The missing ACARS message sent over the VHF link.
Today, Malaysia’s Ministry of Transport quietly released the full ACARS message log for MH370. The new log confirms that the traffic logs presented in previous reports were incomplete and edited, as asserted in a previous blog article. This release comes on the heels of a strong denial from Malaysia Airlines stating that it has “provided full cooperation and assistance to all respective authorities”.
Notably, the new log contains an additional ACARS message that was sent from MAS Operations Dispatch Center (ODC) and destined for MH370 over the VHF link. The message was sent at 18:38:51 and was intended to be displayed in the cockpit on a Control Display Unit (CDU), which a pilot uses to perform tasks such as programming the flight computers. The message was not received by MH370, and was re-sent by MAS ODC at 18:39:52, 18:40:42, and 18:41:52, failing each time. The text of the message was:
DEAR MH370. PLS ACK TEST MSG. RGDS/OC.
The new log confirms that there was a renewed attempt to initiate communications with MH370 using ACARS over the VHF link at 18:38:51. The error messages that were generated confirm that the VHF link was not available at that time, likely because MH370 was not logged into ARINC’s server. ARINC was MAS’ service provider for ACARS over VHF.
Also of note is the new ACARS message was sent about a minute before an attempted telephone call over the satellite link at 18:39:56, suggesting an increase in activity at MAS ODC at this time.
The new log contains other traffic between MAS ODC and other ground computers. This data is under examination for additional clues. The new log also confirms that the name of the MAS ODC employee that sent the ACARS messages was redacted from the message logs in previous reports.
[This first appeared under the previous post.]
@Victor,
I AM @nineinchhair, sorry about the name confusion.
There seems to be no official statement. Just a new link, appeared today on the official Website of the Ministry of Transport Malaysia.(http://www.mot.gov.my/en)
One friend would routinely refresh the Website every day to see if there are any updates. And he shared the link. He also leads a self-organized group of Chinese aviation enthusiasts who gather online to discuss MH370-related topics.
We have some NoK members in the group. They told us that the Malaysian investigation team would visit Beijing around Oct. 17, presumably to provide more detailed information.
@haxi: Thank you for all your help. I hope your group of aviation enthusiasts is aware of the existence of the blog.
It is encouraging to see that MAS responded to the latest pressure to disclose the full ACARS message record. It is hard to understand why that took 4.5 years. I wounder how long it will be before all the radar data is disclosed.
@Victor
Couple questions-
> What is the time span? Obviously we would like to see out to beyond end of flight
> I see references to MAS371
> The flight path waypoints .BITOD..BIBAN..BITIS are easier to understand than the way it is shown in the SIR
Slow but steady progress @Victor. Congratulations. I am assuming you and the IG will carry out examination of the newly available data for additional clues?
The flurry of activity at 18.38/39 by MAS ODC came hours too late to be of any practical use in saving the lives of 239 passengers..
@Julia: All are invited to examine the logs and report back with new discoveries and insights. I know that Don, Richard, and Mike are already on the case. In the article, I only presented the most obvious of new evidence.
@Airlandseaman said: It is hard to understand why that took 4.5 years.
There was only a response because the issue was raised here. Without the scrutiny of Don, Richard, and others, the new release would have never occurred.
@ALSM
@Victor
RE:”It is encouraging to see that MAS responded to the latest pressure to disclose the full ACARS message record.”
What evidence is there to suggest this latest release had anything to do with MAS?
@Victor
Re: 18:40
That is potentially very critical period, because if anyone has their “thinking caps” on (pilot and/or MAS), the satellite phone call is “successfully” unanswered, which means that MAS should realize the aircraft is still flying (which Chief Ops Officer Hugh Dunleavy said they did assume it may still be flying, but he came into the office later) and the pilot would know that MAS/ground (Razak) should be smart enough to realize this flight has been diverted.
Although I tend to agree with you and ALSM and Jean-Luc, that perhaps a descent was in progress at 18:40, the other possibility is a turn towards Banda Aceh at IGOGU was in progress. But the phone call could have been a reason for the pilot to change course, for various reasons.
@Andrew: The new report was created on Sept 7, 2018, using an application that reports all operations-related traffic related to MH370 [9M-MRO] for messages incoming to and outgoing from MAS ODC, including data between ground computers, and also with two ACARS service providers (SITA and ARINC). It seems that data would be owned by MAS. Who else could have generated that report?
@Victor
The SITA copyright appears at the bottom of each page. If the MOT didn’t hold the complete log, then perhaps they requested it from SITA (or directed MAS to do so) following the recent publicity.
@TBill,
You said: “Although I tend to agree with you and ALSM and Jean-Luc, that perhaps a descent was in progress at 18:40, the other possibility is a turn towards Banda Aceh at IGOGU was in progress. But the phone call could have been a reason for the pilot to change course, for various reasons.”
The BFOs related to the first unanswered phone call from 18:39:55 to 18:40:56 do not show the BFO variations one would expect from either a turn in progress or a typical descent (either in progress or just beginning). If the phone call triggered a maneuver (turn or altitude change), it did not begin until after the phone call timed out.
Apparently the ACARS message at 18:38:51 was not received at the aircraft, so the pilot would have no knowledge of it, and it would not have triggered a maneuver.
@ALSM,
You said: “My best guess, based on the KB altitude and radar derived speeds up to 18:22 is that the altitude remained unusually high between 17:30 and 18:22, probably 43000 ±2000 feet. I think there was a descent underway at 18:40.”
If you think a descent was underway at 18:40, it probably was not underway 15 minutes earlier at 18:25. Near that time the BFOs are well matched by a right lateral offset maneuver. Why do you think that maneuver was done then if the aircraft was at 43,000 feet? Or do you think a different maneuver took place at 18:25? If so, what?
@Andrew: Some of that communication doesn’t appear to be with SITA computers, although the communications software application might have been supplied by SITA (AS.V6R9.2). I could be wrong, but since the common destination/source for all the messages is MAS ODC, that would appear to be where the messages are stored.
@DrB
I am saying after 18:40 sat call (not ACARS) may be a time to change course because (1) pilot should know the diversion is now undeniable by Razak (who nonetheless tried to deny it as long as Inmarsat/Obama allowed) such that the pilot can now progress with next phase of plan; and (2) pilot might be spooked that the satellite call might leave “bread crumbs”. I assume the pilot is not trying to hide the diversion is why SDU in ON…otherwise Razak never has to admit to it.
Bobby: You wrote above “…The BFOs related to the first unanswered phone call from 18:39:55 to 18:40:56 do not show the BFO variations one would expect from either a turn in progress or a typical descent …”.
I don’t understand why you are now claiming this after 4 years of general consensus that the 18:40 BFO values indicated one or the other. In any event, I don’t agree with that statement at all. MH370 was definitely either turning or descending at 18:40 (or possible some of both).
I lean now toward the descent scenario simply because the plane was not found around S37, or anywhere near by. All things considered, it seems more likely now that there was some delay before heading south, thus my best guess that it was descending at 18:40. Given the estimated rate of ~2500 ft/min, if the descent started from 43000 ft, it probably would have lasted several minutes. I don’t have any idea when it started or finished. I think we all agree some maneuvering was going on between 18:22 and 18:40.
@DrB, @airlandseaman: If the plane was descending at 18:40, then it was descending at a very steady vertical speed, and likely in V/S mode where the pitch would be controlling vertical speed rather than the airspeed. Or, the plane was in level flight in a southerly direction.
A maneuver after 18:22 is required if we accept the Lido Hotel radar data.
I’m with Dr B on this one (interpretation of the 1840 BFOs).
Not only do you happen to have just the right amount of descent going on at 1840 for it to mimic a southern course consistent with subsequent BFOs, you also need an uncannily and atypically smooth descent. I’ll allow that the above is possible, but it doesn’t smell “probable” to me.
This is the main reason that I believe that the plane must have been flying south by 1840, that the N terminus can’t be right, that the original hotspot path was most probably correct and the final 0019 BFOs must be wrong for reasons not yet understood.
@ALSM,
You said: “I don’t understand why you are now claiming this after 4 years of general consensus that the 18:40 BFO values indicated one or the other.”
