it, and need to discuss it with one particular expert in the field before proceeding.
I still think a couple of seconds delay (now maximum 15s) is likely and improves the fit with other data, so the post is still relevant.
@Viking, thanks for responding.
@Victor Thanks. In any future search it would be interesting to see if the minimum gap size to be searched is based on that 200 by 200.
As to the standards and scrutiny, hopefully a contract with Malaysia would stipulate those.
One more comment about Peter Foley’s talk. He expressed high confidence that there had not been a controlled ditching, observing that the character of the damage to the inboard end of the right outboard flap indicated that had been from impact with the adjacent flaperon while the flaps were retracted.
Yet in the collision envisaged, in which the flaperon’s rear spar penetrated the flap’s fibre glass seal pan there should have been similar damage from the front spar, yet none is evident.
(So while the front spar itself was in fact bent, presumably that resulted from a later collision. In any case its bend was in the opposite direction to the rear spar’s.)
As to the internal damage to the flap’s inboard end, quite evidently with the flap retracted, some believe the explanation for that to be that it had indeed been deployed but then was forced back on water impact, that damage following, so consistent still with the flap being retracted.
Since its hinges would need to be unbroken for retraction, that would indicate its operating jackscrew failed on impact or its ball nut was stripped since it is unlikely that its ball nut would be forcibly spun backwards against a fine pitch – and besides the jack screw is itself braked.
The forces that broke hinges and, if not broken already, the flap, would then need to follow.
To me there can be no conclusive explanation without more detailed investigation. However, supposing the ATSB’s conclusion was supported by Boeing, with its knowledge of stressing, ductility, and the relative strengths of jack screw and hinges, that would explain Peter Foley’s confidence.
The satellite pings are obviously falsified. It shocks me when I see so many otherwise intelligent people ignore basic common sense.
There’s no corroboration for the satellite pings. No other countries or companies can confirm the data. There’s no witnesses who saw the plane anywhere along the path. There’s no acoustic detections in the water despite multiple detectors in range. We found zero debris in the area indicated by the pings.
How are some of you so stupid?
You really think the computers just came back online at 18:25UTC randomly? The ‘suicidal pilot’ decided they wanted to let people track them all of a sudden?
Katherine Tee – REAL WITNESS saw the plane after this and it had no navigation lights. You’ve decided to completely ignore her and put all your faith in some falsified rows on an excel spreadsheet instead.
MH370 is NOT in the South Indian Ocean. It’s physically not possible for debris to have drifted to Africa from there. 100% impossible based on simple physics. The currents don’t even flow West, they flow East. The debris would have washed up in Australia.
Please, use basic common sense. It’s embarrassing to me how me, as a random guy, was able to figure out more about this plane in a few months than all the other ‘experts’ combined.
What a joke.
Why does everyone only consider chord wise (fore aft) forces on the flap ?
Why does no one consider lateral (span wise) forces as well ?
Either in isolation, or combined ?
@David, you said:
“(So while the front spar itself was in fact bent, presumably that resulted from a later collision. In any case its bend was in the opposite direction to the rear spar’s.)”
If both spars are bent, and in oposite directions, (in the one event – near simultaneously – as opposed to two seperate events) then to my mind, that spells clasic beam buckling from compressive loading, which in this case, indicates a span wise force doing the loading, compressing, and bending.
Comments ?
@ventus45
Clever comment! I tend to think you are right. However, in that case both must have come off simultaneously.
@All
How far do you think the airplane could have remained airborn after loosing both control surfaces on one wing? Is 4km realistic?
@All: Due to an extended internet and power outage at my home, the comment from @Ashton Forbes was pending for a number of days. It now appears above.
@Victor
My goodness – thanks for the comic interlude.
@Andrew: You’re quite welcome. I guess if we weren’t “so stupid” we’d understand the true nature of the disappearance.
On a somewhat related note, Jeff Wise has started a Kickstarter campaign to buy a flaperon, instrument it, drop it on the 7th arc, and study the drift path and barnacle growth. The goal is to raise $50K this month, with about $700 raised so far.
https://www.kickstarter.com/projects/findingmh370/the-finding-mh370-project
All,
So, where on the 7th arc is this Kickstarter funded flaperon going to be dropped?
More importantly, how will the vastly changed differences in atmospheric weather and marine temperature induced changes to surface circulation be taken into account?
As for the barnacle growth etc., the flaperon recovered from Reunion had been beached for an undetermined period, and there is a possibility it had been detained in shallow warm water on its journey, in coral reefs surrounding the Ile de Rodrigues. Any part of the preceding statement could be relevant or irrelevant; who knows?
But I’ve been getting ahead of the facts in this story. Someone needs to reconstruct the 7th arc with the now non-existant pings!
@Barry Carlson: Everything you said and more.
@Ventus45. Apologies for the delay.
For my part I have speculated that the wing broke upwards and backwards causing the spar and flaperon damage by flap bending and inboard movement:
https://docs.google.com/document/d/1HRwXZCTyG7xHlXz1hi6-Y5VytC9t1Ndd/edit?usp=sharing&ouid=105452605762640315384&rtpof=true&sd=true
That was on the basis that the flap operating mechanisms would not reverse under excess force and also the supposition that there could be no retraction if the hinges broke. Implicitly though it did suppose the operating mechanism did not yield and that remains a question.
Bending might well have broken off the flap’s inboard end but I do not see it relevant to the flaperon’s with its different constraints.
There is unexplained damage a-plenty, including the denting at both ends of the flaperon’s leading edge. Then there are the multiple cracks along that. Those with its upper skin distortion might indicate torsion overload at some point. Also there is the leading edge damage to the flap, its trailing edge failure (maybe also in torsion) and the damage to the flaperon’s inboard end.
The last raises what contact it had with the inboard flap.
Also as I have mentioned there is the unexplained lack of damage to the flap seal pan from the flaperon’s stiffener forward of the rear spar (not from its front spar as per my earlier memory). You will see from the French photos I was alluding to, of the upside-down flaperon’s outboard end at the plate 25, SIR page 572/1423, (see below), these show how the rear spar’s outboard end was bent aft, the stiffener damage being forward.
Not being a spar, splaying of those, even if it could happen, is no longer relevant though it remains likely to me that that the stiffener damage was from another impact. https://drive.google.com/file/d/1dvFyk2E0h6Ao7ncIzoAj2QwW6IaLHLwv/view?usp=drive_link
@David
Many thanks for the reply David, I will need time to digest that.
In the first document you mention:
“The above can fit other scenarios as I wrote about (pages and pages) some months ago in more detail,
particularly about sequence. In those, the forces came principally from flaperon separation and collision
with the flap or shock on wing break.”
I am very interested in the ‘other scenarios’.
Are there links to those other ‘(pages and pages)’ ?
@Ventus. Based on dates it was this:
https://docs.google.com/document/d/1u6zAfgfZr0_cRMIxylncpGuEp2lKEfEx/edit?usp=sharing&ouid=105452605762640315384&rtpof=true&sd=true
However I thought I was referring to this. In case it is of like interest:
https://docs.google.com/document/d/1ihS5uxdY_s08H1QN-upbCcDVEJYp4q6_/edit?usp=sharing&ouid=105452605762640315384&rtpof=true&sd=true
Finally, another scenario, about the consequence of flaperon trailing edge loss in flutter:
https://docs.google.com/document/d/13SuchJe7drQWDptb8Q59Yswe-mkkUiyl/edit?usp=sharing&ouid=105452605762640315384&rtpof=true&sd=true
@David
Thankyou, it is bedtime.
@David
In your opinion, could the damage to the Pemba Flap and/or Reunion Flaperon be explained by a ground collision event, as per https://mh370.radiantphysics.com/2018/03/18/mh370-search-update-mar-18-2018/#comment-13864
OI’s Armada 78’s #04 and #08 are still waiting patiently in Singapore.
https://www.vesselfinder.com/?imo=9924273
https://www.vesselfinder.com/?imo=9924314
@Ventus. About your 2018 post suggesting the recovered part outer flap and flaperon were planted, having been removed after being damaged in a taxiing accident:
• There was a range of other items recovered that would have needed to be planted also, all of whose drift, in the conditions then prevailing, except for one outlier, was consistent with a common and feasible crash origin.
• I do not see how the replaced flap part was broken off at the flap’s inboard support mechanism in flap compression as a whole: that fracture appears to have been in bending and torsion. Besides, I do not see why instead its outboard thinner section would not have compressed sacrificially, being thinner, relieving the compression inboard.
• But putting that aside, it could be the entire flap that would need replacement. So would the flap operating mechanisms.
• Judging by the damage to the flaperon it hit the inboard flap, in which case that too would have needed repair or replacement; as might such as closing panels.
• All this would need to be concealed – and I suppose that might have meant no claim on insurance.
• Besides, a big job, likely entailing on-site hangaring etc and parts supply to there, once more difficult to conceal from any enquiry – and not just during the accident investigation, so not without risk.
• Quite likely, after such major forces, Boeing advice would be needed as to whether there was more structural damage, not just that evident.
• Supposing that the taxiing accident was just that, it would have been with an end in mind that these two particular items were put aside, to come in handy for such a nefarious purpose. Why just these and did the planning go back to then?
• As to the nature of this event there remain rational explanations as to how the aircraft crashed, or was crashed, without the need for such plantings, even though it has not yet been found.
Incidentally the flaperon was sourced quickly in France by the Boeing engineer in attendance, not by the French. He was very familiar with it and its markings.
@David, @Ventus
My previous comment only concerns the possibility that both items fell off simultaneously near the 7th arc, and shortly before the crash. I do not believe in planted parts.
However, perhaps there could be a tiny link to the old accident. What if the repair was not done perfectly (e.g. without consulting Boeing officially)? Then a violent flutter might have torn these parts off before anything on the other side. I know that airplanes are more carefully inspected than other things, but tiny cracks are extremely difficult to find, particularly if they go in an unexpected direction (due to the old damage).
By the way, do you have any idea if the airplane could have remained flying without these two parts for 4 km?
Per Victor’s suggestion, here’s a link to an interview with 9News Meteorologist Cory Reppenhagen 8 days ago. Aired last Thursday and Friday. Cory got most of it right, but as usual with TV journalists, sometimes pieces of the interview are (unintentionally, or not…) conflated. No harm done here, but some viewers thought I was soaring (engine off) for the whole leaf peeping tour, at 1000 ft AGL!! Of course, that was not the case. The engine was on for the whole tour.
https://www.9news.com/video/weather/weather-colorado/73-41025d4a-f06c-4b02-9b31-fa36fbfa8eb7
@Viking. I think the force to put a crack in the flap front hinge would leave visible flap damage. Likewise with stressing of the flaperon hinges enough to crack their bases (the actuators can swivel some).
As to flutter, to me the flaperon hinge design makes it the more prone to that, as per Boeing’s comments about it in engine ground runs and it needing protected during take-off. Failure in flutter could start with loss of its trailing edge. Of itself that might increase drag a little but would not affect overall control much.
If the flaperon front section then separated and hit the flap, the ATSB has implied that could lead to its separation, though I join some others with doubts.
To me, while that strike could have caused the flap’s internal damage from its inboard end’s support track, yet for it to be break as it did at its inboard operating mechanism would require more damage from that track than is evident. The same goes for it having fluttered to failure consequent to support track failure. Thence my favouring forced flap movement inboard by wing break as its separation cause.
More generally as to the right wing having lost these control surfaces in flight, supposing the hydraulic system had been degraded by engine failure its flaperon might be yet more vulnerable to flutter and its stresses. That is because its inboard actuator’s load would be more than doubled, its outboard now being inactive, added to by with the left flaperon itself being inactive.
Since the fractures of the right flaperon’s two actuators’ attachments look similar, I think most likely they failed in the same way, not the left in fatigue. Hence I would suppose that the load on the left hinge, in a couple with the heavy actuator load, would have been what failed first, that in turn overloading the inboard hinge in bending and torque.
Yet while that scenario might make the right flaperon susceptible to flutter and loss at high speed, in a phugoid say, again it would not account for the outboard flap part’s separation.
Irrespective, coming to whether the aircraft could fly on with both control surfaces gone from the right wing, I would have thought so, unless the hydraulics were degraded (the left flaperon inoperative0 and the ailerons locked at high speed (and I haven’t looked into what roll control the spoilers might contribute then).
As to whether the aircraft could stay airborne and stable after loss of the right flap and flaperon earlier in the flight, engines alive, left flaperon and rudder active, again I would have thought so, though profile and lift drag would increase fuel consumption.
I am unclear though on how your “4 km” criterion arises so may be missing something.
@David
Thanks for a detailed explanation, and for confirming that remaining airborne for a short distance (after loosing these two control surfaces) is possible.