My position on this has not changed. What I have said is that a “typical” descent will generate larger ROD and BFO variations than the very steady pattern measured at 18:40 (only +/- 2 Hz peak to peak, due to ALL sources of BFO noise, including the portion due to ROD variation). It may be possible to fly a more constant ROD (with smaller BFO variations) by commanding a vertical speed instead of simply changing the desired flight level, but my understanding is that setting a vertical speed is not a normal practice (Andrew may correct me if I am wrong on this point), and I have not seen a proposed rationale for doing so in this situation.
Here is another argument that the BFOs seen at 18:40 represent straight and level flight. We have a second unanswered satphone call at 23:14. I believe the aircraft was flying on A/P at that time with a constant track, airspeed, and flight level. I am unaware of any proposed route which has an ongoing maneuver at that time. So, for the sake of argument, let’s say that was indeed the case. The BFOs at 23:14 have a peak to peak of +/- 3 Hz and a standard deviation of 1.7 Hz, and, by our assumption, this is what we should expect for straight and level flight at constant speed. We can compare the 23:14 BFO errors with the 18:40 BFOs, which have a peak to peak of +/- 2 Hz and a standard deviation of 1.3 Hz. Thus, the noise statistics of the two events are similar, and the 18:40 BFOs shows slightly SMALLER variability than the 23:14 BFOs. If a descent were ongoing at 18:40, the BFOs then should vary MORE than for level flight, since the ROD variation is in addition to the other sources of BFO variation seen with the assumed conditions at 23:14. Taking into account the finite number of BFO samples, the standard deviation at 18:40 is 1.3 +/- 0.18 Hz and at 23:14 it is 1.7 +/- 0.33 Hz. Thus, the standard deviations of the two events are consistent with both being the same value to better than 1 sigma. This means there is no statistically significant difference in their BFO noise levels. I assert that if a descent was ongoing at 18:40, the BFOs then would show a measurably LARGER BFO variability than seen at 23:14. In fact, the BFO noise is the same (or less), not measurably larger, and therefore it is most likely that the aircraft was also in straight and level flight at constant speed at 18:40.
If you claim the BFOs at 18:40 represent a descent, then one must conclude either (1) that a descent was also occurring at 23:14, or (2) that the descent at 18:40 held ROD so steady that its contribution to the BFO noise was negligible. To be negligible, the BFO variations due to ROD fluctuations would have to be about +/- 1 Hz or roughly +/- 60 fpm peak-to-peak. I doubt this is possible for a minute of time at the average descent rate needed, regardless of the descent method used.
It would be illuminating to analyze data from another flight or with the same aircraft type to see how steady the ROD actually is during descents. The difficulty is that the only scenario that offers roughly 1-second sampling of BFOs over a minute of time is an unanswered satphone call, and I am unaware of any such BFO or ROD data records besides these two cases in MH370.
You also said: “I lean now toward the descent scenario simply because the plane was not found around S37, or anywhere nearby. All things considered, it seems more likely now that there was some delay before heading south, thus my best guess that it was descending at 18:40.”
I agree there was delay such that the FMT occurred well after 18:40, but I don’t see why that conclusion, or the fact that the aircraft was not found at 37S, makes a descent more likely to have occurred at 18:40 relative to other possible maneuvers, such as a turn southward prior to 18:40.
@ALSM,
@Victor Iannello,
One can also look at the 15 BFOs at the gate from 16:00:32 to 16:29:52, which show a +/- 3 Hz peak to peak and a standard deviation of 1.8 +/- 0.45 Hz. This data set is also consistent with the BFO noise levels measured at 18:40 and at 23:14. This consistency means that the dominant contributor to the BFO noise during the satphone calls is system related and not due to aircraft motion. That is, the measured BFO noise is the same (within the measurement errors) at the gate when the aircraft is motionless as it was during BOTH satphone calls. I would not expect horizontal speed variations over one minute nor track changes (assuming no turns were underway) to contribute significantly to the BFO noise, based on the relative insensitivity of the BFO to horizontal speed and track produced by the AES frequency pre-compensation. However, the BFO is highly sensitive to vertical speed because it is uncompensated by the AES. My conclusion is that it is highly unlikely that a descent was ongoing at either 18:40 or at 23:14.
@TBill asks “What is the time span? Obviously we would like to see out to beyond end of flight”
The time span is at the top of each page, until 2:00:01 on 8th March 2014.
References to MH371: yes, the later messages in the Traffic Log relate to the subsequent sector that 9M-MRO was expected to fly, ZBAA back to WMKK, the MH371 service.
This, new, expanded Traffic Log is not expected to include specific detail for satcom datalink traffic. I have seen some uninformed mumblings on social media that are incorrectly conflating the two.
This Traffic Log comprises operational messaging internal to various MAS locations and functions, MAS and external service providers, and MAS and the aircraft. Functions such as flight planning, flight tracking, dissemination of events such as OOOi/OUT and OOOI/OFF, and so on.
@Don
Time I was confused UT or local time, as I was thinking MH371 was prior flight.
@DrB: We’ve been through this before. Descending planes in VNAV or FLCH autopilot modes will have higher variation of vertical speed than a plane in a V/S descent. Planes following flight plans are typically in LNAV/VNAV autopilot mode, so I don’t think you look at the vertical speed variation of a typical commercial flight to draw conclusions about MH370.
Frankly, I’m not even sure we should expect ANY increase in vertical speed variation for a V/S descent versus level flight, other than differences in turbulence. In both cases, the plane is in quasi-steady equilibrium. The vertical speed, vertical acceleration, and pressure altitude are measured no more accurately around 2500 fpm than 0 fpm. I would expect any differences in vertical speed variation to be more a function of turbulence. A plane flying level through turbulent air could have more vertical speed variation than a V/S descent through clear air.
If you are going to base your argument on the fact that pilots don’t typically use V/S descents, the discussion is going to quickly become absurd when applying how pilots typically fly to MH370. By all measures, this was not a typical flight.
Bobby: Thanks for explaining the basis for your comments. I understand the logic, but I’m not convinced a descent from a high altitude with no turbulence would produce significant BFO noise, even if a V/S descent was not used. There are not many thermals at 43000 feet at 2am. It was probably very smooth air.
But let’s say you are correct, and the data is indicative of a level turn to the south circa 18:40. If that happened, we should have found MH370 ITVO S37, unless you think more turns and maneuvers followed.
@airlandseaman says: unless you think more turns and maneuvers followed.
Which is another reason that it is very hard to link what happened at 18:40 to a position on the 7th arc. We might be in a situation where even if we knew the position at 19:41 we could not estimate the impact site with sufficient precision.
@DrB, @Paul Smithson
@DrB stated “The BFOs related to the first unanswered phone call from 18:39:55 to 18:40:56 do not show the BFO variations one would expect from either a turn in progress or a typical descent (either in progress or just beginning).
@Paul Smithson stated “I believe that the plane must have been flying south by 1840”.
The BFOs between 18:39:55 UTC and 18:40:56 UTC for a level flight at an assumed altitude of 35,000 feet and an assumed ground speed of 498 knots result in a track either around 217°T or 144°T. Neither BFO solution is anywhere near 180°.
@DrB has proposed a CMT track at 181.22°M (180°T) but requires a S-curve holding pattern to fit the satellite data from 19:41 UTC onwards. @DrB states that MH370 turned toward waypoint ANOKO at 18:37:20 UTC maintaining the previous Lateral Offset and just before the SATCOM call creating the BFOs between 18:39:55 UTC and 18:40:56 UTC took place, just as MH370 was passing near waypoint IGOGU on a track of 206°T the call took place. MH370 then turned northwards entering a holding pattern at 18:41:10 UTC only 14 seconds later and at 18:52:45 UTC finally turned southwards on 180°T.
@Paul Smithson says “Not only do you happen to have just the right amount of descent going on at 1840 for it to mimic a southern course consistent with subsequent BFOs, you also need an uncannily and atypically smooth descent. I’ll allow that the above is possible, but it doesn’t smell “probable” to me.”
I could similarly say of @DrB’s proposal, that not only do you need the right timing to turn toward waypoint ANOKO for it to mimic a level flight on a track of 206°T, you also need the right timing to enter a holding pattern northwards 14 seconds later and finally the right timing to turn southwards on 180°T to match the satellite data at 19:41 UTC. I’ll allow that the above is possible, but it doesn’t smell “probable” to me.
@DrB
Actually your latest flight path embodies what I am saying, after the 18:40 sat call the pilot is then possibly making another maneuver under the new circumstances. In your latest case you make 2 turns instead of a descent, but you end up pretty much equivalent to where a descent would end up at 0894E.