4km is not a random number. I have found some new (never considered before) data that indicates ‘something with considerable weight’ fell off the airplane roughly 4km before the crash. I will return later with details, but I am not ready yet due to ongoing discussions with an expert in the relevant field.
@David
The “4km is not a random number. I have found some new (never considered before) data that indicates ‘something with considerable weight’ fell off the airplane roughly 4km before the crash” may be significant to your previous comments.
Could it have been the No.2 (starboard) engine that @Viking is describing as “something”?
@Barry Carlson
That would fit better with the weight, but I do not know if it is likely the engine came off. I presume everybody would have been alert for that type of failure after the El Al plane crashed in Amsterdam in 1993?
@airlandseaman That scenery is amazing. Thanks for sharing it with others.
@Viking. 4km after a ‘considerable’ weight loss, presumably following MEFE: maybe not, particularly if the chord of a turn:
To me the most vulnerable item to in-flight loss, supposing high and increasing speed, would be that right flaperon, its inboard hinge failing from overload or fatigue.
But not enough weight by itself by the sounds.
As raised by Barry Carlson, I suppose an engine might separate in flight due to the wing flutter propounded by Don Thompson and ALSM, at yet higher speed (IMO) but 4km to crash after that would be an ask. Turn might well have increased at flaperon loss.
No report of either separation in simulations though these possibilities might be beyond their scope.
@Viking,
The 4km distance is interesting and I look forward to your work. My Flaperon study looked at EOF scenarios from Mike Exner (2014), Victor Iannello (2017), and Boeing (2018) simulations based ATSB conditions as obtained by Victor.
See Section 5.0 Conditions of Overspeed
https://www.dropbox.com/scl/fi/ju3aa3q34kkjguvxznzah/MH370-Right-Flaperon-Consolidated-Studies.pdf?rlkey=p39hulhtwzdha7g5xsktr3xk9&e=1&st=0f93it1b&dl=0
The graphs show EOF fuel exhaustion altitudes of 35000 ft and 40000 ft (10.5 to 12km) vertical distance.
If your 4km study involves total horizontal travel (or perhaps total net positive horizontal travel) the graphs in the report may help visualize the relationship of overspeed vs. loss in altitude.
– An EOF scenario with a rapid loss of altitude reduces the opportunities for horizontal and net positive horizontal travel.
– An EOF scenario with a higher altitude overspeed condition that is sufficient to cause shedding of aircraft component(s) can lead to increased opportunities for longer horizontal and net horizontal distances.
Just food for thought…
@Kenyon
Thanks for your input. If you only give me those two possibilities, it is closest to 4km total travel, but not exactly (more complicated). The true total horizontal travel distance was near 5km.
I am now almost sure the engine did not fall off, only the two control surfaces.
@All
Does someone know the precise weight of the flap and the flaperon respectively? 10% precision is fine.
@Viking, the recovered Flaperon weighed 40kg.
Ref: Safety Investigation Report MH370 (9M-MRO) pg 564 of 1423
Infrasound is a subject that seems to have been drowned out by other noises in recent years. As the quest for “new evidence” remains, I think a fresh look at Infrasound records is warranted, as a possible source of such evidence, since it appears that at the time, such data was mostly secret.
Has that data been declassified for 7/8th March 2014 yet ?
@ventus45
Good question.
@ventus45:
The CTBTO infrasound was released from Oct 21, 2019 back to Apr 3, 2014.
Here’s the current IRIS catalog entry for Cocos Island:
https://ds.iris.edu/mda/IM/I06H5/
The dates are the same for the I04 Narrogin Infrasound Array in Australia.
https://ds.iris.edu/mda/IM/
I’ve not pursued becoming a CTBTO contractor after failed attempts, but scientists affiliated with universities and government researchers do have contract access. I would certainly be very interested in any release of the infrasound data for analysis.
I do believe that infrasound could be key to detecting the passage of MH370 using beamforming methods. We do have a LANL conference poster that indicates a pressure wave passing Cocos Island at around 22:46:40Z on Mar 7, 2014. Perhaps a researcher with CTBTO access could take a look to see if this might match with a wake vortex from MH370 passing over the atoll at 22:22:22Z as seen in the infrasound range via the public seismometer at the airport there.
https://370location.org/2018/08/cocos-island-infrasound-may-be-key-to-locating-mh370#jan2020update
@All
Here is a potentially interesting newspaper artichle on MH370:
https://www.ladepeche.fr/2024/10/25/disparition-du-vol-mh370-pourquoi-la-malaisie-hesite-a-relancer-les-recherches-12283759.php
Sadly it is in French and behind a paywall. For principal reasons I do not want it to install all kinds of unwanted apps on my PC. If one preseeds without accepting, it is only possible to read a tiny bit.
From that tiny bit, I learned that Ocean Infinity has 2 ships positioned and ready to go in Singapore and one more performing exercises in the South China Sea. However, the Malaysian government is hesitant to give the go-ahead.
I wonder why?
@Viking, all
I will suggest that story with La Depeche is a regurgitation of the speculation posted, recently, at blogs managed in Australia and Malaysia.
It’s open knowledge that three of the Armada 78 vessels have recently been undergoing work in the ST Engineering and Mooreast facilities in Singapore. That’s evident by the vessels’ positions given by AIS vessel trackers. These three vessels are of similar configuration to those already in service in Europe. Those in Europe have undergone periods of engineering work after handover from Vard Vung Tau, the building yard in Vietnam. The vessel pictured is Armada 78 03 which, along with 05, spent some time at Damen Shipyard in Amsterdam. Again I suggest, those 3 vessels having remained in Vung Tau after handover from the constructor have relocated to Singapore for catch-up to the configuration of their equivalents in Europe.
An example of the work might be the installation of an MBES (multibeam echo sounder) gondola, fixed externally, under the hull. To test an MBES, operating in water depth greater than the few 10s of metres around Singapore might be useful. The area where Armada 78 08 indicated ‘sea trials’ on its recent excursion from Singapore is the location of a shelf to deeper water that is closest in proximity to Singapore.
@Victor
A question regarding the figure in your last posting (the 2024 proposed search area):
How much of the previously searched area within S33-S36 would be included?
I was estimating based on the 15,000 km2 and I’m a bit puzzled.
@Niels: I think the approach is to go back and search the areas that have missing or low quality data, and then search outwards from what was previously searched. I would be surprised if the areas with missing or poor data amounts to more than 10% of the 15,000 km2, but I don’t know OI’s plans.
Here is Viking’s article in full: https://archive.is/y6VEk
@Victor
Thank you, Victor. Looking at the figure I estimated the “outside” areas to be close to the 15,000 km2, so that would match with your expectation.
Now the latitude range is typical for the single setting autopilot flight to fuel exhaustion for the final hours of the flight. Also the final BFOs were noted and analyzed. So personally, I would be focused on thorough (re)searching the areas close to the 7th arc (mainly the “GoPhoenix” area), rather than spending most effort further out of the arc. This should include the area around S35.5 near the 7th arc, based on the “Pleiades” hypothesis.
Based on more final results that I obtained by fitting 20 million random paths, I agree with the S33-S36 range, though I would rather stretch it towards S31.5 – S37.5. I’ll try to share some more details in coming weeks.
This season’s ‘weather window’ in the SIO has opened.
OI have three of their Armada 78’s (#4, #6 & #8) moored side by side in Singapore Harbour, ready to go.
https://www.vesselfinder.com/?imo=9924273
https://www.vesselfinder.com/?imo=9924297
https://www.vesselfinder.com/?imo=9924314
How long is OI going to wait for Malaysia to make up it’s mind?
@ventus45: If you recall, the last time OI searched, the contract with Malaysia was signed in the 11th hour as the ship was in transit. If a new search is conducted, there might be a similar scenario with last minute negotiations. So, it’s hard to predict what will or will not occur.
@ventus45
What makes you claim that Ocean Infinity is to commit three Armada 78 vessels to a search and how do you conclude the vessels are ready to go?
Elementary logic and deductive reasoning.
Ship owners do not (as a rule) leave their brand-new ships sitting around doing nothing (idle) for long. It is not good business practice, let alone an efficient use of expensive assets.
Positioning three out of eight of your fleet in one port for many weeks (when your other five ships are on various missions elsewhere) does not make any business sense, unless they are being readied to sortie for some mission.
Why else would these three ships be sitting there for so long ?
It is logical to conclude that they are fitted out & provisioned for a mission.
It doesn’t take a great leap of logic to guess what mission.
@370Location
Did you get my mail?
@Ventus
The information is out there if you look for it:
https://www.offshore-energy.biz/industry-first-dnv-approval-for-remote-vessel-operations-goes-to-ocean-infinity/
The 3 vessels are undergoing type testing. 1 has already received approval from the DNV. The other 2 are working towards it. As far as I understand it all 3 are based in Singapore so it explains why they are there. I’m yet to read anything official to suggest that they will be used in a search for Mh370.
@All
As I don’t find the time to write a full paper soon, I hereby share some of the key results from recent BTO/BFO path fitting attempts. A short summary of methodology is included. Please see the link below. I treated the BTORs and BFORS as independent random errors, which perhaps needs more consideration; any feedback is welcomed.
In section B, I describe results from evaluating the set of random flight paths against the measured BTOs and BFOs, where we should realize that there is an unknown random error incorporated in these values. Some broadening of the latitude probability distribution is caused by the range of frequency bias offset values considered (figure 1).
To estimate the impact of the possible random errors in the measured values, I deposit different “extra” errors on top of the BTOs and BFOs (25 sets of 10 errors randomly picked based on a normally distributed probability density), similar to what @sk999 has implemented; Section C. This leads to some additional broadening of the latitude probability distribution (figure 4).
While calculating the latitude probability distribution (fig. 1), the probability weighted mean BTORs and BFORs were also recorded. Intuitively, one could think of these as a best estimate of the actual errors in the BTO and BFO measurements. Therefore, I tried to use these values as an extra set of possible random errors in an extension to the procedure in section C; see Section D. It gives remarkable high values for the summed path probabilities, figure 7. The peak is within the bin S34.0 – S34.5.
There seems to be a dominance in the probabilities by the higher FLs (>350) and by LRC and ECON as thrust modes.
While a more detailed analysis and comparison would be desirable, it seems at first sight there is a reasonable consistency with results obtained by “UGIB” and @sk999,
https://www.dropbox.com/scl/fi/oac9fl3gyuhv5hwkqxvur/Path_fitting_Nov_2024_v1b.pdf?rlkey=bum0wnyai40y3wly3dzo3ii02&st=cs5p4b9z&dl=0
National Geographic’s documentary film ‘Endurance’ is now on general release. ‘Endurance’ describes the loss, in 1915, of Sir Ernest Shackleton’s ‘Endurance’ and the 2022 sub-sea expedition that located the wreck.
The film premiered at the London Film Festival on October 12th with a limited distribution in UK independent cinemas. It is now available globally on Disney+ and Hulu. I recommend it.
The sub-sea team included personnel involved in Ocean Infinity’s MH370 search in 2018.
The documentary describes how the 2022 expedition, during the initial period allotted for the search, had failed to locate the wreck. An additional 10 days was granted to continue the search. Lasse Rabenstein, the expedition’s scientific advisor, then revisited the possible drift path of the Endurance between two positions logged on separate days by Shackleton’s navigator in 1915. Rabenstein deduced that the drifting ice had carried the Endurance in a southerly loop between the two position logs. It was decided to task the AUV to operate in one of the remaining ‘boxes’ corresponding to the loop. The wreck was located.
I thought this notable as it reminded me of David Mearns’ description of the detailed wind and current modelling contributions from David Griffin in the search for HMAS Sydney.
The eventual success of the 2022 expedition to locate Endurance was not unlike the search for ARA San Juan: both missions were running down the clock on a time extension and both involved a review of the pre-existing information directing the searches and re-focus within the pre-defined search area before success was achieved.
@Niels
I was curious.
Niels Tas Figs 1 4 7 combined.pdf
https://www.dropbox.com/t/lU2piJaQFJL0B3qZ
From Bernama.com today:
Chong Zhemin (PH-Kampar) will seek an answer from the Transport Minister about whether the government still has the intention to continue to search for flight MH370 during Questions for Oral Answers session.
https://www.bernama.com/en/general/news.php?id=2359612
An update based on a translation to English of the Bernama.com article:
In response to the question posed by Chong Zhemin, Transport Minister Anthony Loke said that Malaysia will consider further searching for MH370 if there are clues that are authentic and robust. They have been negotiating with Ocean Infinity since June.
It sounds like a search is still on the table, but Malaysia has made no commitment to search, and can still backout with the claim that there is no new credible evidence.
https://bernama.com/bm/am/news.php?id=2359790
Further details of the proposed agreement betw. Malaysia and OI:
https://www.straitstimes.com/asia/se-asia/search-for-mh370-to-restart-based-on-credible-proposal-says-malaysia
Curious note:
“Dr Lam Choong Wah, a defence expert in international and strategic studies at Universiti Malaya, suggested that Kuala Lumpur collaborate with Beijing in the new search for MH370, especially given the upcoming 50th anniversary of their bilateral relationship.
“As the majority of the flight’s passengers were Chinese, Malaysia should not be prevented from collaborating with Beijing in the search for the wreckage. China has the most advanced seabed exploration technology,” he told ST.
Dr Lam referred to Chinese oceanographic survey and research ships, such as the Xiang Yang Hong, which have frequently operated in the Indian Ocean since 2016.”
@George G
Please note that fig. 4 may not yet be “fully developed”. 25 trials for a set of 10 random variables is probably not enough. Furthermore, the tail to the south is probably influenced by the fuel limit kicking in more heavily.
@Niels
A work in progress. Understood.
You write: “Furthermore, the tail to the south is probably influenced by the fuel limit kicking in more heavily.”
Your work concerns your recent BTO/BFO path fitting attempts.
In context, fuel limitations will provide a quite distinct cut-off of the tail to the south, when considering overall probability.
This is, I think, additional and separate to the task of fitting a probability curve to the BTO/BFO data.
@George G
I did include a basic pre-1941 fuel model for path pre-selection in the results I just shared (see link below for the model).
When calculating the 20 million paths (quite long ago) I re-iterated the initial fuel (just before 19:41, as the simulations started a bit before the 2nd arc) by subtracting the fuel left at the moment of expected fuel exhaustion from the initial fuel amount and re-run the path simulation. At the time I didn’t have a pre-1941 fuel model by I did record the 19:41 weight resulting from the above procedure for each path. Later I developed the basic pre-19:41 fuel model (which gives the max. available fuel at 19:41 as a function of 19:41 Lat, Lon) which I now can easily apply for a pre-selection, based on the recorded required 19:41 fuel amounts and recorded 19:41 position for all paths when combing through the “old” set of 20 million paths.
https://mh370.radiantphysics.com/2024/03/05/ocean-infinity-proposes-new-search-for-mh370/#comment-36390
I’ll rerun for fig. 1 and 7 without the fuel limit to see the difference.
@George G
It is good that you raised the point, as the difference with and without fuel limit is quite significant, see link, especially for fig. 1. It would be good to find out in detail which type of paths are filtered out where and why. The large difference with and without fuel limit also indicates that the peak in the latitude probability distribution will be sensitive for possible errors in the fuel model.
Note, by the way, that the fuel limit comes with a restriction in the 19:41 latitude as well (>S3 and <N10) because that is the range for which the pre-1941 fuel model is valid. The 20 million paths were derived with a uniform random 19:41 latitude distribution between approx. S4 and N14 (initial latitude between S3 and N15 @ 400s before 19:41).
https://www.dropbox.com/scl/fi/qpkxkkxkqgqbzfmhtz0jn/Fuel_limits.pdf?rlkey=lcac4ck5fxdlcbbb98ezol16p&st=m4zea9ud&dl=0
@Niels
Your Fig 1b vs Fig 1a comparison:
The influence of inclusion of your fuel limit considerations is quite evident.
Your Fig7b vs Fig 7a comparison:
This shows that your analysis leading to Fig 7 is a refinement upon that producing Fig 1.
The influence of inclusion of your fuel limit considerations is noticeably less.
In your “Path_fitting_Nov_2024_v1b” you do note that: a preselection of paths based on a “soft” fuel limit was employed. It is in the last sentence of your methodology. My apologies for not noticing this earlier.
I have for some time been inclined to consider a close to “hard” fuel consideration might be appropriate.
It is noted that the interdependency of the fuel usage with the Flight Level, the Thrust Mode and ambient conditions, including wind, as well as the starting fuel load at 19:41 is complex.
If the statistical approach were to be taken with BTO and BFO as you are taking, then having arrived at the resultant path probability, a “post-processing” of that path to determine the fuel situation for that path might be applied.
The “post processing” would be to determine whether the “remaining fuel” at the time of fuel exhaustion, for which you are assuming 00:17:30, is realistic.
Whether or not the “remaining fuel” is realistic would depend upon the Flight Level, the Thrust Mode, ambient conditions, as well as the starting fuel load at 19:41, and on the fuel model used. The Flight Level and Thrust Mode will be as used in the path probability calculation. The uncertainty on the starting fuel load and subsequent fuel usage over the (assumed) four hours, thirty-six and a half minutes would have been determined prior.
It is considered realistic to have determined upper and lower limits on the fuel uncertainty.
A flight path requiring more than the pre-determined estimated fuel available PLUS the maximum fuel uncertainty could then be removed from any analysis of overall uncertainty. In other words, for this particular path there would simply not be enough fuel to complete the flight via that path.
Similarly, a flight path for which the estimated fuel usage is lower than the pre-determined estimated fuel available LESS the maximum fuel uncertainty could also be removed from any analysis of overall uncertainty. In other words, for this particular path there would be an unexplained excess fuel situation. The path may be considered invalid.
In your “MonteCarlo_v10” of 24/3/2024 you wrote in the fourth paragraph: “As mentioned before, paths requiring a higher weight (to be consistent with 00:17:30 fuel exhaustion) are rejected in the final statistical analysis.”
I’m not sure if you have continued along that path, or you have drifted a little. I would expect a more “dramatic” drop off in overall probability than evident in your figures. Here, I am referring to path probability enveloped, and limited, by realistic fuel expectations as the “overall” probability, in the context of this discussion.
To sum up: My thoughts are that the fuel limitations may be considered as providing quite distinct cut-offs when considering overall probability. Whether or not it is appropriate to soften these by applying a statistical approach between the upper and lower uncertainty limits to “soften” the cut-offs may be a moot point.
@George G
Many thanks for your feedback; much appreciated!
First,to clarify: I still follow the same approach as described at 24/3/2024. So the cut-off is hard; the “soft” limit refers to the pragmatic approach to assume a relative efficient path between 18:30 and 19:41 to avoid rejecting in the end paths which would be valid, at the risk of suggesting a slightly wide latitude range for LEP. However, looking at fig. 1a vs 1b I have to consider if even as a pragmatic approach this works out well. So, as a first step I might start estimating error margins in more detail, based on comparing my 19:41 fuel estimates with those estimated by “UGIB” for different scenario’s. I have done that, of course, but not in all detail and I haven’t reported it.
In a later posting I’ll come back to the apparent difference in impact of applying fuel limits between figs. 1 and figs. 7. I think that better understanding those will also help to develop the right strategy.
Can I just ask something obvious here. We all have read about Malaysia stalling on a search for Mh370 without further evidence coming to light. What kind of evidence is Malaysia looking for? To my knowledge all we do have is the Inmarsat Data. Which has been pretty much exhausted in terms of analysis. People in this group & elsewhere are still debating an end point based on the data, amongst other factors like fuel exhaustion, glide possibilities, debris drift etc. Ocean Infinity themselves appear to be pretty much committed to beginning a new search. I believe they are looking at a no fee unless they find Mh370. If that is the case then can anyone tell me why Malaysia are reluctant to agree to a new search going ahead?
I’ll take that last comment back. I’m now reading that Malaysia is in advanced talks with Ocean Infinity over a plan submitted in June 2024. Apparently Ocean Infinity is seeking a $70 million fee if Mh370 is found. Once a proposal is finalised it will need final approval by the Malaysian cabinet. So let’s hope this is sooner rather than later.
https://www.9news.com.au/national/mh370-location-new-search-for-mh370s-wreckage-could-begin-within-weeks-off-wa-coast/d75a645a-c998-496a-8508-46c641757952
@ventus45
I suppose that news outlets will run with whatever “talent” they can get in front of a camera but trotting out talking head GT to spruik a torrent of palpable nonsense does little for the story from a credibility perspective.
@Mick Gilbert: True, but in some corners there is the belief that pseudo-science claims may help motivate a new search.
Read again. (?)
The 9News item includes this sentence:
“Experts say they are almost certain they’ve found MH370’s final resting ground, just 30 kilometres from where the last search took place.”
I offer no comment re the “Experts say they are almost certain” part.
@Mick, @Victor,
Perhaps I’m not reading what you are ?
@George G: I think Mick was referring to the video that can be found at that link.
@Victor Iannello
Mmm, pseudo-science as a search motivator … well 9M-MRO would, like me, be a Taurean. Perhaps I should be paying closer attention to our horoscopes. Apparently, with the Moon in Aquarius, we’re looking at a generally good week ahead.
@George G
G’day George. Yes, as Victor said, I was commenting on the accompanying video of one of our local “aviation expert” talking heads, making utterly preposterous claims while waving the latest output from the WSPR ouija board about.
@Mick, @Victor,
Reality check setting found on this computer.
@George G
Did you intend to end your last contribution with a link or did I misunderstand something?
By the way, I have no idea why many people still beleive in WSPR for long distance applications.
@Viking,
Quote: “Reality check setting found on this computer.” END Quote,
was all there was.
Reference: link posted by @ventus45
on November 9, 2024 at 12:37 am
When I first looked at the linked 9News item, the embedded video would not open, as if it had been removed, if there ever had been one.
Later, after seeing reference to the video elsewhere, I looked on another device.
Hence, I seriously looked at this computer for a “Reality Check Setting”, or other such setting.
As a result, I seem now to have removed the “reality” screen.
In context, my comment seemed to me to make an appropriate summary.
@George G
Sorry for my misunderstanding. I also klicked at the link after finding it with Google, and experienced the same problem as you. However, a few hours later one of my personal friends in Australia sent me a deep link, and that worked. When the link appeared in this Blog I did not need to use it, since I had seen the video already. Therefore I got confused: ‘Was there perhaps something more?’
@Viking, 😊
@George G
Please see through the link below a new version of figs. 1 and figs. 7 (with and without fuel limit). Note that the b figures had to be slightly corrected, as together with the fuel limit I introduced 19:41 latitude limits (S3 – N10 deg) which initially I left out when deriving the figs. 1b and 7b. I have now corrected this. Also I included probability weighted mean latitudes for all the figures, and the areas under the curves.
The best estimates for BTORs and BFORs change when fuel limits are omitted and more paths contribute to the evaluation. Therefore, I added a run (without fuel limits) where I corrected for these new random error estimates (fig. 7c).
While comparing fig. 1a vs. 1b and 7a vs. 7c we see slight differences in the effect of omitting the fuel limit. In both cases there is a shift to the south and the total probability increases by a factor 2 -3. I don’t think it is easy to explain the slight differences, given all the factors involved.
As a follow up strategy I propose to estimate a min and a max available fuel weight at 19:41 (as a function of lat, lon), where I expect that the current model is close to a best estimate (given certain assumptions). Estimating the proper amount of fuel available at 19:41 is complicated. Even if you show that the fuel model is rather accurate, practically there is a dependency on the assumptions made for the unknown flight path and modes between 18:28 and 19:41. I tried to keep those as simple as possible.
As I am assuming fuel exhaustion for all my 20 million flight paths, the fuel available at 19:41 should actually include both a min and a max. Every path having the 19:41 required fuel amount between the min and max available fuel amount, will be counted valid and included in the probability analysis. I expect the resulting latitude probability distribution to be rather close to figs. 1b and 7c.
For the error analysis I’ll use information from “UGIB” to determine the error margin (in tonnes); it may take a while to complete.
https://www.dropbox.com/scl/fi/5o5xhwema285eh4ttdy4z/Fuel_limit_2.pdf?rlkey=zsk9o25mvqwr02iqtnxrp5xsz&st=4rrhgqk3&dl=0
Just wondering if anybody was aware of this:
Back in Jan 2014, MAS decided to suspend all flights to continental USA, LAX being the last remaining destination. The suspension was to take effect Apr 30, 2014. Knowing that extra security precautions are taken when flying to USA, could this MAS route change have something to do with MH370’s disappearance on Mar 7, 2014?
https://theedgemalaysia.com/article/malaysia-airlines-scraps-los-angeles-focus-asia
According to this photo montage of 9M-MRO, it was flown there as late as Feb 28, 2014.
https://www.raischstudios.com/malaysia-airlines-flight-370/
If ZS was a senior pilot experienced in flying B777s for MAS, he must have certainly been familiar with the KUL to LAX route, as well as all the security measures unique to flying to CONUS. Does the report into the captain mention when he last flew to LAX?