@Victor
We might be in a situation where even if we knew the position at 19:41 we could not estimate the impact site with sufficient precision.
I agree. Take a guess consistent with a landing “preference”, and you are left with a path toward the Cocos or CI.
@ALSM,
You said: “But let’s say you are correct, and the data is indicative of a level turn to the south circa 18:40. If that happened, we should have found MH370 ITVO S37, unless you think more turns and maneuvers followed.“
As I said in my last paragraph addressed to you: “I agree . . . that the FMT occurred well after 18:40 . . . .” So, yes, I absolutely believe the FMT was made well after 18:40, and possibly between 18:50 and 19:28. I also agree with Victor that even knowing the position at 19:41 might not allow us to precisely locate the terminus, because that may not produce a unique solution.
@TBill,
You said: “Actually your latest flight path embodies what I am saying, after the 18:40 sat call the pilot is then possibly making another maneuver under the new circumstances. In your latest case you make 2 turns instead of a descent, but you end up pretty much equivalent to where a descent would end up at 0894E.”
That is correct; there are multiple paths that satisfy the 18:40 BFOs. In my CMT case, exiting a possible HOLD cannot produce a southbound route after 19:00 that matches the remainder of the sat data, so I don’t use a HOLD at all, but I do need turns to reach the pair of waypoints that gives a 180.0 degree true track, which becomes 181.2 degrees magnetic track if TRK HLD is then applied. A descent at 18:40 puts the aircraft near 8N94E somewhat earlier than my suggested route does.
@Victor
RE: ”Some of that communication doesn’t appear to be with SITA computers, although the communications software application might have been supplied by SITA (AS.V6R9.2). I could be wrong, but since the common destination/source for all the messages is MAS ODC, that would appear to be where the messages are stored.”
My understanding is that MAS is contracted to SITA for ACARS messaging. All ACARS messages are processed through SITA, but some are then routed through ARINC in areas where SITA has no VDL coverage (at much greater cost, as @Don previously mentioned). It stands to reason that SITA would retain the data that was processed through its systems. Indeed, the following article suggests that SITA has been assisting the MH370 investigation since the beginning: https://runwaygirlnetwork.com/2014/03/14/sita-aids-mh370-investigation-expert-explains/
Perhaps SITA was responsible for the initial ‘filtering’ of the ACARS data, simply because they weren’t asked for the complete logs.
Victor wrote: The new log confirms that there was a renewed attempt to initiate communications with MH370 using ACARS over the VHF link at 18:38:51. The error messages that were generated confirm that the VHF link was not available at that time, likely because MH370 was not logged into ARINC’s server. ARINC was MAS’ service provider for ACARS over VHF.
How are broadcasted those message via VHF? To all the world? Or only where the plane” could be”? If the “system” had the plane above Cambodia or Viet-Nam, could it have been broadcasted from a place too far from Malacca Straits to be “heard”?
PL
@Richard Godfrey,
In characterizing my proposed CMT Route, you said: “. MH370 then turned northwards entering a holding pattern at 18:41:10 UTC only 14 seconds later and at 18:52:45 UTC finally turned southwards on 180°T.”
You also said: “I could similarly say of @DrB’s proposal, that not only do you need the right timing to turn toward waypoint ANOKO for it to mimic a level flight on a track of 206°T, you also need the right timing to enter a holding pattern northwards 14 seconds later and finally the right timing to turn southwards on 180°T to match the satellite data at 19:41 UTC. I’ll allow that the above is possible, but it doesn’t smell “probable” to me.”
I would agree with you if what you said was true, which it is not. You seem to misunderstand my proposed CMT Route. There is no HOLD. To quote from my paper on p. 37 regarding the FMT route shown in Figure 19:
“There are three turns. The first turn is to waypoint ANOKO (while maintaining the 15 NM right lateral offset). This produces a 206° true track at 18:40, matching the phone call CBFOs then at MRC speed [29].
The second turn begins toward AGEGA when ANOKO is reached. The third turn is to BULVA when in the vicinity of AGEGA. This set of turns matches the back-track position at 19:00 near AGEGA travelling due south. The exact path near AGEGA depends on when the speed reduction occurred from M0.84 (across the military radar track) to MRC (from 19:41 onward). In Figure 19 MRC was commanded at the same time as the altitude was reduced by 500 feet (circa 18:29). If the slow-down occurred later, then the FMT Model allows a path circling around AGEGA instead of turning just before reaching it.”
To clarify that last point, it is possible to generate this route by entering ANOKO/AGEGA/BULVA as LNAV waypoints circa 18:37 while flying on N571 past NILAM with a 15 NM right lateral offset. There is only one additional action required after 18:37, and that is to set TRK HLD near BULVA circa 19:28. Thus, you don’t need precise timings entering and leaving a HOLD since there is no HOLD. The relative timings between the 206-degree course turn and the 180-degree course turn are provided simply by the distance between waypoints and the aircraft speed. There is no need for the pilot to make any turns (after the first one) at specific times. In fact, for my CMT Route there is no need for the pilot(s) to do anything between 18:37 and 19:28. Let that sink in for a minute; it surprised me.
@PaxLambda
You make a very good point.
It would be instructive, if it could be determine from the system logs, which actual ground stations (antenna sites) those ACARS VHF messages were past to, and transmitted from, and to determine their geographic coverage, to determine whether or not, 9M-MRO could have been within their coverage, at the appropriate time(s)
@Andrew: It looks as though some of the data in the traffic logs isn’t really even ACARS data, and therefore would not be routed by SITA. Perhaps @Don Thompson can weigh-in.
@PaxLambda: The VHF link was established at 12:50:22 and lost at 15:54:53. The VHF link was not established again. I don’t think there were any VHF transmissions after 15:54:53.
DEAR MH370 ?
Isn’t that a bizarre ACARS message ?
Sounds very strange to my non-anglophone ears.
@DrB
You suggested “It would be illuminating to analyze data from another flight or with the same aircraft type to see how steady the ROD actually is during descents.”
Below is a link to the results table from an analysis of the BFO data during the MH371 flight, including the turn of 33° at waypoint EPDOS (LNAV) and the descent phase at -2,055 fpm (VNAV).
A snapshot of the MH371 cruise at 40,000 feet for 13 mins including 159 BFOs gives a SD = 1.76.
A longer snapshot of the MH371 cruise at 40,000 feet for 1 hour including 257 BFOs, but including the turn at waypoint EPDOS, gives a SD = 3.75.
The descent phase of MH371 at -2,055 fpm for 73 seconds including 164 BFOs gives a SD = 4.33.
Finally from MH371 Landing to the Gate for 93 seconds including 132 BFOs gives a SD = 1.82.
As you point out, the MH370 BFOs at 18:40 UTC for 61 seconds including 51 BFOs giving a SD = 1.26 is neither showing a typical LNAV turn or a VNAV descent.
https://www.dropbox.com/s/vtj15f3qqh7o9z9/BFO%20Comparison.png?dl=0
Unfortunately, I do not have any data of a V/S descent, which is Victor’s proposal for MH370 at 18:40 UTC.
Richard: Very interesting results. What do you conclude about MH370 at 1840? Do you think it had already finished a turn to the south, or was it descending with V/S?
@airlandseaman
My view is that MH370 was descending with V/S.
I have held this view since August 2016.
Quite clearly this was not a standard VNAV descent.
@sk999 studied this some time ago (July 2017) for MH371. Clearly the BFO variation is higher during a typical VNAV descent than during level cruise. That’s why I refer to a possible V/S descent at 18:40, which I think would have a similar BFO variation to level flight for the same level of turbulence. In both cases, the elevator is controlling vertical speed and the autothrottle is controlling airspeed.
That said, we really don’t know what was happening at 18:40. I won’t say it couldn’t have been level flight in a southerly direction. There are many possibilities for the flight pattern between 18:22 and 19:41.
@Andrew, Victor
In an earlier comment, I wrote that this ‘Traffic Log’ “comprises operational messaging internal to various MAS locations and functions, MAS and external service providers, and MAS and the aircraft“.
This messaging application/platform is not solely for ACARS correspondence, it’s routing messages out through ‘gateways’, via other mediums, to other applications. I expect it’s a hosted SITA service, by that I mean it’s hosted in what we’d now call a cloud, SITA did cloud way before Google, Amazon Web Services, and Microsoft Azure came along.