@Niels,
Your Fuel_limit_2.pdf
Specific Comments:
A: “Figure 1b corrected (without fuel limit, with 19:41 Lat limits)”
B: The previous “Figure 1b (without fuel limit)”
Comparison of A with B shows a tendency for a deviation at the most southern latitudes, specifically for your Latitude Bins 28 and 29. The corresponding A vs B comparison for Figure 7b is similar.
I do not know your reasoning behind your decision to apply your 19:41 Lat limits. But, perhaps your use of South 3 degrees as one Lat limit is the main reason for the above characteristic.
Figure 7c is significantly different to your Figures 7b and 7bcorrected.
Of note is that the comparison 7c to 7a more closely resembles the comparison 1b to 1a, in contrast to the comparison 7b to 7a.
Perhaps you should conduct checks to determine whether Figure 7c, or perhaps Figures 7b and 7bc, may be suspect, if you plan to proceed further.
General Comments:
None of the above alone contributes much to narrowing down the possible range of area coverage which may eventually be necessary before the immersion site of MH370 is found. But of you find a means of narrowing the range, by further restricting fuel limitations, taking into consideration that not only will the aircraft will simply run out of fuel if some paths were to be taken, but that other paths may well be eliminated as probable paths if the path would arrive at your 17:30 fuel expiration time still with fuel in the tanks.
For the record it is noted that: “For each path the flight level (FL), navigation, and thrust settings remained constant after 19:41 UTC.” as per your introduction to “Path_fitting_Nov_2024_v1b”. It would seem this is the only practical path, and at least provides limits to as far as the aircraft may have travelled. Others have all been forced down this path, as we know.
https://www.dropbox.com/scl/fi/zs47fy1b1aznles1dhj4e/Fig1BvsFig1A-Fig7CvsFig7A.pdf?rlkey=54rix2hzekgdk4qfhdz0o3ywv&dl=0
@Niels,
You write:
“While comparing fig. 1a vs. 1b and 7a vs. 7c we see slight differences in the effect of omitting the fuel limit. In both cases there is a shift to the south and the total probability increases by a factor 2 -3. I don’t think it is easy to explain the slight differences, given all the factors involved.”
You write “the total probability increases by a factor 2 -3”. I guess that this would be rectified by you normalising your data comparisons. In effect I have done this by using your charts as they come, and by quite deliberately ignoring the vertical scale. This is how I have treated all your charts in the current discussion. As you have clearly treated the data similarly for each analysis, I felt “comfortable” in taking this approach.
The “shift” to the south when comparing fig. 1a to fig. 1b or, in other words when comparing a data set for which fuel usage has been considered with-or-to data for which no fuel usage considerations have made is quite understandable on the southern side of the probability peak, but I have not (yet) taken time to explain the northern side, or have I ?
For the southern side of the peak it is quite understandable that for some otherwise possible, or probable, paths the fuel availability limit will be reached before the subject path can be completed. Thus those paths which cannot make it are eliminated. They are declared not valid. A qualification should be made here: “all otherwise possible, or probable” refers to paths with probabilities simply determined by your BTO/BFO analysis. Some of these seemingly possible, or probable, paths will pass the fuel availability filter and remain in the group being classified.
The elimination of the paths just discussed (on the southern side of the peak) will have affected where the general peak is. This will move NORTH for each southern candidate eliminated. Hence the peak probability of the overall sample moves north.
So, EVEN THOUGH the available data appears the same it is biased northwards by the removal of any southern candidate. As you have been plotting your analysis each removal of a candidate reduces the size of the overall sample remaining, or the size of the remaining group being classified. The overall probability of any non-filtered and remaining candidate seems to progressively increase.
Hence, the removal of any southern candidate apparently improves the probability of any other candidate remaining. If this is plotted as you have done, and I have acquiesced, then the seeming probability of candidates remaining north of the peak will increase. Hence my plots of “Figure 1b vs Figure 1a” and “Figure 7c vs 7a” each clearly show a higher result NORTH of the peak than they would if all the original candidates were taken as the size of the sample.
The above “Figure” comparisons show more “north” tendency for the fuel limit applied results north of the peak simply due to the elimination of many results south of the peak. I suspect.
Example:
Figure 1a: FUEL LIMIT APPLIED, 41928 Paths contributed to the graph;
Figure 1b: NO FUEL LIMIT, 115215 Paths contributed to the graph.
115215 is 41928 times 2.748 which falls in your 2 to 3 factor range.
And this is why you have plot ranges in the ratio 5 to 1.75 = 2.86.
P.S. Yes, there is a typographical in my “Figure 7c vs 7a” comparison. Fig 1a (red) should read Fig 7a (red).
Hello all,
I did not write for a while here and unfortunately, my previous try to post a comment here regarding the search proposal did not appear (hopefully due to technical glitch) so I will try again with some rewording. I don’t expect that anyone here wastes time on explaining me something which was discussed many times before but still feel a need to say something about this search proposal.
Expanding the area north of the seventh arc and not searching further south is not a good idea and here is why.
1. Flight path derived with the assumption of staying at FIR boundaries and continuing makes a lot of sense considering early stages of flight and crossing the 7th arc at 34,2S is also in line with previous studies, which I am not going to critique since I don’t have sufficient knowledge
2. Proposed search area mostly consists of A1 area (high probability search area) presented in UGIB 2020 paper. That one assumes uncontrolled glide and final resting place close to the seventh arc. However, the highest probability areas within the A1 area have already been mostly searched (which are closest to the seventh arc), with the exception of one relatively small area with rugged terrain.
3. Area A2 from UGIB 2020 which assumes controlled glide south after fuel exhaustion was not taken into consideration. Some of you may disagree with Larry Vance who is convinced that there was no high-speed impact, because pieces of debris would be much smaller, but I think there is some truth in that. Controlled glide is more likely to yield softer landing.
I really don’t know why Ocean Infinity opted to search northern and southern part of the A1 area and excluded the A2. In my opinion we cannot any close our eyes anymore and exclude possibility that the plane was in control to the very end.
It would be more logical to search the southern remaining part of the A1 and continue with the A2. Probably the last opportunity to find the MH370 could be wasted.
@Marijan: Your previous comment was submitted under the nick “xmarijan”, and was therefore treated by WordPress as a new commenter, which required my manual approval, which I had not done.
@Victor Iannello
OK, thanks, I made a mistake and used nick from my email!
@George G
I have to think a bit longer about some of the points you bring forward.
For now to clarify some points:
– I don’t apply any normalization. I just add all probabilities in the right latitude bin for all path that pass the selection criteria. The formula to calculate the probabilities was given before when I summarized the methodology.
– Indeed the S3 selection limit at 19:41 will probably lead to a drop in probability when getting close to S40 @ 00:19. In the latest results I applied the S3 limit in all figures shown. There have been previous discussions on the most southerly latitude that could have been achieved at 19:41. Perhaps it is good to refresh that discussion.
– To check I derived again the figs 7b and 7c. The only difference between the figures is the different BTO and BFO error corrections applied. The script in both cases is identical. I got the same figures again. Fig. 7c is the more relevant one, from a methodology point of view.
The fact that I don’t normalize is perhaps useful: I’m thinking to apply the whole procedure to artificial BTO and BFO datasets generated using emulated flight paths. Perhaps it is possible to compare straight paths with paths that include turns. But it’s for later, after the proper implementation of fuel limits is settled.
@Niels,
It is now a fresh morning.
Having freshly reviewed my previous post/comment of November 14, 2024 at 7:23 am
I find that one major intended explanation (or expansion) was lost during writing/editing.
(along with some typo’s AND a missing “been”).
Above (in the previous comment) was written:
“The elimination of the paths just discussed (on the southern side of the peak) will have affected where the general peak is. This will move NORTH for each southern candidate eliminated. Hence the peak probability of the overall sample moves north.”
This clearly needs further expansion/explanation/correction.
Let’s first go back to your basics.
In your “Path_fitting_Nov_2024_v1b” you took 20 million random path samples and for each candidate, or sample, you calculated the probability of being a path fitting the BTO/BFO data for the last flight of 9M-MRO on Flight MH370. Then each candidate path was allocated to the appropriate “Latitude Bin” based on the estimated location of the aircraft at a declared time of fuel exhaustion, for which you declared 00:17:30 UTC on the 8th March 2014. The “weight” added to the bin for each candidate path was not unitary but weighted by the calculated probability. Hence each and every contribution to any specific bin was always less than unity.
AND, candidates with a calculated probability less than 1e-4 were not allocated to any bin.
After all candidates have been accounted for the total “weight” in each bin can be charted on a latitude based chart.
Note that here I may refer to “post processing” as my summary of the process to be described.
Also note that, for this explanation only, I am describing candidates for which fuel calculations determine that for the candidate path the aircraft would have run out of fuel before completing the candidate flight path. (More later)
Referring to my previous post/comment of November 14, 2024 at 7:23 am
The “post processing” may be described as such:
For example: For the southern side of the peak it is quite understandable that for some otherwise possible, or probable, paths the fuel availability limit will be reached before the subject path can be completed. Thus the candidate is not a valid candidate. It is necessary for the “weight” allocated to the respective latitude bin to be removed. The total number of samples contributing to the final determination is reduced by one.
And I subsequently wrote “The elimination of the paths just discussed (on the southern side of the peak) will have affected where the general peak is. This will move NORTH for each southern candidate eliminated. Hence the peak probability of the overall sample moves north.”
This as written, is misleading, and deserves expansion/explanation/correction.
As the process of elimination of paths proceeds there will be more and more samples (candidate paths) for which their weight be removed from the appropriate bin and the sample count will reduce by one. The failing in the above misleading description is that as more and more candidate paths are eliminated the affected bins will be several or many, but (as I see it) it is likely that bins “South” of original peak will be more affected than bins “North” of the original peak, and thus the peak will move northwards in the process. This is where the post processing concerns the max fuel limit.
I also SHOULD have written: “The removal of ANY candidate apparently improves the probability of any other candidate remaining.” Perhaps this should be rephrased that the relative probability of any remaining candidate will be altered (even if only slightly) by the removal of any other candidate. (I may have “drifted off” into a world where I was considering unweighted candidacies when I wrote that.)
P.S. @Niels, I have just checked and seen you have replied above.
I am posting this now and will read your comment after and separately
@Niels,
As you will have now seen I had decided to summarize the bare basics of your Methodology simply for my own ends to make sure I had it right. Hopefully I have. You may need to correct. Specifically, I wrote just above (November 14, 2024 at 6:05 pm) that the completion of the candidate flight paths were determined at the time of fuel exhaustion, but your mention of “getting close to S40 @ 00:19” reminds me I got that wrong as well … ! Of course the BTO/BFO data itself ends at 00:19:30 UTC, (subsequently abbreviated by you as 00:19; refer your MonteCarlo_v10 of 24/3/2024).
Above, you write: “There have been previous discussions on the most southerly latitude that could have been achieved at 19:41. Perhaps it is good to refresh that discussion.” I doubt that discussion needs much further mention here, at this time. Your existing data and analysis tend to indicate that 36 degrees south may have been achievable and 38 would have been unlikely. Methinks this is in general agreement with many others. Whether you will achieve any improvement in additional southern constraint (and/or any others) by proceeding with your fuel analysis, if you continue, we do not yet know.
Re Fig 7c. Thanks.
I don’t see that anyone asked, what are the three orange, one dark red and one purple dot on the map?
@George G
Thanks again George. My apologies for not going in all detail; I’m in a really busy period at my job. I think one main point is to qualitatively understand the shift in the peaks when removing the fuel selection criterion. I agree it seems that more paths are removed towards the south when the fuel limit is applied, and that it impacts the appearance of the latitude probability curve. To fully understand/explain it is not that easy; at least I like to be careful there. One factor can be that in my fuel model there is a trend for less fuel available at 19:41 when going further south in 19:41 position; another factor can be the wind field.
@All
Interview from Ireland worth listening to:
https://www.rte.ie/radio/radio1/clips/22460520/
I do not understand how WSPR got through a Peer review.
@Niels
My gut feeling is that it is mostly due to wind fields.
@CMR
I just found an interesting 9M-MRO photo collection at various airports including LAX on 28-Feb-2014. Interesting to focus on flaperon behavior during different take-offs/landings.
I believe we have the records, detailed archives on historical flights by 9MMRO as well as historical flights by ZS. I do not recall any ZS USA flights, but please do not quote me on that. Some of that info is in FI and SIR reports.
https://www.raischstudios.com/malaysia-airlines-flight-370/2014/3/30/wqupf9xsv2emht1x020isa25b2wlg0
@Marijan,
The three orange dots are historical. They are three CSIRO priorities. They come from 2017-2018. They were described in the reference as “Within the area searched so far, there are three warm spots that CSIRO has designated as priorities, based on satellite images of objects that could have been MH370 debris, and from drift models that estimated the points of impact from the location of these objects”. Reference: https://mh370.radiantphysics.com/2018/02/16/mh370-search-update-feb-16-2018/ by Victor Iannello.