Some of the messages relate to flight planning, routing out to MAS’ flight planning service provider (Sabre-FWZ), others to the flight following service provider (Sabre-Flight Explorer), and the infamous engine reports which are parsed out to local files and likely to be picked up by another process for forwarding to RR.
I’m surprised that the MAS ODC staff were so bamboozled by the Flight Explorer information they were viewing after 17:07Z. In SE Asia, in 2014, Flight Explorer relied on the FMC reports for its position updates.
@PaxLambda, Ventus45
This expanded message Traffic Log shows that the messages routed directly to BKKXCXA, after 18:17Z, were ‘forcing’ a Ground Location of KUL.
AN 9M-MRO/FI MH0370/GL KUL/MA 991IFrom ARINC 620: “For an uplink message that requires delivery to an aircraft not active in the system, DSPs start with the Geographic Locator (GL) or Airport Locator (AP) specified by the ground user (airline or other host computer). Delivery through each possible locator is attempted a configurable number of times.”
The messages submitted after 18:17 all failed with:
- NO ACK 311ADDRESSEE: 9M-MRO
Note that text MAS-L/S/F relates to Message Assurance, not Malaysia Airlines.
@Don Thompson
RE: ”I’m surprised that the MAS ODC staff were so bamboozled by the Flight Explorer information they were viewing after 17:07Z. In SE Asia, in 2014, Flight Explorer relied on the FMC reports for its position updates.”
Didn’t they initially believe the position information in Flight Explorer was aircraft generated, then realised later that it was only projected?
@Don Thompson said: and the infamous engine reports which are parsed out to local files and likely to be picked up by another process for forwarding to RR.
That was something I learned from reading the log. I had (incorrectly) assumed that the service provider forwarded the reports directly to Rolls-Royce. It looks as though they are stored on an MAS server and forwarded to RR.
@Don Thomspon
How does the IFE flight progress map display get the flight path information?
@Dr B said: However, the BFO is highly sensitive to vertical speed because it is uncompensated by the AES. My conclusion is that it is highly unlikely that a descent was ongoing at either 18:40 or at 23:14.
It may be that the data is telling us something very different. I’ve made the argument that for a V/S descent, the BFO variation might be comparable to level flight, depending on the turbulence levels.
Perhaps the (abnormally) low BFO variation at 18:40 is telling us the plane was neither in a VNAV (or FLCH) descent nor flying level. It may be that lowest BFO variation is achieved for a V/S descent. It is the one mode in which the vertical speed is directly controlled. (For level flight, it is the altitude and not the vertical speed that is primarily controlled.) A side effect of that mode is that once the airspeed increases past the MCP speed, the TAS remains nearly constant (which increases IAS on a descent). So V/S is controlled to a constant value, and TAS tends to be constant, both of which steady the BFO, understanding that the BFO is most sensitive to vertical speed, so changes in tailwind/headwind won’t have a big effect.
Why choose V/S as the descent mode? Using V/S for climbs and descents is an acceptable mode, with the understanding that the airspeed might fall in a V/S climb and it might increase in a V/S descent. However, a V/S descent to FL200 at 2500 fpm won’t result in an overspeed, and the pilot might have understood that it was safe to use this mode for these values of final altitude and vertical speed.
@Victor
For a constant track, at a constant TAS in V/S mode controlled by the auto throttle , with constant winds and no turbulence, at a constant GS, at a constant ROD in V/S mode controlled by the pitch, you have a BFO data series with the lowest standard deviation at 1.26. This is even better than SD = 1.76 in the cruise and 1.82 on the ground.
@Richard Godfrey: There was some discussion on PPRUNE claiming that some 747-400 pilots cruise with VS=0 instead of ALT HOLD, the logic being the flight will be smoother at the expense of more altitude error. The claim is also that when turbulence is high, Boeing recommends V/S mode for climbs and descents to reduce pitch changes.
@Richard Godfrey said: For a constant track, at a constant TAS in V/S mode controlled by the auto throttle
I think my statement might have been misunderstood. There is no autothrottle mode that controls TAS. What I mean is, for a V/S descent, once the speed increases past the target (MCP) IAS, the autothrottle goes to idle. (If the speed falls below the target IAS, the throttles advance.) If the (L/D) ratio is approximately constant, and if the vertical speed is held steady by the autopilot, then the TAS will remain approximately constant for an idle descent. That means that the IAS will increase during a descent.
I’ll also add that the abnormally low BFO variation at 18:40 makes me question whether all the BFO values were independent measurements.
@Victor
Thanks for the correction!
I am intrigued by the BFO data series at 18:40 and how come the standard deviation is so low.
@Victor
We did quite some experimentations and simulations around 18h40 to descend to FL270 and understand the a/c behaviour to determine the best ROD. Indeed your statement is in line with our findings:
1- if only a target altitude is selected on the MCP and validated, the a/c tries to reach it within 2 minutes (by default in the FMS). This leads to large vertical speed values close to -3500fpm and raising the a/c speed. To avoid overspeeding, the FMS raises the pitch leading to a kind of phugoid descent. This takes more time than selecting a V/S as below:
2- We found that actually the V/S ~ -2600/2700 fpm is the maximum ROD to avoid the phugoid type of descent. By selecting -2500fpm on the MCP with the new FL, the a/c beautifully behaves and reaches the target altitude via a perfect glide.
Let’s remenber that on that day and in that area the wheather was calm with a windspeed less than 1 knot (i.e. extremely calm 🙂
A smooth descent is thus very possible and in line with small BFO SD.
@Jean-Luc Marchand: I’m not sure what you did in (1). In addition to changing the MCP altitude, to descend you have to choose a mode, i.e., FLCH, V/S, or VNAV (with a change the VNAV altitude). I have noticed that there is what I believe is a bug in the PMDG777 model. If you descend in a mode in which the Mach number is held constant, you easily enter an overspeed condition. I suspect that partly explains what you are seeing in (1).
@Richard Godfrey,
The table of MH371 BFO statistics is enlightening. Thank you for posting it.
@Richard Godfrey,
@airlandseaman,
@Victor Iannello,
Here is what I think we know concerning the BFO variation (“noise”) during periods when 9M-MRO was (a) on the ground, (b) in level flight, and (c) descending. Let me know if I get something wrong.
1. The BFO standard deviation (sigma) on the ground is ~1.8 Hz.
[MH370 at the gate was 1.8 +/- 0.4 Hz. MH371 from landing to the gate was 1.8 +/- 0.1 Hz.]
2. In level cruise flight the BFO sigma is ~1.8 Hz.
[MH370 at 23:14 was 1.7 +/- 0.3 Hz. MH371 at 04:56 was 1.8 +/- 0.1 Hz.]
3. There is no discernible increase in BFO noise in level cruise flight compared to being on the ground (while either stationary at the gate or at taxi speed). This conclusion bounds the vertical speed variation in level flight to be less than ~60 fps 1-sigma, which is negligible in its effect on the BFO variation.
4. There is a large increase in BFO sigma during a LNAV turn, to 3.8 Hz at 04:29 in MH371.
5. There is a large increase in BFO sigma when descending in VNAV in MH371, to 4.3 Hz at -2,055 fpm for ~1 minute, and to 9.8 Hz over a full descent over ~5 minutes.
[See Richard Godfrey’s Table for specific MH371 descent values.]
6. The BFO sigma during the 18:40 phone call is 1.3 +/- 0.2 Hz. This is slightly smaller (by ~2 sigma when compared to the MH371 data) than the value expected in level flight (1.8 Hz), and it is very much smaller than the VNAV initial descent value (4 Hz).
7. It seems extremely unlikely that a either a LNAV turn or a VNAV descent could have been ongoing at 18:40 because the BFO noise then was 3-4X lower than measured in MH371.
8. It has been postulated, based on general principles, that a V/S descent would have smaller BFO noise than a VNAV descent and would behave similarly to level flight. The question is, what is the additional ROD variation (and thus the additional BFO variation), if any, during a V/S descent compared to level flight?
9. In order for a V/S descent to have occurred at 18:40, the ROD variation would need to be less than +/- 60 fpm peak-to-peak. This very small ROD fluctuation is needed in order to make its contribution to the BFO noise negligibly small.