I cannot find any other reference in those that I kept from that time.
Perhaps someone else can help with the purple spot.
If you mean by “one dark red” dot the small red square labelled as “IG Hotspot”, then this is the Last Estimated Position (LEP) resulting from the investigation and report “The Final Resting Place of MH370” by Ulich, Godfrey, Iannello and Banks (UGIB) in March 2020. Reference: https://mh370.radiantphysics.com/2020/03/09/new-report-released-for-mh370-search/
@Viking
You “do not understand how WSPR got through a Peer review.”
If you are referring to the eleven words in the time 2:16 through 2:21 starting with “has now been” and ending with “experts”, then please take this with a grain of salt (or even the whole Indian Ocean’s worth of salt if you wish).
Refer previous discussion above, November 11/12.
@George G
I was referring to those 11 words. I like your comment about the whole Indian Ocean…..
@All
I’m working on the error budget for available fuel at 19:41 UTC.I hope someone can help with the following question: What is the typical accuracy of the aircraft gross weight as indicated in for example the take-off report and the climb report? How is it measured and/or calculated? How accurate would be the aircraft zero fuel weight estimate, made before take-off. How is it measured and/or calculated?
I’m trying to estimate the possible error in the 43,800 kg fuel indicated for 17:06:43, which is just before the last ACARS report (“UGIB” p. 19).
The FQIS fuel indicator on Boeing airplanes is accurate to within plus or minus 1% at full scale and 0 to 0.5% below 10% for the center tank.
In 2020, the FAA issued a rule requiring operators to perform a validation process for the FQIS on Boeing 777s with certain fuel tank capacities. The FAA took action after identifying at least 25 in-service events related to inaccurate fuel quantity information. In some cases, crews continued to fly despite a misstated fuel load, leading to diversions.
@alsm
Thank you! So assuming a calibrated system 1% possible error on the last measured fuel quantity could be reasonable. However, probably we don’t know how well MAS FQIS were calibrated around 2014.
@Niels
@ALSM
The FQIS quantity indication is accurate to within ±1% with full main tanks. Under FAA regulations, the system is calibrated to read zero when the fuel remaining is unusable (14 CFR § 25.1337). I don’t know the exact relationship, but I think it’s fair to say the accuracy improves as the fuel remaining decreases.
In the MH370 case, the aircraft departed with 49,100 kg, all of which was loaded in the left and right main tanks. The centre wing tank should have been empty. At the last ACARS position report, the recorded fuel remaining was 43,800 kg. Assuming a linear relationship between the FQIS accuracy and fuel remaining, the last recorded quantity should have been within about ±0.8% of the actual fuel remaining (Full Main tanks = 58,200 kg).
Regarding your other questions about zero fuel weight calculations, etc, the following document provides some insight:
FAA AC 120-27F Aircraft Weight and Balance Control
@George
Thanks George! I figured out what two remaining dots are. The small purple dot and the dark red one just south of it come from the Blelly-Marchand paper from 2022 (https://www.mh370-caption.net/wp-content/uploads/Blelly-Analysis-of-Flight-MH370.pdf). The purple one refers to “minimum crash zone” and the red one to “maximum crash zone” calculated for two indicated airspeeds along the presumed airplane course. Interestingly, paper also contains the OI proposed search zone from 2022, which was refined in the meantime.
@Andrew
Thank you. Given that FAA regulations proper calibration would then be essential, and relative errors near empty tanks could become large if not.
In addition for the linearity of the relation: would you know which type of sensor system is used?
@alsm
Would you have a link to the 2020 FAA rule? Perhaps it helps to understand calibration issues?
The FAA’s 2020 AD refers to the Center Wing Tank (CWT). 9M-MRO’s fuel load to operate the MH370 KUL-PEK service did not require any fuel to be loaded into the center wing tank. The aircraft fuel system draws fuel from the center tank, if any is available, before drawing from the main tanks.
From the AD:
‘The FAA has received reports involving discrepancies between the airplane FQIS, specifically the CWT fuel quantity, and the refueling truck uploaded fuel amount, followed by a FUEL DISAGREE EICAS message at an early stage of flight (e.g., within 3 hours after take off), and/or with an INSUFFICIENT FUEL EICAS message.’
Niels,
The FQIS appears to be self-calibrating. The FQIS involves 78 tank units measuring fuel height in an integrated still well (28 tank units per main tank and 20 in the CWT).
Ten of the tank units have calibrating targets within their still well. The tank unit measures the fuel level distance and the calibration target distance (by reflection time of ultrasonic pulse off fuel surface and off the target).
Adding…
The referenced B777 AD describes a procedure that inserts a delay in the refuelling check process that involves closing the refuelling panel door and then reopening, thus causing the FQIS display to reset. More a settling process after fuel loading than calibration.
@Niels
Adding to Don’s comments, the B777 FQIS has an ultrasonic gauging system. It has a number of tank units that measure fuel height by reflecting sound pulses off the surface of the fuel. The pulse travels through the fuel inside the tank unit to the fuel surface, then back to the receiver. The travel time of each pulse is used to determine the fuel height. As Don mentioned, some of the tank units have targets at a fixed height. The sound pulses reflected off the targets allow the system to measure the velocity of sound (VOS) in the fuel.
The system also has one densitometer in each tank. The fuel density and fuel volume measured by the tank units are used to determine the fuel weight in each tank.
Don also mentioned that the 2020 AD only affected the CWT and was not relevant to MH370. The issue was caused by a design flaw where the FQIS used a ‘stale’ VOS during CWT refuelling. The AD required operators to cycle the refuelling station door after CWT refuelling/defuelling. That procedure reset the system, forcing it to use a ‘fresh’ VOS for the CWT fuel quantity calculation. A subsequent AD issued in 2022 fixed the problem with a software update to the fuel quantity processor unit.
@Andrew, Don Thompson
Thank you! That clarifies a lot. Regarding the ten tank units that have a calibrating target included, how would those be distributed over the units composing the main tanks?
@Niels
There are three tank units with targets (aka ‘velocimeters’) in each main tank and four in the CWT.
@Niels,
Will send you an explanatory image directly.
Don
Based on an estimated cruise speed of 683 mph & assuming Mh370 was flying in a straight line then my estimated location at 19.41pm is around 3°13’51″S 92°53’58″E
What’s everyone else estimating?
@Michael John
Ah ha, the vexed question of the 19:41:00 intercept raises it’s ugly head yet again.
As a general comment, I would say at the outset, that the unresolved problem of the ASTB/DSTG’s “geographic inconsistency” with the 19:41 arc intercept, is most likely a direct by-product of the “unverified and unverifiable Malaysian Military Radar Data” (which by necessity produces the FMT north of Ache) and that is, in my frank opinion, the root cause of our inability to find MH370. But more importantly, this unverifiable radar data has produced a location that also causes significant difficulties interpreting the First Arc log on BFO’s. As we all know, many theories have been examined over the years, for example, the crystal oscillator oven warm up effects, the proposed ‘right offset of 15Nm from N571’ for possible traffic avoidance, and even the more recent Air France Captain’s scenario. None of them have produced a definitive outcome. More to follow on this matter, see below.
I take a purely practical approach to solving MH370, as a pilot, as a navigator, but more particularly, as a mission planned. We have to accept that MH370 was the Captain’s last flight, that he knew it was going to be his last flight, that he had been preparing for it for a long time, many months in fact, and that he had planned it in meticulous detail, with nothing left to chance.
The official Malaysian government narrative is the direct opposite of all of that. Furthermore, as I have explained before, I do not accept ‘the Lido Slide’ as being the true path of MH370, certainly not past Pulau Perak Island anyway.
My track to the 2nd (19:41:00utc) arc is based on my theory that MH370 DID NOT proceed all the way up the Malacca Strait, but rather, made a LEFT turn off the Vampi Track just South-East of Pulau Perak Island, in other words, just after the end of the civil radar data. (I totally dismiss the remainder of the radar data, the so-called military radar data to Vampi and beyond, as unverifiable)
Why I believe Zahari had to cross Ache can be simplified down to one thing, fuel. I have explained it all in detail in other places before, but the essential point is that his original plan had been made off a flight where he had significantly more fuel than MH370 had. He absolutely had to “cut his track miles” to get to his planned destination, so he had to “escape to the northern Indian Ocean” much earlier than his original plan. Therefore, he could not go up to the north of Ache, he had to cut across southern Ache.
My route from the end of the civil radar data is as follows: (generated from Barry Martin’s spreadsheet).
(a) Turn left just SE of Pulau Perak Island (from heading 288T to 240T) towards waypoint TASEK
(b) Thence (to OVERFLY ACHE) – Track DCT waypoint IVRAR (formerly Nagan – just north abeam of WITC)
(c) Within 10Nm of reaching IVRAR – Repower All Electrical Systems
(d) Go “Feet Wet” (cross the coast – out into the ocean – never to see land again)
(e) Thence track DCT waypoint MABIX
(f) Thence – Cross the 1st arc – continue
(g) Thence – (within 5Nm of waypoint MABIX) commence the FMT
(h) Complete the FMT at 18:39:30 at (94.7030 East / 3.0680 North) at FL400, Heading 192.8963 degrees True, GS 467.52kn
(i) 1st Satcall (do not answer) at 18:40:00 at (94.6891 East / 3.0046 North) at FL400, Heading 191.8956 degrees True, GS 467.28kn
(j) At 19:01:30utc pass 27.252Nm East ABEAM of waypoint ISBIX (closest approach to ISBIX) (94.1188 East / 0.2782 North) at FL400, GS 467.2kn
(k) Cross Equator (19:03:45utc) at (94.0591 East / 0.0078 South) at FL400, Heading 191.8791 degrees True, GS 467.71kn
(l) Cross 2nd arc (19:41:00utc) at (93.0666 East / 4.7467 South) at FL400, Heading 191.9203 degrees True, GS 467.71kn
In other words, my 19:41:00 latitude is 4.7467 SOUTH, which is way further south than most other people’s calculated latitude [and I cross the 7th arc (00:19:30utc) at (86.2442 East / 39.226 South)].
To return to the beginning (knowing that most will dismiss what follows, but, tin hat on).
My suggestion is that we should concentrate our attention on resolving ‘the elephant in the room’, the ASTB/DSTG’s “geographic inconsistency”, something that the ATSB seemed quite happy to just leave ‘swinging in the breeze’, with apparently no real attempt having been made to ever resolve it.
It has to be resolved, and to that end, I propose that ‘the number crunching experts’ should re-examine the circa 18:25utc log on BFO’s, but, this time, on the assumption that they occurred in a different geographic location to that currently accepted, specifically, my location, near WITC.
To begin, it is readily apparent by simple inspection alone, that one thing is very obvious.
The line of sight to the Inmarsat Satellite from Sanob is about 260 degrees true, and from WITC it is about 264 degrees true.
At Sanob, the aircraft is generally considered to be heading about 290 degrees true, but at WITC, my heading is 240 degrees true.
This produces an obvious and significant difference in the aircraft’s relative heading to the satellite for the two locations.
Up near Sanob, the aircraft heading (290 – 260) is about 30 degrees TO THE RIGHT of the line of site to the satellite.
Down near WITC, the aircraft heading (264 – 240) is about 24 degrees TO THE LEFT of the line of site to the satellite.
I have a ‘gut feeling’ that this difference in relative headings is significant, and that it could be ‘the smoking gun’ that we need to solve ‘the riddle of the circa 18:25 Log-On BFO’s, and the whole case.
Number crunchers, on your marks, get set, GO.
@ventus45
You are indeed opening Pandora’s box with an ugly head inside. I am not going to write a long contribution this time, but I would like to comment on 3 important details:
1) I qualitatively agree with all you are saying, but quantitatively I disagree strongly. The reason is that I think you underestimate Shah and the quality of the radar data. I think he was even smarter than you assume, and the radar data is indeed inaccurate, but not useless.
2) If you look at the Inmarsat data together with the end position for the radar data, this leaves two options for a northern route and a southern route after radar contact ended. After 19:41, only southern routes are possible due to BFO. In-between routes are only possible if they include a lengthy period of flying in circles. I am sure someone would have noticed that. In particular since the circling would have had to take place near a naval base belonging to India. It would have been the equivalent of sticking your finger in a wasps nest. Shah was too clever to do that.
3) You are obviously choosing the pure southern route. Mathematically that is ok, but the question is if it is the correct solution? Several things speak against it. To mention two, there are no contrails (should have been since such a route passes through a small area near Aceh with excellent contrail conditions). In addition, the BFO value at 19:41 speaks against it.
I personally prefer the northern route. That means an extremely northern position at 19:41, a U-turn soon after, and a final destination near Christmas Island. This agrees with infrared contrail evidence from 3 independent satellites from France, Japan and China, and with the BFO at 19:41. As a pilot, you should listen to contrail evidence.
@Viking
…”infrared contrail evidence from 3 independent satellites from France, Japan and China….”
News to me. I thought that all the previous ‘contrail’ candidates had been debunked moons ago.
@ventus45
A few small details have been debunked, but not the stronger part of the evidence. Take a look at the last version of my paper. You can see the deviation for the BFO at 19:41 illustrated in figure 2.
The contrail decending below the clouds near Aceh is in figure 11.
The Mushroom cloud near the 7th arc is shown in figure 14.
@Viking
“…. the last version of my paper.”
Link please ?
@ventus45,
Your comment of November 19, 2024 at 9:16 pm
In your comment you refer to: QUOTE: the ASTB/DSTG’s “geographic inconsistency” with the 19:41 arc intercept END Quote.
As time goes by my memory may be getting slower. And having been forced to replace my computer with another, researching through my saved references becomes slightly more onerous. Hence would you please kindly provide reference to your source concerning the “geographic inconsistency” that you mention.
@George G
It was in either the Bayesian Report or the ATSB Search Report, I forget which, but off the top of my head, but I think it was in the Bayesian.
Ventus45 GeorgeG
https://www.atsb.gov.au/sites/default/files/media/5733804/Bayesian_Methods_MH370_Search_3Dec2015.pdf
Is it referring to pages 18 and 19?
@ventus45
http://arxiv.org/abs/1811.09315
@Joseph Coleman,
Thank you. You ask: “Is it referring to pages 18 and 19?” I don’t think so. These pages of the DSTG Bayesian Methods report/book are basically discussing the choices available for the initiation point or state for the subsequent statistical analysis. Also see later.
@ventus 45,
I don’t think the ATSB Operational Search report (AE-2014-054) discussed any matters approaching such as you may have been discussing. I have not found any specific reference in the DSTG report, but perhaps you may have been referencing to where in that report, whilst discussing BFO measurements and errors, is written: “The residual error is clearly not zero-mean, and the mean
varies with time. Substantial effort was made to characterise this structured bias. It
was found to have a geographic dependency but it has not been possible to determine
a quantitative function to compensate for this change in bias.” Reference: Page 28 of the DSTG report/book for which @Joseph Coleman provided a link. I would not term this an “inconsistency”, rather a consistent small contribution to the data, the reason for which had not yet been found.
@Viking,
In your report/paper Ref: 1811.09315 you discuss that the BTO values “look like they are located on a straight line with constant speed” whilst you consider the BFO values also do so (with less precision) and with the exception of the BFO at 19:41. You then go on to discuss four ways that the apparent deviation at 19:41 may have occurred. To return to the DSTG document (link provided by @Joseph Coleman), reference may be made to Fig. 10.7 (on Page 91). Here both BTO and BFO data are plotted. The BFO results are for 18:28 through 00:10 time points. The result at 19:41 may be compared to the rest.
@ventus45,
In summary, I do not know where your quote “geographic inconsistency” comes from.
@George G, Viking, Joseph
Yes, Bayesian Methods in the Search for MH370 – page 28 – under Fig5.4 – where they were talking about a Mumbai to KL flight, that’s it, I remember now that I see it again. I’m not sure how I associated that with 19:41 at the time, but I did, and it stuck in my head. I will retract the idea for now.
In general, with the passage of time, the reported radar position data out to 1822 seems valid. It matches very well with Inmarsat Arc1, and it matches with many proposed flight paths that a B777 flight simulator could fly south to the SIO. It is the reported truth, and keep in mind that means that USA, Boeing, NTSB, Australia etc. seems to agree with it (and they had more access to secret data than we do as public). It makes good sense (and is consistent with pilot’s sim cases) that MH370 may have have avoided Indonesian FIR (until later) and experience indicates no apparent fundamental flaws in the radar + Inmarsat Arcs.
Therefore the simplest form of the “official narrative”, that the radar and Inmarsat data are correct evidence and point to a flight path to the SIO over Arc7 somewhere, as verified by debris finds, stands the test of time, and would be hard to overturn without some equally solid evidence to the contrary.
@TBill: I’d say the validity of the radar targets between 18:02 and 18:22 are likely but not certain for the following reasons:
1. The radar targets after 18:02 shown in the Lido Hotel image were never supplied to the ATSB, other than the final point at 18:22.
2. Malaysia has never officially stated that the radar targets shown in the Lido image were valid, despite showing the Chinese families the data.
3. To match the BTO data at 18:25-18:28, after the last radar target at 18:22, a maneuver is required, e.g., the right offset from N571. For this reason, the DSTG chose 18:02 for the starting point of the reconstructed paths and ignored the target at 18:22. That’s not to say a maneuver did not occur.
Malaysia could put these questions to bed by simply releasing the military radar data. Whatever security reasons there were in 2014 for not releasing the data are irrelevant at this time. Instead, 10+ years later, we are still speculating.
@Victor,
Of interest, the DTSG actually used the radar point preceding that at 18:02 as their initiation point. They actually chose 18:01. Presumably they considered the existence of the data from the next point (namely 18:02) together with that from the prior data points as providing more confidence in, or information about, the data for the 18:01 point than was provided for/by the 18:02 data point, itself. Reference: “Instead, a prior was defined at 18:01 at the penultimate radar point” from Page 19 of the linked document provided by Joseph Coleman, November 20, 2024 at 6:58 pm.
Actually, self proof reading actually does not actually work properly, until “Submit” has actually been actioned. !
@George G: On page 19 of the DSTG report, it says:
The radar data contains regular estimates of latitude, longitude and altitude at 10 s intervals from 16:42:27 to 18:01:49. A single additional latitude and longitude position was reported at 18:22:12. Figure 4.1 shows the radar data overlaid on a map.
It’s clear that the last radar target the DSTG had was 18:22:12, and the penultimate target was 18:01:49. I rounded up the time from 18:01:49 to 18:02. In the text you cited, the DSTG rounded down the time to 18:01. It’s the same target.
@Victor,
Noted. Fully Agreed.
Yes, Clearly referencing the one and the same data point.
Just goes to show what one can retrieve (or I did retrieve) from a partial read whilst searching for and reading a rather specific section, and not properly reading the whole section or document. (!)
P.S. To reduce confusion for any other readers, Page 19 of the (final) DSTG report/book is the same page as Page 17 of the previously linked Pre-Publication Draft.
@Victor. No doubt you have considered that the “Lido” data and leading up to 1822 “last known position” may come from a non-MY source, shared on a non-quotable/attributable basis? Hence their inability to release it or explain it in any detail. Speculating…
In its final report from Oct 3, 2017, the ATSB stated the following on p. 10:
“Radar data shows the aircraft then headed to the northwest, eventually aligning with published air route N571 from IFR waypoint VAMPI. The validity of this section of the radar data was verified using the track of a commercial flight that followed N571 about 33 NM behind MH370.”
This description matches the first part of the Lido Hotel portion of the route. The commercial flight following MH370 was likely UAE 343. Indeed, if it were possible to extract the primary radar data alone for UAE 343 in the same form as that for MH370, they would provide a valuable cross-check.
@sk999 thanks for highlighting that snsnippet. Even if flightaware data for UAE343 were available it would provide a useful indirect corroboration of time/position based on that “33NM” separation.
https://www.flightaware.com/live/flight/UAE343/history/20140307/1710Z/WMKK/OMDB/tracklog
@sk999, Paul Smithson, Victor Iannello, George G
Re UAE343, so far as I remember, it was only speculated that it might have been UAE343, but it was never confirmed to be UAE343 ?
{(Edit) I see that Paul has just put up the link to UAE343}
But, to test the 33Nm in-trail theory, we would need all of the adsb data for all of the other flights that were in the area between Penang and Sanob for the time period 17:50 to 18:30 to be able to positively identify it as the flight in question (and not possibly some other flight).
{(Edit) @Paul, could the data for all the other flights still be obtained from any of the adsb tracker web site archives ?}
Re the primary radar data:-
(1) can it be agreed by all, that the ATSB’s choice of the second last Butterworth PSR hit at 18:01:49.1 can be accepted as “.. the last ‘verified’ and ‘agreed’ PSR data point ..” and that we agree to standardise on the use of that point as “the start point” for all analysis going forward (so that everyone stays ‘on the same page’) ?
(2) can it be agreed by all that the ATSB’s choice to ‘ignore’ the 18:22:12 hit (per Victor’s point 3 above: “… the DSTG chose 18:02 (meaning 18:01:49.1) for the starting point of the reconstructed paths and ignored the target at 18:22….”) is in itself effectively a ‘vote of no confidence’ in that hit by the ATSB themselves ?
(3) if (2) is agreed, is there any “compelling” logical reason (or reasons) to disqualify consideration of any other physically possible (i.e. within the B777 performance limits) track to the first arc, other than that generated by the ATSB (and hitherto effectively accepted as ‘gospel’) ?
@Ventus45
EK343 has been brought up many times before. Looking at flightaware anything after 02:07 for this particular flight up the Straights seems predicted at the time. Was there any Secondry radar coverage for EK343 after 02:07 up to 18:2212 from Malaysia perspective. How was validation done, was it Malaysia or Other Counties may have seen or recorded EK343 on Secondry. How would Malay military radar distinguish between the two if let’s say Malay Secondry radar was out of range any time after 0207 (predicted EK343).
https://mh370.radiantphysics.com/2017/10/12/simulator-data-from-computer-of-mh370-captain-part-1/#comment-8310
UAE343 revisited. I have translated the flightaware tracklog data to a .kmz file with times in UTC that opens in Google Earth here https://shorturl.at/qrCEA
The “real” position data had the aircraft tracking directly towards VAMPI. The “estimated” data is a straight line from last ADSB toward destination. Assuming that UAE343 did join N571 at VAMPI and followed it thereafter towards IGOGU at the groundspeed last recorded (501kt) I get the following projections of distance and time:
From 18:07:22 to VAMPI: distance 80.1NM, time 00:09:36, ETA 18:16:58
From VAMPI to MEKAR: distance 68.2NM, time 00:08:10, ETA 18:25:08
From MEKAR to Arc 1 crossing: distance 45.4, time 00:05:26, ETA 18:30:34
If MH370 followed N571 beyond Arc 1 to MEKAR then it was only ahead by 00:02:19, or a distance of 19.3NM. That’s quite a big difference from the “33NM” separation quoted by sk999 above.
Interestingly, the difference between the two estimates (33NM = 00:03:57, 19.3NM = 00:02:19) is 98 seconds. Which matches the mysterious offset between MH370 ETA at Arc 1 per radar data and its actual time at Arc 1 per Inmarsat log. Perhaps there was a timebase offset between radar data and satcom log?
correction: If MH370 followed N571 beyond MEKAR to Arc 1… [not vice versa]
I’ve tracked down the flightradar screenshot of EK343 projected position at 1822, viz 6.37N, 96.94E https://imgur.com/APHqKTz as posted by “Beggarman” on [page 25, 2014-03-22, 10:25] https://forum.flightradar24.com/forum/aviation-forums/general-aviation/7298-malaysia-airlines-flight-goes-missing-en-route-to-china-flight-mh370/page25?view=stream
This is consistent with a distance of ~33NM if MH370 was 10NM beyond MEKAR at this point. I don’t know whether Flightradar has access to any realtime position info for EK343 or if it is an extrapolation based on flightplan.
And finally….with reference to my statement above “Assuming that UAE343 did join N571 at VAMPI”
With a skyvector overlay, it is clear that the UAE343 track during ADSB coverage places the aircraft directly on N571 heading towards VAMPI. So it was already flying along N571 and it is only the “estimated” data beyond 18:07:22 that departs from N571 in a straight line towards destination.
@Victor
On your point2: “Malaysia has never officially stated that the radar targets shown in the Lido image were valid, despite showing the Chinese families the data.”
The 18:22 radar point is officially reported in the FI and “final” SIR reports. I would say that is official stated, admittedly we would like more openness from Malaysia on their military radar details, gaps etc.
@TBill: The last point at 18:22 was part of the data set supplied to the ATSB, but the interim points between 18:02 and 18:22 were never officially acknowledged. We would have more confidence in the 18:22 point if the interim points shown in the Lido Hotel image were released.
@ventus45,
Re your question/s concerning “ignoring” the 18:22:12 primary radar point:
Well, that depends. See all ‘here’ following your posting of your question/s and what follows here in this post/comment. Confusion reigns.