10. MH371 ACARS altitude reports in level cruise flight show fluctuations of at most 20 feet in the pressure altitude from the nominal value. These reports are spaced by 5 minutes, so the time scale of the altitude variation is only known to be shorter than 5 minutes. However, we can look at how fast these altitude changes must occur in order to generate 60 fpm (1 fps) ROD. Assuming a sinusoidal variation, the period would need to be shorter than ~126 seconds to generate at least 60 fpm of vertical speed.
11. At present we have no BFO data, or historical ROD data, to indicate the expected BFO noise during a V/S descent.
12. Therefore, at present we have no measured data with which to test the assertion that a V/S descent is compatible with the 18:40 BFOs. It may be consistent, or it may not be consistent.
13. In addition, it is not clear why a V/S descent might have been chosen at 18:40 instead of a typical VNAV/FLCH descent, even if a V/S descent were deemed safe. (Perhaps Andrew will give us his opinion on this question.)
Oops. Item #3 above should be “60 fpm peak”, not “60 fps 1-sigma”.
@Victor
I’ll also add that the abnormally low BFO variation at 18:40 makes me question whether all the BFO values were independent measurements.
Ive never seen a detailed description of how BFO is actually measured. Maybe you or ALSM can enlighten me. It would seem important to characterize the inherent accuracy of the measurement process. Determining the frequency of an L-band signal to a few Hz with a snapshot measurement is non-trivial.
@DennisW
…also 23:14 BFO starts out near-constant values, then gets a little bump about midway through the series…I assume much higher winds there up to 70+ knots
Dennis: The BFO value is “measured” by the demodulator (Square Peg CUs) at the GES (Perth). The RAW number from the demod is effectively the difference between the measured carrier frequency and the nominal channel center frequency (good to 10^-14). It starts with the “station clock” (rubidium or cesium typical). The whole RF chain is frequency locked (but not phase locked) to the station standard. But the system is not perfect. The GES AFC function in particular introduces errors we have discussed extensively. They have been calibrated to a few Hz. In short, the calibrated system is probably good (accurate) to a couple of Hz.
@ALSM
Thx. I was not conerned about the AFC error which is a slowly varying function. I was more concerned with the error associated with individual measurements. In a “broad” principle a meaurement error of 1Hz would require a measurement duration of one second at the carrier. Demod measurements at lower frequencies would require an even longer time period to get an accuracy of 1Hz at the L-band carrier.
@DrB: Regarding (13), in the past I have asserted that although a V/S descent might not be preferred during a typical commercial flight, the concern about overspeed would not occur with a 2500 fpm descent to FL200.
Here is a possibility why a V/S descent might be preferred. Let’s suppose the intention was to alert the passengers as little as possible that a descent was occurring. As I have stated in the past, a V/S descent would provide a “smoother” ride than an FLCH or VNAV descent. But also, the start of the descent could be initiated in a less-noticeable manner. The MCP target altitude could be dialed down to FL200, for instance, the mode switched to V/S, and the descent rate could be increased over time from zero to 2500 fpm, perhaps in increments of 200 fpm. In addition to the smoother ride, the throttling down of the engines and the pitch down would be much less noticeable than for a FLCH or VNAV descent. If desired, the same thing could be executed for the level out at the lower altitude, i.e., the descent rate could be progressively reduced to make the throttle up and pitch up less discernable.
@DennisW,
You said: “I’ll also add that the abnormally low BFO variation at 18:40 makes me question whether all the BFO values were independent measurements.”
That could be part of the explanation. For instance, the 23:14 BFOs are an average of 2.2 seconds apart, whereas the 18:40 BFOs are an average of only 1.2 seconds apart. Perhaps there is a significant BFO noise component that has a decorrelation time somewhere between 1 and 2 seconds. Then the 18:40 BFOs would be more highly correlated on average than the 23:14 BFOs.
To test this notion I found the sigmas for 18:40/23:14 using two subsets of the BFOs. First I only used BFOs that had a “neighbor” within 1 seond. Second, I only used BFOs that had no “neighbors” within 1 second. The sigmas are 1.0/2.0 Hz at 23:14 for the two neighboring conditions. Thus, it appears in that 23:14 data set that neighbors closer than 1 second are indeed more correlated, because the sigma is only half of what it is for more distant neighbors. The results for 18:40 are puzzling: 1.2/1.2 Hz. That is, at 18:40 the time difference of the neighbors doesn’t seem to matter.
Has anyone figured out why there are many more BFOs at 18:40 than at 23:14, even though the phone call durations are equal?
@DrB
The quote in your post is attributable to Victor, but I had similar concerns. I also pondered the question you raised relative to the number of reported BFO’s at 23:14 versus 18:40.
How can a plane flying a V/S descent have a more steady vertical acceleration (SD 1.2) than a plane at the gate on the ground (SD 1.82)?
@DrB
At 18:40 there were 2 calls simultaneously, hence the larger sample of BFOs compared with 23:14.
@sinux
The BFO is dependent on horizontal speed, vertical speed and track of the aircraft relative to the motion of the satellite. The computation is a 3D relative speed (not acceleration) between aircraft and satellite.
On the ground from landing to the gate there is the horizontal speed and track of the aircraft (even if the vertical speed is close to zero) relative to the satellite.
Even when stationary on the ground, the satellite is still in motion.
sinux asked: How can a plane flying a V/S descent have a more steady vertical acceleration (SD 1.2) than a plane at the gate on the ground (SD 1.82)?
Part of the BFO error on the ground could be due to drift of the fixed frequency bias (FFB), which would be smaller over shorter time intervals. Are the time intervals comparable?
Apologies in advance if I’m going off topic a little here, but this is where I think it would be useful to have sample data from other aircraft with the same equipment. I would have thought it would be possible to understand BFO/BTO errors for aircraft on the ground and following take off from many data sets.
I assume after time the units temperature will stabilise.
What I’m not so sure on is how external factors such as aircraft temperature may affect what we see.
I suppose the point I’m trying to make is do we know everything we need to know in order to accept BFO/BTO values as they are or to apply appropriate warps / filters?
Coming back to the original point of this article, the Malaysian authorities should be doing all they can to get data out in the open.
@Richard Godfrey
Understood, but the satellite didn’t stop moving at 18:40 😉
The fact that the SD is also close to 1.8 in level flight, is an indication, to me, that the minimum system standard deviation expected is close to that figure.
Suddenly getting a better value seems very odd!
Imagine an analogy of a student asked to measure weights.
What would you as a teacher say is the cause of a student bringing back his assignment with a better SD than the other students?
Maybe he used a different scale than the other students (one that has 2 digits precision instead of only 1 for the rest of the class).
Or it could also be that the student got lazy and thought he could game you by inventing values instead of doing all the measurements (and in the process overestimating the precision of the scale).
@Victor
“Finally from MH371 Landing to the Gate for 93 seconds including 132 BFOs gives a SD = 1.82.”
“the MH370 BFOs at 18:40 UTC for 61 seconds including 51 BFOs giving a SD = 1.26”
That’s 61s vs 93s. Do you think FFB would have such an impact on those time scales?
@sinux: Not likely. Thanks for running that down.
@All: I’m delighted to see that the Google Earth image of an aircraft flying over the Cambodian jungle has received zero attention on this blog. This really is a very informed group. Compare that to the “enlightened” discussions on Twitter and Facebook where this is considered to be a real story.
I’ll say that before the story ran, I was contacted by the Daily Star asking for my opinion. I chose to not respond at all. I’ve been burned in the past by making a sensible statement and having it deliberately contorted by a reporter into the exact opposite meaning to sensationalize a story. I did not want my name in any way to appear in this story.
@sinux
Changes in mean and variance are to be expected relative to a process which is not stationary. You cannot draw any conclusions relative to the mean and the variance of BFO samples.
@Andrew asked: Didn’t they initially believe the position information in Flight Explorer was aircraft generated, then realised later that it was only projected?
So now we have an idea of the level of sophistication of the planning and tracking tools used by MAS. Knowing this, how in the world could MAS ODC have thought the plane was over Cambodia? That doesn’t correspond to the flight plan.
@Don Thompson? Why Cambodia?
@Victor
Projected track based on the aircraft’s last known position and track? The following image shows the flight planned track in red. The projected track from IGARI takes the aircraft over Cambodia.
https://www.dropbox.com/s/yncu692gb5t7c82/IMG_1304.jpg?dl=0
@Andrew: Don’t you think a sophisticated tracking tool would take into consideration the filed flight plan instead of just extrapolating forward based on the last known track? What you are describing is no better than how Flightradar24 estimates position after its ADS-B receivers are out of range.
@Victor
I do not know how Flight Explorer works, or what inputs it was designed to use back in 2014. I only suggested that it MIGHT have used the last known position and track to determine a projected position. That would fit with statements that were passed to ATC by MAS. Beyond that, I don’t know.
@DennisW,
Yes, I see now that the quote I attributed to you was actually your quote of Victor’s comment about the possibility of BFO correlation.
@sinux,
@DennisW,
@Richard Godfrey,
@Victor Iannello,
I will try to address the apparent intent of @sinux’s question regarding how the 18:40 BFOs could show lower noise than BFOs in level flight at other times as well as when the plane is on the ground. The short answer is that it is certainly possible if the BFO data are partially correlated over a time scale near 1 second, and this appears likely to be the case.
Consider the BFO reading to vary due to two causes. One cause is random reading noise in the electronics of the system. This random (stochastic) process will generally be stationary. It is stationary when its probability function (generally close to a gaussian exponential curve) does not vary with time. That means its mean value and its variance are constants. I can measure it now and I will get the same answer as I got yesterday.
As DennisW points out, the frequency of the oven-controlled crystal oscillator in the SDU drifts with time. That drift is complicated and unpredictable. The Allen Variance is used to characterize it, and it is not a simple monotonic curve like a gaussian.
The time scales affecting the random electronic “read noise” are quite short – probably several seconds or less. That is, readings farther apart in time than a few seconds are effectively independent, and readings close together in time are not independent (i.e., they are correlated). That could be why the 18:40 BFOs appear to show a smaller standard deviation when one assumes each data point is independent, because those readings are closer together in time. It is possible to correct for the partial correlation of the BFO readings if one already knows the correlation coefficient or correlation function. Alternatively, one can estimate the average correlation coefficient by taking the ratio of the measured close-sample standard deviation and dividing it by the wide-sample standard deviation (c = 1.3/1.8 = 0.7).
Fortunately, the typical OCXO drift times are much longer – on the order of an hour or longer. That large difference in time scales (seconds for read noise and hours for drift) allows us to perform some useful statistical analyses on the BFO readings. Essentially all the BFO subsets Richard and I compared were 1–30 minutes long, so in this case I expect the OCXO drift to be smaller than the read noise. That allows us to assume stationarity and to compute meaningful statistical properties, assuming the BFO behaves like a gaussian random variable on short time scales, which we have done.
Now, back to the question at hand of the low standard deviation of the 18:40 BFOs over a 61 second period of time, what is different/unique about this data set? Well, the average time between BFO readings at 18:40 is 1.2 seconds. At 23:14 it is 2.2 seconds. At the gate at 16:00 it is 117 seconds. For MH371 at 04:55 it is 5.3 seconds. These last three cases, with separations of 2.2/117/5.3 seconds, all had standard deviations near 1.8 Hz. However, the MH371 data set after landing at 07:28 also had a standard deviation near 1.8 Hz, but the average BFO time difference was only 0.71 seconds, which is even shorter than MH370 at 18:40. This one case does not seem to fit the others. Perhaps the read noise depends on the channel type and on the channel unit ID. We may not have enough BFO data to sort that out. I will note that if one takes a subset of the 07:28 MH371 BFOs, in particular the R-Channel Rx only, the standard deviation drops to 1.26 +/- 0.28 Hz. These particular BFOs occur in rapid bursts with sub-second spacing in groups spaced by several seconds, so characterizing the correlation function is more complicated than one can unravel in a minute or so of data. Still, it does seem likely that rapid BFO readings in one channel show a smaller standard deviation because of some low-pass filtering effect in the electronics that correlates readings made close together in time. That effect is probably contributing significantly to the lower measured BFO noise at 18:40.
@DrB
RE: “13. In addition, it is not clear why a V/S descent might have been chosen at 18:40 instead of a typical VNAV/FLCH descent, even if a V/S descent were deemed safe. (Perhaps Andrew will give us his opinion on this question.)”
An enroute cruise descent is typically flown in VNAV, which provides a descent rate of approximately 1,250 ft/min, with the speed controlled by elevator and the descent rate controlled by thrust. VS mode is sometimes used for small altitude changes, or for situations where the pilot wants to descend at a reduced rate. The use of VS mode for a large altitude change would be unusual, but certainly not impossible. VS mode is also recommended by Boeing in cases of severe turbulence, but that does not seem likely in this case.
The pilot might elect to use VS mode if he wanted to slowly increase the rate of descent, as Victor suggested. That would make the attitude and thrust change much less noticeable to the passengers. If the intent was to avoid alerting the passengers, he might also need to manually control the cabin altitude to prevent pressure changes in the cabin, and descend at a constant IAS to prevent changes in the air noise.
@Andrew
Why is the cabin manual pressure control needed? Wouldn’t that be on auto? Do pilots normally intervene with pressure or are you saying as special idea to handle a faster than usual descent?
@DrB
Good post. The measurement noise is stationary as you point out. The BFO noise is comprised of that and a component due to the oscillator. The oscillator component is similar to a random walk and that part of the variance will grow with time (the number of samples root(n)). Yes, I agree the temporal separation of the samples is another factor complicating our understanding.
@TBill
The outflow valves would normally be left in AUTO during descent. However, the pressurisation system would then decrease the cabin altitude during descent, resulting in pressure changes that the passengers/cabin crew would feel in their ears because of the difference in pressure between the outer and middle ear. If the pilot was truly intent on masking the descent he might consider taking over manual control of the outflow valves to keep the cabin altitude constant. Mind you, that’s easier said than done!
@Andrew: For a descent from FL340 to FL200, do you really think there would be much of a change in cabin altitude?
@Victor Iannello
@Andrew
Re The Cambodian Conjecture.
Short of knowing what version of Flight Tracker (and what filters and layers) MAS were using on the night in question and interviewing the relevant operations staff we’ll never know exactly what transpired to give rise to the ‘somewhere over Cambodia‘ report. I’d make the following observations though:
1. The phrase somewhere over Cambodia‘ appears to have originated from KL ATCC not MAS Ops. There is no record of MAS Ops actually stating to KL ATCC that MH370 was ‘… still flying … somewhere over Cambodia‘; that’s just how KL ATCC reported the conversation to HCM ATCC at 1804:39 UTC. The KL ATSC Duty Watch Supervisor stated that MAS Ops had informed him (incorrectly) that their Flight Tracker showed the ‘… aircraft in Cambodian airspace‘. That was then relayed to HCM ATCC as ‘… somewhere over Cambodia‘; equally incorrect but not the same thing.
2. Flight Tracker is a Sabre product and it does show the flight’s planned route per the lodged flight plan. That is evidenced by the 1835 UTC exchange with KL ATCC where MAS Ops placed MH370 at N14.90000 E109.15500 at 1833 UTC; that location matches the flight plan for that time. The MAS Ops staff had mistakenly read off the half-hourly projection as being an actual position report.
3. You can get a feel for the sort of map and flight route displays provided by Flight Tracker from its website. Almost all the map displays I have come across are devoid of country name labels. When MAS Ops advised the KL ATSC Duty Watch Supervisor that Flight Tracker showed the ‘… aircraft in Cambodian airspace‘ MH370’s projected position would have been in Vietnamese airspace. Perhaps the MAS Ops person didn’t know the difference between Vietnam and Cambodia.
4. I don’t know how MAS had been routing their KL-Beijing flights in the period leading up to 8 March 2014 but a quick check of their flights in recent times shows that some of them are being routed via Vietnam in a similar fashion to MH370 (BITOD L637 TSN W1 BMT W12 PCA G221 BUNTA) whereas others are being routed via Cambodia (what looks like BITOD M755 PNH B329 VIBUN R575 PAPRA A1 BUNTA).
Needless to say none of the involved organisations (Vietnamese ATC, Malaysian ATC and MAS) seemed to have fielded their A teams that night. Consequently none of them exactly showered themselves in glory.
@Mick Gilbert: Your comment implies that the MAS Operations had the tools to properly track the flight according the flight plan, but made an error. That could be. There seems to have been a lot of errors.
re: Pilot Repeated “FL350” Transmission @0107 (MYT) Implication
In the “MH370 SIR slides” – diagram “History of the Flight”, it specifically highlighted the repeated “FL350” transmission @0101 (MYT) and @0107 (MYT) respectively.