My initial response to your question/s was to be that due to the range and being the sole (known) point at that large range the estimated position would have to be considered ‘likely to have rather poor accuracy’ in the terms of the DSTG. And I would have qualified that by offering the opinion that it could not be totally ignored, even if the sphere of uncertainty might be large.
The discussion I offer following (in two separate comments/posts) somewhat confuses my initial response.
@Joseph Coleman,
You ask “How would Malay military radar distinguish between the two” ?
How indeed I am beginning to ask in general.
@Paul Smithson,
With your Google Earth view and your SkyVector overlay, how far away from the radar installation at Butterworth might two aircraft following the same path have to be to be indistinguishable on the Butterworth radar display, or replay? For the purposes here let’s consider one aircraft being (say) approximately 33Nm behind the other. Perhaps we also should consider the hypothetical (even conjectural) situation where the leading aircraft might be flying along a path offset slightly south of the path being flown by the second. ?
If two aircraft became indistinguishable on a radar display the question arises for just how long this situation might last. Being totally unfamiliar with radar displays I would have to ask those familiar with radar systems just how two returns along (to and from) the same direction are handled.
@Paul Smithson (Gerore G)
If you look at Skyvector closely, you can see that N571 is defined all the way from waypoint GUNIP to waypoint RUKOR (which is only 23Nm from OMDB) (Dubai International). (See link below)
As your adsb KMZ data file clearly shows (thanks very much for that effort) EK343 had departed WMKK and overflew Gunip and was heading for Vampi (and not Tasek on B466), so it seems reasonable (almost certain) to assume that EK343 had ACTUALLY ALRAEDY JOINED N571 AT GUNIP (so was not “joining” at Vampi) but was rather already on N571 and simply tracking direct to the next waypoint, which was Vampi. I think it is reasonable to assume that EK343 would remain on N571 all the way to Dubai.
Therefore, as George G has indicated at 6:23pm, the question of distinguishing between EK343 and MH370 in the area from Vampi to the first arc becomes more critical than ever, since your analysis so far indicates that they would have been much closer together than previously thought.
https://skyvector.com/?ll=4.276662935171122,100.77355957416002&chart=304&zoom=3&fpl=%20WMKK%20AGOSA%20R467%20GOBAS%20R467%20GUNIP%20N571%20VAMPI%20N571%20MEKAR%20N571%20NILAM%20N571%20IGOGU%20N571%20RUPTI%20N571%20LAGOG%20N571%20BIKEN%20N571%20IDASO%20N571%20VATMO%20N571%20GURAS%20RINTO%20VABDI%20PADBI%20VIRAM%20G208%20DORAM%20G208%20BOGAT%20G208%20AGELA%20N571%20ENTUG%20N571%20BEKUT%20N571%20SUGID%20N571%20DOGET%20N571%20PARAR%20KIPOL%20RAGMA%20SODEB%20VUSET%20N571%20TOVDI%20N571%20KIROP%20N571%20ASNIB%20N571%20MENSA%20N571%20LUBAT%20N571%20ENEGA%20N571%20RUKOR%20OMDB
@George G. Radar distance resolution is in fractions of NM while azimuth resolution might be as much as a degree. Given the distance resolution there’s no way EK343 and MH370 could be confused/conflated with this separation even on similar radial to radar head. The reason for my interest in this is the timing anomaly of circa 100s between “radar” and Arc1 arrival time. The same problem exists with 1802 as with 1822. And since 1802 Butterworth PSR is preceded by a contiguous stream of points I dont feel that the last Butterworth position is in question. The more I gnaw at it the more I am inclined to query posdibility of timebase offset on the satellite data after power-up.
@Ventus “EK343 had ACTUALLY ALRAEDY JOINED N571 AT GUNIP (so was not “joining” at Vampi) but was rather already on N571 and simply tracking direct to the next waypoint, which was Vampi.I think it is reasonable to assume that EK343…” Fully agree, hence my follow-up clarification.
In the Malaysian Safety Investigation Report (SIR) Issued on 02 July 2018 the subject of Malaysian Military Radar is addressed in Section 1.1.3 starting on Page 3 of that report. This is part of Section 1, Factual Information, of the report.
Observation: It is noted in the SIR that the “the altitude and speed extracted from the data are subjected to inherent error”. This is very evident from some of the data reported.
Data which may be extracted from the report includes:
At 1752:31 UTC the “blip” was observed to be at 10 nm south of Penang Island;
At 1801:59 UTC [0201:59 MYT] the data showed the “blip”;
At 1803:09 UTC the “blip” disappeared, only to reappear at 1815:25 UTC until 1822:12 UTC, about 195 nm from Butterworth; and
The tracking by the Military continued as the “blip” was observed to be heading towards waypoint MEKAR on Airway N571 when it finally disappeared at 1822:12 UTC, 10 nm after waypoint MEKAR.
Here, it is interpreted that “when it finally disappeared at 1822:12 UTC” means the last radar contact was at 1822:12. This interpretation is simply in context that the DSTG (and subsequently others) have referred to it as “The final primary radar return”.
The above summary leads to a (large?) degree of confusion. As far as the DSTG was concerned, “The radar data contains regular estimates of latitude, longitude and altitude at 10 s intervals from 16:42:27 to 18:01:49. A single additional latitude and longitude position was reported at 18:22:12.” Reference: Page 19 of the (final) DSTG report/book and Page 17 of the Pre-Pub Draft per the link provided by Joseph Coleman on November 20, 2024 at 6:58 pm.
Was “18:22:12” the only data point in the range “1815:25 UTC until 1822:12 UTC “ for which specific details were released externally by Malaysia ?
Let’s return to 1752:31 UTC where per the SIR the “blip” was observed to be at 10 nm south of Penang Island.
For this UTC time the FlightAware data for EK343 shows that it was still within ADS-B range, hence it’s position fairly well known. For those with Google Earth (I’ve temporarily lost mine) and/or SkyVector, etc, a check of the relative positions for the two aircraft may be interesting.
@Paul Smithson, may be another timing check concerning your “mysterious offset” ?
@ventus 45, returning to your question/s. I don’t know.
@ventus45, @Paul Smithson, Thanks.
Clearly I posted the above before I saw your recent comments.
Paul, your information concerning radar distance resolution is informative.
That leads me to quite another gnawing “problem” for which I will let you gnaw too in addition to that upon which you are already gnawing.
@ventus45, @Paul Smithson, @Victor, @SK999
@Ventus, like a dog with a bone, I had to gnaw further.
The outcome is that I see no reason to ignore the reported radar data at 18:22:12. This is particularly as it would appear to be part of a set (as per the Malaysian SIR). Due to the long radar range some uncertainty might be assumed, and allowances made. As a general indicator, it would seem appropriate to retain it.
Separately, With regards to the presence of EK343 in the region, as Paul Smithson has explained “there’s no way EK343 and MH370 could be confused/conflated”, and this is particularly evident early on in the relevant time period. SK999 had plotted the path and timing of EK343 some time ago (February 2016). It would appear that at (approx) 1800 UTC the two aircraft were in the order of one degree latitude apart.
@Paul, my apologies. You can go back to gnawing on that timebase bone, without any other gnawing distraction.
Re: EK343
Presumably NTSB/Malaysia et al would have asked Emirates for the EK343 aircraft’s detailed data recording of location for the flight, assuming they needed that level of detail.
TBill,
RAT-31DL has secondary radar – that should be sufficient to provide a cross-check of primary radar.
@Anyone https://www.radartutorial.eu/19.kartei/02.surv/karte012.en.html#:~:text=The%20detection%20envelope%20a%20360,up%20to%20250%20nautical%20miles. An Interesting read for what seems the Mobile Version.
“ The RAT-31DL/M includes a fully integrated SIR-M secondary monopulse radar with a maximum range up to 250 nautical miles”
@All
Interesting reference document.
https://www.icao.int/apac/documents/edocs/cns/gmst_technology.pdf
Putin is targeting planes:
• MH17
• Prigozhin
• possibly MH370 if the spoofing theories pan out
• 19 July 2024 in Leipzig:
« If not for a delay in a connecting flight, the incendiary bomb would likely have burst into flames in the belly of a plane flying high above the European Union. Instead, it ignited on the ground in Germany’s Leipzig airport, setting fire to a DHL air freight container. Western intelligence officials believe the attack, which took place in July, was a trial run by Russian agents who planned to place similar bombs on flights to the United States. »
https://www.politico.eu/article/europe-russia-hybrid-war-vladimir-putin-germany-cyberattacks-election-interference/
https://www.bbc.com/news/articles/c07912lxx33o
• It was a trial run for TODAY:
+++ cargo plane crashes near Vilnius airport +++
Yes, you guessed right: It was again a DHL cargo plane taking off from … Leipzig.
https://simpleflying.com/dhl-boeing-737-400-crash-liepkalnis
https://edition.cnn.com/2024/11/25/europe/lithuania-dhl-cargo-plane-crash-vilnius-intl-hnk/index.html
Prompted by recent discussion on EK343 and its apparent confirmation of the “timing anomaly” at Arc 1, I revisited various strands of evidence around MH370’s path from disappearance to the first Arc.
I conclude that all available evidence – including triangulation with independent data sources – hangs together very well. This also gives me a high level of confidence in the veracity of “military radar” and the Lido slide. If the timing offset is real, how can it be reconciled? My write up is at the link below.
https://tinyurl.com/4pap2dj6
@Victor
Re: your point3 above (N571 offset)
I certainly feel IG has made many significant contributions to understand the flight, and the proposed N571 Offset maneuver is one such observation that helps to unify the 1822 Radar + Inmarsat data. I am ambivalent if the right offset is exactly correct, but some similar maneuver.
@sk999
I know I believe Joseph Coleman gave me the EK343 data, but at some point approx NILAM region it becomes extrapolated data.
@ Paul Smithson,
An excellent write-up and summary.
I believe the conduct of the flight was to disappear, and as such there was no point in having known radar watching you make any change of track that could indicate your ultimate intentions. Descent, resulting in a slowdown that eats up the 90 – 100 seconds would fit.
@Victor @All
Have you ever considered to construct the probability map with the assumption of controlled glide after fuel exhaustion with maximizing the flight range?
So far, maps showed highest probability maps around 7th arc when uncontrolled, spiral flight was assumed. Everything beyond that was given equal chance, no matter if it was 50, 80 or 120nm away from the 7th arc.
However, if it was really a controlled glide would it be logical to assume that the flight range extended to the maximum?
Here is the example. If the fuel was exhausted just before reaching the 7th arc at 34-36degS, and if the plane was not found within the 25nm range from the 7th arc, then the next area we should look is e.g. 80-100nm south from the 7th arc (or 90-110nm, it does not matter) if controlled glide with maximizing range is assumed. In other words, area between 25nm and 80nm from the 7th arc is less probable since the plane was under human control.
With that assumption potential search area would still be large, but still smaller and it could be further refined with ocean drift studies.
@Barry, by my calculation, with straight flight along N571, you would need a descent to ~FL250 and commensurate slow-down to ~400TAS (0.84/280/250 profile) completed just before 182212 to get the timing aligned. Alternatively, a slightly greater descent (to FL180 or FL200) completed shortly after 182212. You would need to start your descent latest about 1815 for this to work out, otherwise you can’t shed enough speed before 182212. All assuming the “DSTG/last Geosource” position is reasonably accurate. So I concur that a straight line descent/slow-down can do it, but it’s a squeeze…
For those interested in the flight path of EK343 relative to MH370, this graphic may be of interest, where the path of EK343 was calculated from the SSR data sets from the Butterworth and Langkawi radar sites, which were obtained through a back channel.
https://www.dropbox.com/scl/fi/r5gstrzhhdnfqsoqovsm9/2024-11-26-Radar-Data-for-MH370-and-EK343.png?rlkey=uiquwit0zhfq409vyagnbgc50&dl=0
At 18:22:12, which is the last radar point for MH370 in the Lido Hotel slide, MH370 is 10 NM past MEKAR on N571, and EK343 is around 24 NM before MEKAR also on N571, for a distance between targets of around 34 NM.
Some observations:
1. EK343 was tracking on airway R467 between the VKL VOR and GUNIP, and then joined airway N571 towards VAMPI and MEKAR.
2. EK343 was out of range of the Langkawi radar site just after passing MEKAR.
2. Malaysia’s military radar should have had no problems distinguishing between the MH370 and EK343 targets due to the relatively large spacing, which was around 34 NM at 18:22:12z.
3. The targets from the Lido Hotel slide are close but not exactly positioned in the figure due to difficulties in accurately extracting the coordinates from the original photograph.
4. As we’ve known for years, the data from the Lido Hotel slide is noisy and sparse, which may explain why Malaysia has been reluctant to share it.