Question:
Is this “repeated transmission” a typical/normal behaviour?
To the extent that it is somewhat atypical/abnormal, can one propose any ‘meaningful’ implication from this? [Besides the explanation ‘meaningless’.]
Observation :
Notice that the repeated transmission happened at the same time as the first/last ACARS report. Are the two events correlated?
Suggestion/Conjecture:
Maybe the pilot (from his experience) saw/sensed something ‘abnormal/atypical’ (say on the display screen [in a flash]), it was not significant enough to raise an alarm but as a ‘safety’ measure to ensure everything’s normal, he chose to contact ATC to ‘test’ the [communication] system …
@Victor
Re: BFO’s 18:40 and 23:14 satellite phone calls
Another issue is the calibration of Channel C, which Holland seems to take as a reliable reading at around 216 for 23:14.
The passive flight path proposals are not hitting that 23:14 BFO very well, unless liberty is taken to omit or manually adjust to a higher value. To me the 38 South proposed paths fail around about Arc5, where the lack of BTO for 1.5-hrs leads to very ambiguous forward path projections (if one assumes the pilot could still be active).
I presume it means an apparent 37-38 south flight path can divert to north of 25 South if so desired by the pilot. I think of it as heading to a trench below Zenith plateau, but a chilling recent thought is heading northeast to simulate heading into Beijing re: sunrise.
@mash
If a problem occurred with the aircraft, sufficient to cause a departure from the flight plan and turn back toward Malaysia, it is not compatible with the aircraft continuing for 7 hours until fuel exhaustion.
Radio check is standard practice, but it is announced as such “requesting radio check”, not as a repeat “maintaining flight level” message.
It’s interesting that we have evidence of a low-bank turn around Penang and a possible V/S descent at 18:40. Both of these are consistent with maneuvers that would be less noticeable to passengers and crew than a typical turn and descent.
@TimR’s understanding is that the plan was for the flight to proceed without the passengers’ awareness of the detour at IGARI. (Of course this flies in the face of sharp turn at IGARI, which I suspect was never captured by military radar.) Perhaps these observations are all related.
@TBill: I don’t know what you mean about the problem with the BFO at 23:14. My automated flight past Cocos Island has very low (<1 Hz) BFO error at 23:14, with no adjustment of the BFO value. Considering that some flights have low BFO error at 23:14, I see no justification for adjusting this value in evaluating and reconstructing paths.
@Victor
I was not targeting your recent path.
Good though, it means your path may contain some of the behavior needed to match 23:14 BFO, which is some change heading etc.
@mash
Did you think about something like this:
https://www.aeroinside.com/item/8810/canada-b772-near-fort-mcmurray-on-dec-18th-2016-audio-control-panels-dont-like-fluids
Andrew, Victor, and Mick.
When did “Cambodia” enter the discourse? That question also relates to the initial entry of MAS Ops into the discourse.
At 1758:37Z, during a telephone line conversation between KL ACC Sector 3+5 Planner and HCM, KL ACC committed to call MAS. That call to MAS Ops is not recorded or transcribed.
At 18:04, during another telephone conversation between KL ACC RADAR and HCM, the KL ACC party relates, “company Malaysian Airlines the aircraft is still flying is over somewhere over Cambodia.”
I agree with Andrew that an extrapolated path, using the heading and lat-lon from the 1707Z final FMC position report, would track over Cambodia. However, the processed and filed flight plan puts the aircraft over Ho Chi Minh City, or slightly east, at 1757Z.
At 1833:50Z, another telephone conversation began between KL ACC RADAR and MAS ODC. MAS ODC appears confident the aircraft is still flying. While the ODC did not get a reply to their earlier message to -MRO, they claim “ the… message went through successfully“. MAS ODC quotes a specific location of 14.9º 109.155º. MAS ODC reiterates “Aircraft still sending the … movement message“. The position is consistent in time and for the planned and filed route, i.e. where they expect -MRO to be. KL RADAR ACC officer asks MAS ODC to renew attempts to contact the aircraft.
But the MAS ODC ‘URGET REQUEST’ message had failed via all possible paths at 1819:34Z.
Later, the DCA Watch Supervisor’s Log Book records that the positions were projected. Apparently, “at 1930 UTC [0330 MYT] MOC called in and spoke to the Radar Controller, “…admitting that the ‘flight tracker’ is based on projection…”
No transcript of this conversation between MAS ODC (aka MOC) and the Radar Controller. No follow up to the request for renewed attempts to contact the aircraft, presumably the SATVOICE call and the “DEAR MH370” missive.
Over SE Asia, Flight Explorer derived position using the operational comms from the aircraft. Over US, AirServices Aus, NZ, and Eurocontrol airspace, Flight Explorer augmented this information with the ANSP surveillance feed. In 2015, Flight Explorer added Planefinder’s ADS-B feed. I’m mystified that the ODC staff were not aware of how Flight Explorer derived its data, the mechanism of updates in various operating regions, and what was extrapolated vs confirmed track. Flight following is an explicit requirement by the regulator.
@TBill asked “How does the IFE flight progress map display get the flight path information?”
Airshow involves the PIIC, the passenger in-flight information computer. The PIIC has (unidirectional) A429 data bus connections from AIMS over which relevant FMC and ADIRU words are passed.
Don’t forget the cabin IFE components will be disabled if the Main AC Busses are isolated, or the IFE/PASS SEATS switch is released.
Victor:
Re your point above: “It’s interesting that we have evidence of a low-bank turn around Penang and a possible V/S descent at 18:40. Both of these are consistent with maneuvers that would be less noticeable to passengers and crew than a typical turn and descent.”
I understand your point, but it is not even remotely consistent with the 35 deg bank, climbing turn to 43,000 feet at 17:21-17:30, slowing down about 60 kts in the process, then accelerating at full throttle to 500 kts. These figures might not be exactly correct, but the turn back at 17:21 was the opposite of what happened at Penang and 128:40 as far as being noticeable by passengers.
@mash: On Twitter, @Edward_767, a 767 captain, has closely studied the audio recordings of the MH370 radio transmissions. He is convinced that the second altitude call at 17:07:56 was NOT made by the captain, but by the first officer, possibly because the captain was not in the cockpit at that time. His theory is that the first officer made the call because he was not sure whether the captain had made the previous altitude call at 17:01:17.
He makes another interesting observation regarding the final exchange where MH370 is handed off to HCM. He claims the typical radio procedure would be to receive the new frequency (in this case, 120.9 MHz), dial it into the radio as the standby frequency, read back what frequency was dialed in as part of the confirmation of the handoff, listen for the final transmission, hit the switch which swaps the active and standby frequencies, and make the call on the new frequency.
In the transmission that acknowledged the handoff to HCM, the captain did not read back the frequency. Ed believes this is because the new frequency was never dialed into the radio, as the captain never anticipated making a call to HCM ATC. Ed believes the missing frequency during the acknowledgement of the handoff is yet another red flag.
@mash,
My take on the slightly unnecessary “maintaining FL350” calls were that they might be a gentle reminder to ATC that they were approaching the boundary. Other controllers might normal respond with “Roger, at IGARI contact 120.9”.
It would be good to hear from any controllers working this sector as to what is normally said.
@airlandseaman: You assume we have military radar data of the turn at IGARI. I question all the military radar data, but in particular, the record of that turn. As for a full throttle climb, even if the TAS dipped, I don’t think a pitch up and full throttle climb would be perceived as odd during that part of the flight.
Also, the low available thrust at high altitude would mean the acceleration would be very low and barely perceived.
@Tim: At 17:07:56, MH370 had a contact with ATC just six minutes before and was still 14 minutes from IGARI. It is hard to imagine why at that point they would need to “gently remind” ATC that they were approaching IGARI.
Victor: No matter how you try to explain it, the turn at IGARI was anything but subtle. That turn would have been noticed by everyone on board, even if they had been dozing off. Thus, I think there is zero probability that the turn at Penang, or the turn/descent at 18:40 had anything to do with an attempt to keep passengers and cabin crew in the dark.
@Don Thompson
…the PIIC is a new revelation to me. Do we have a way to access the output and/or turn that PIIC off?