Thanks for sharing that, Victor. Interesting to see that separation confirmed through to 1822.
@Marijan. Interesting point that quite likely the aim of a glide would be to get maximum distance south. So in any glide area search, that should be the priority.
@David, @Marijan,
David, as you may recall this matter has been discussed on this (Victor’s) site quite extensively. As I recall, both you and (certainly) I may have discussed the matter before. It also was the basis for the second recommended general area for search, being Area A2, subject of “The Final Resting Place of MH370” by Authors UGIB in March 2020.
This report may be accessed by the link to: https://mh370.radiantphysics.com/2020/03/09/new-report-released-for-mh370-search/
As I have commented before, if there was a still functioning human pilot on board and in the cockpit, and with a wish to proceed as far as possible then it makes a degree of sense that that pilot guide the aircraft as far south as possible. Unless, as I previously suggested, that that pilot wanted to revel in the rising sun and guide the aircraft slightly eastwards towards it on the way down.
All the above is regardless of, and subsequent to, any specific areas of interest, that Ocean Infinity, or others, may have considered. This would be particularly with respect to Ocean Infinity with their detailed knowledge of, and data from, the previous search, and plenty of time since for extensive analysis.
Note: Any current analyses being undergone might just provide additional data to restrict or constrain the apparent generally large area outside any specific areas of interest. Fuel availability analyses and drift analyses do assist in “drawing” those constraints. Unfortunately, there will always be statistical considerations concerning any and all constraints in additions to that in any analysis of the satellite metadata. High or low confidence levels may be derived for any specific area or region. Only after the aircraft or most of it is found will there be something approximating certainty.
All, sorry for that doubtful note.
@David: “the aim of a glide would be to get maximum distance south.”
Going for the maximum distance is not compatible with the nose dive observed in the BFO data, no ?
@PeterNorton
Those who continue to assert with ‘religious ferver’ that the final BFO’s are ‘slam-dunk proof’ of a ‘death-dive’ close to the 7th arc, are not helping to find the aircraft.
There are alternative explanations for them.
To cite one, read this.
https://www.mh370-caption.net/wp-content/uploads/Blelly-Analysis-of-Flight-MH370.pdf
Just look at figure 2, you should get the idea.
@ventus45, Some simple questions:
1. Who is asserting with “‘religious ferver [sic]’ that the final BFO’s are ‘slam-dunk proof’ of a ‘death-dive’ close to the 7th arc”? That certainly was not the claim in UGIB 2020, although some have falsely made that assertion.
2. Where in Figure 2 does it show that there was an average downward acceleration of 0.7g, increasing the descent rate from 4,000 fpm to 14,000 fpm, during the 9 seconds between the two transmissions at 00:19? If there was a long glide, it would have occurred after a recovery from a steep descent.
3. In Figure 2, it shows that ELMS shed the SATCOM while the APU was running and both engines had shutdown (the right from fuel exhaustion and the left from a manual shutdown). Where is the documentation that ELMS would behave in this way? It certainly is not a scenario that Boeing has proposed.
@ventus45: I looked at fig.2
Of course, it’s technically possible to recover from the dive suggested by the BFO data. But the lost altitude (depicted in fig.2 as well) is contrary to the goal of reaching “the maximum distance south” as argued above.
@Victor Iannello:
Is this a new development ?
https://www.perthnow.com.au/news/mh370/malaysia-airlines-search-for-missing-mh370-flight-set-to-resume-off-coast-of-wa-after-credible-proposal-c-16677311
What do you know about the current state regarding the decision to restart the search ?
@George G. Thanks for that reminder.
@Peter Norton, I had in mind the possibility of failure of non-glide searches. We not know what of UGIB is in OI’s thinking vs other proposals/research it might have before it, where the piloted possibility stands relative to those, or the extent of searches it will have agreed to.
Then there is the sequence it will have planned. If a search of a high priority area fails and that is proximate to another lower down but on its to-do list, it might be economical to take the opportunity of looking there before relocating to a more distant high priority area or replenishing, saving on travel.
@Victor
From the earliest days of the ATSB’s ghost flight scenario, it has long been held by the media and non-pilots alike (and certainly not dispelled by the ATSB) that the final BFO’s confirmed an accelerating rapid decent which equates to a dive which equates to a crash close to the arc.
The media often use additional terms such as spiral dive and or death dive, presumably for dramatic effect. That is what the public at large has been served up by the ladle full for years, and that is what most people accept as fact.
That is religious fervor, and that is what annoys me, because it closes people’s minds to serious consideration of any other possible scenario, and it seems that this is deliberately peddled by some to cement the notion that the wreckage must be very close to the arc.
Even today, the ATSB cautiously discounts any other possibility as “highly unlikely”, which in the public’s mind equates to, “yeah, well, they got it right the first time, anything else is a conspiracy theory”.
However, the fact that there was a very rapid descent for some seconds does not prove that it continued to finality. Even the iron bird simulator studies show phugoids without any pilot intervention whatsoever, so the notion of a continuously accelerating dive to impact is on shaky ground to begin with.
Recovery from a dive (however initially caused) is entirely possible as any pilot knows. If any pilot was in the seat and conscious he would have done so, or put the stick in the panel, take your pick depending on what you think his intention was.
If his intention was to glide onwards, he would have recovered, and after recovering, he would have glided on.
Figure 2 in the reference is one person’s descent scenario. Other people may have their own.
What the final configuration of the aircraft was, whether or not the last engine was deliberately switched off or flamed out, whether or not the apu start was auto or manual, whether or not rat deployment was auto or manual or even at all, or what ELMS did or did not do, and or when, is all open to conjecture, but all of it is irrelevant to what the physical control column was, or was not, doing, or should I say, was having done to it.
@ VictorI
You wrote “Where is the documentation that ELMS would behave in this way? It certainly is not a scenario that Boeing has proposed.”
Over on RG’s website I found this comment below:
“The Electrical Load Management System (ELMS) automatically sheds non-essential electrical loads when the APU starts. The ELMS protects the electrical power system. It sheds (disconnects) electrical loads to keep the load levels below the power supply levels. In-flight the ELMS monitors the status of these three power sources for load shed control:
– Left Engine Integrated Drive Generator (120 kVA).
– Right Engine Integrated Drive Generator (120 kVA).
– APU generator (120 kVA).
These are the loads that the ELMS can shed and the general sequence it sheds them:
– Galley loads.
– Utility buses.
– Equipment cooling vent fan.
– Galley chillers.
– Recirculation fans.
– Lavatory/galley fans.
– Electronic passenger seat equipment.
– Hydraulic pumps.”
https://www.mh370search.com/2024/03/16/mh370-a-new-hope/
Where did RG get this sequence from?
@CMR: That’s in the FCOM. I don’t see mention of shedding the SATCOM.
@ventus45: At this point, in light of the negative search results, I don’t know anybody (including the ATSB) that says a manual glide was not possible.
@Peter Norton: It’s up to the Malaysian parliament to fund a new search. I don’t know the status of that vote. People make strong claims that later show they are not as informed as they present themselves to be.
@David @PeterNorton @GeorgeG @ventus45
Please correct me if I am wrong because I was not 100% involved in discussions here, but I thought about something you are discussing.
I don’t recall that the BFO after the SDU bootup sequence was ever EXPERIMENTALLY validated when the SDU was without power for a couple of minutes.
The 1st arc BFO data was dismissed as unreliable due to the prolonged period of time where SDU was without electrical power (and cooling of oscillator). However, 7th arc BFO was interpreted as a steep descend, but what if actually that was not the case and that even couple of minutes without power needed to start the APU and boot up the SDU was enough to produce data which were only “partially erroneous” just enough so they could be taken as accurate while in fact they were not.
Consequently, that would mean that the plane might not have been in steep descent at 7th arc after all.
Again, I apologize if I missed something.
@Victor @Peter Norton
The current meeting/session of the Malaysian parliament ends on Dec 12. If there is going to be any voting this time about MH370 it will probably be before that date.
@Marijan: Before the temperature of the crystal oscillator stabilizes, the frequency excursion tends to be positive. The final BFOs were progressively negative. It’s true that BFO stability at power up was investigated for only long power down times (as far as we know), but it would be odd if a short power down would produce such different behavior in direction and trend of the frequency excursion.
Thank you for the information about the Malaysian parliament. In light of the limited window to search due to seasonal weather, you would think a vote would be scheduled sooner than later.
@Andrew:
I am probably misunderstanding the FCOM cited above, but why does the ELMS shed the electric load for something essential such as the equipment cooling fans BEFORE non-essential loads (such as “electronic passenger seat equipment”) ??
Happy Thanksgiving to those in the US!
It should be noted that the load shedding list given by RG comes from the Continental Airlines 777 Training Manual, section “Electrical Power”, p. 147.
@Peter Norton,
I’ve just checked, and the equipment cooling system uses 2 fans-the supply fan and the vent fan. Perhaps the short-term loss of the vent fan won’t be an issue.
Tim, thanks! But still more important than electronic passenger seat equipment, Andrew ?
Hello all and Happy Thanksgiving to the US crowd and best wishes to you all.
Glad I dropped in today as the discussion has returned to one of my favorite topics…..the dive or glide.
Some of you may remember that Barry, Geoff and I, many years ago proposed the glide hypothesis and it was roundly dismissed. At the time it was not politically correct to discuss the possibility of an active pilot at the 7th arc. If I recall this was before the failure of the first search expedition.
But the strength of the push back at the time, including the ATSB analysis has biased the thought process regarding any future searches. Looking South of the 7th arc is more likely to be productive than any other blended search strategy.
Again, let me point out that the original IG 7th arc intersection is still a valid option based on a least fit error to a simple TT scenario. Dr.B has validated the fuel model for the specific best fit path.
I welcome the additional analyses and discussions that have ensued as they continue to turn up interesting facts and inferences. One hopes that this has not been in vain.
Happy thanksgiving to all celebrating same.
@CanisMagnusRufus
@Peter Norton
What the MAS B777 FCOM states about the ELMS is as follows:
“Electrical Load Management System (ELMS)
“The ELMS provides load management and protection to ensure power is available to critical and essential equipment. If the electrical loads exceed the power available (airplane or external), ELMS automatically sheds AC loads by priority until the loads are within the capacity of the airplane or ground power generators. The load shedding is galleys first, then utility busses. Utility busses are followed by individual equipment items powered by the main AC busses. When an additional power source becomes available or the loads decrease, ELMS restores power to shed systems (in the reverse order). The message LOAD SHED displays on the electrical synoptic when load shed conditions exist.”
Elsewhere the FCOM notes the following:
“The main busses power individual equipment items such as:
• cooling vent fan • electric hydraulic pumps
• recirculation fans • IFE
• lavatory/galley fans
“Each main bus also powers its associated busses (typical loads are shown in parentheses):
• transfer bus (DC system transformer–rectifiers, AC standby bus)
• utility bus (forward galley heater, chiller boost fan, gasper fan, captain’s and first officer’s foot and shoulder heaters, door area heaters, lavatory water heaters and shavers)
• galley busses.”
I’m sure that Andrew can elaborate.
Noting that the IFE component of the SATCOM log-on sequence is for the SMS e-mail application and Built-In Test Equipment application, Don Thompson may be able to shed some light on the structure of the IFE with regards to the division between the head equipment and the passenger seat equipment.
@Victor @all
You are welcome and Happy Thanksgiving to you and all readers and contributors of this blog who celebrate it. Enjoy your holiday
@Peter Norton
RE: “I am probably misunderstanding the FCOM cited above, but why does the ELMS shed the electric load for something essential such as the equipment cooling fans BEFORE non-essential loads (such as “electronic passenger seat equipment”) ??”
I don’t have a definitive answer, but as Tim mentioned the equipment cooling system has two supply fans in addition to the vent fan. One supply fan operates at a time to provide blow-through cooling, while the vent fan draws warm air away from the equipment.
The B777 MEL allows the aircraft to be dispatched with the vent fan inoperative. The only restriction is that for ground operations with the OAT ≥ 30ºC, both packs must be selected on or the aircraft supplied with conditioned air. Accordingly, I think it’s fair to say the vent fan is not so ‘essential’.
As for the passenger seat equipment having a higher priority, it might be to cover the case where load shedding occurs on the ground before engine start. The higher priority would usually allow passenger safety briefing presentations to continue uninterrupted.
@Sid Bennett. You said “Dr.B has validated the fuel model for the specific best fit path.” If you are referring to the original IG spot at 38.0 adjacent DSTG, it was my understanding that Bobby’s subsequent fuel modelling puts that out of range.
@Andrew: If the MEC fans fail, doesn’t the equipment automatically reconfigure to override mode, in which a vent to the outside from the MEC opens, drawing air through the MEC in flight?