@mash
…at the moment I am temporarily adopting Ed Baker’s hypothesis via voice analysis that the second call may have been the FO’s voice, and FO might have called in “maintaining FL350” due to not recalling if the Captain had made the initial call. In that scenario, Captain is presumably temporarily out of the cockpit perhaps for clandestine reasons.
And an important point which Is omitted from any report. That is the quality of the last 3 transmissions from MH is lower. I think this is because they are now using the ‘hand mikes’ and not the headsets. Something that is perfectly normal once in cruise. But I think it needs a mention anyway.
@airlandseaman: As an exercise, ignore the military data, including the inexplicable discrepancy with the civil radar data near IGARI (the flyover versus the flyby) and the impossibly sharp turn that has been graphically depicted. You are left with the final SSR point and the first PSR data point. I submit it is possible to connect those two points in time and space without a sharp turn.
@Tim: Or perhaps the S/N ratio is degrading due to increasing distance from the VHF ground antenna.
Victor: There is simply no way to gain 6000 feet and reverse course within a few minutes and not have that be noticed. Very different from the turn at Penang.
@airlandseaman: I didn’t say not noticed. It is a question of whether it would be perceived as out of the ordinary at that point in the flight. It would just be a turn and a climb.
Victor: I think we are loosing sight of my original point, which was that I do not believe we can assume V/S mode was used instead of VNAV at 18:40 to reduce passenger awareness of what was going on.
As you know, my STDEV calc’s of the MH371 descent into KL on the prior flight show that the BFO STDEV during a known VNAV descent circa 07:05 were quite low (1.7Hz), about the same as level flight. The bottom line is, there is no evidence that V/S mode was used at 18:40, nore is there any consequence. Either mode could have been used (if there was a descent at all) and the BFO STDEV could be the same.
@airlandseaman: When did I assume there was a V/S descent at 18:40? I’m not sure there was ANY descent at 18:40, although I think it could explain a lot. If there was a descent, I propose a V/S descent as the mode that will minimize the BFO variation. If the BFO variation is consistent with FLCH or VNAV, I don’t have a problem with that. What I disagree with is the claim that a descent could NOT have occurred at 18:40 due to the low BFO variation.
Victor: I did not ever “…claim that a descent could NOT have occurred at 18:40 due to the low BFO variation”. I agree that a descent may have been in progress at 18:40, or it could have been a turn to the south as we originally proposed here:
http://bit.ly/2MfoHVj
It remains ambiguous. I do not believe any of the recent discussion about the STDEV of the 18:40 BFO values provides any evidence one way or the other. We may never know until the MH370 debris field is found.
@airlandseaman: Yes, we are in agreement that the evidence neither rules out nor confirms that there was a descent at 18:40.
@ALSM
I think I see what @Victor is saying, which is a hypothesis (thought exercise) that the PAX were being fooled into thinking they were on a flight to Beijing.
Possible Elements of Decoying PAX-
(1) Somehow IGARI turn has to be justified to the PAX or made smoother
(2) Smooth turn at Penang and hope PAX think it is Vietnam out the window? Or maybe tell the PAX resuming path to Beijing after hold??
(3) Smooth descent or turn at 18:40 with control of cabin pressure
(4) (My add) possible Northeast track at end to simulate approaching Beijing during twilight/sunrise
(5) PAX are alive in this scenario; FO is ???
@TBill: Although I believe the hard facts are most consistent with a deliberate diversion under the control of the captain, I have to admit that I am biased towards believing the diversion began with the intention to deliver the patients safely to an alternate airport. I am not a fan of the scenario in which there was a controlled depressurization over Kota Bharu, but again, I freely admit that could just be my bias. Despite my bias, I’m trying to keep an open mind.
The first officer could have been incapacitated by the captain. For instance, upon returning to the flight deck, the captain could have injected him from behind with a sleep-inducing drug.
After a deliberate diversion even if the captain safely landed somewhere with passengers alive, it would be the last flight in his life.
@Victor
You stated “On Twitter, @Edward_767, a 767 captain, has closely studied the audio recordings of the MH370 radio transmissions. He is convinced that the second altitude call at 17:07:56 was NOT made by the captain, but by the first officer, possibly because the captain was not in the cockpit at that time.”
Prof. Dr. H.J. Künzel a speech scientist and forensic expert of audio analysis wrote a report included in the RMP report in the ‘Folder 6 Audio and Other Records’, where he concludes that all cockpit voice communications with LUMPUR RADAR including both the “maintaining level three five zero” messages at 17:01:17 UTC and 17:07:56 UTC were spoken by Zaharie Shah, with high likelihood ratios (LRs) and not Fariq Hamid where low LRs were encountered.
An extract from the conclusion of Künzel’s report “the speech segments contained in ATC DELIVERY, LUMPUR GROUND and LUMPUR TOWER originate from the co-pilot Mr. Fariq whereas the speech segments contained in LUMPUR APPROACH and LUMPUR RADAR (AREA) originate from captain Zahari. (sic)”
An extract from the analysis on Künzel’s report “For the two sets LUMPUR APPROACH and both models of LUMPUR RADAR (AREA) the distribution of LRs is almost reversed. Matching the respective unknown test audios with the voice of captain Zahari produces high LRs, also for the two subsets of LUMPUR RADAR.”
I would rather trust the expert on forensic audio analysis, than Captain Edward.
@Greg: No doubt that would have been understood.
@Tim,
@Richard Godfrey,
@Victor Iannello,
Identifying the speaker of the last three radio calls is difficult for several reasons. Edward Baker may be right or he may be wrong that the second call was made by the FO. I took a look at the recorded background noise, and the second call is different from the other two. To me that means that either it used a different microphone, or the microphone was in a different location that changed the background noise. I am not aware that any audio expert has analyzed the background noise. They tried to use the speech pattern to identify the person speaking, but there may be other features of the recordings, such as the background noise, that can provide additional clues.
@Richard Godfrey, @DrB: I proposed Ed’s theory as his, not mine. I am not a voice identification expert, and neither is he. I will say that after listening to the audio sample that Ed assembled, to my untrained ear, the second altitude call does sound like a different speaker than the first.
I encouraged Ed to contact Dr Kúnzel and solicit a comment on his work. Dr. Kúnzel initially said he would get back to Ed, but never did.
Regardless of who made the second altitude call, the missing read back of frequency in the handoff acknowledgement transmission could be significant.
@airlandseaman,
You said: “As you know, my STDEV calc’s of the MH371 descent into KL on the prior flight show that the BFO STDEV during a known VNAV descent circa 07:05 were quite low (1.7Hz), about the same as level flight. The bottom line is, there is no evidence that V/S mode was used at 18:40, nore is there any consequence. Either mode could have been used (if there was a descent at all) and the BFO STDEV could be the same.”
You also said, regarding the possibility of a descent at 18:40: “I do not believe any of the recent discussion about the STDEV of the 18:40 BFO values provides any evidence one way or the other.”
@Victor Iannello,
You said: “What I disagree with is the claim that a descent could NOT have occurred at 18:40 due to the low BFO variation.”
Let me disagree with Mike and add a caveat to what Victor said.
Richard Godfrey’s analysis (posted here) of the MH371 descent showed a 4.3 Hz BFO standard deviation from 07:17:54-07:19:08. I have independently computed this, and I found the same number as Richard – a very high 4.3 Hz. Mike, I don’t know where you got 1.7 Hz for the MH371 descent. Would you please give the time range for calculating your number. It is important that the elapsed time be on the order on 60 seconds so that it is comparable to the 18:40 phone call duration. By the way, there are no data between 07:07:21 and 07:17:54, so I don’t see where you got the time of “7:05”.
Using about a minute of MH371 data, I think you will find that the VNAV descent has substantially increased the BFO variation compared to level cruise and to the 18:40 data, as Richard said. Assuming no counter-example is found by Mike, then I think we can say that it is quite unlikely that a VNAV descent was underway at 18:40. That is my quibble with Victor’s statement. I think we can say that a VNAV descent most likely did not occur then, but we cannot rule out the possibility of a V/S descent then. As Andrew has pointed out, a V/S descent is possible but not typical. Normally it is used when lower than 1,250 fpm ROD is desired, which is much lower than the ROD needed to match the 18:40 BFOs assuming the 297 true track was still being used then.
@Victor, @DrB
Here is another excerpt from Künzel’s report regarding the background noise during the second altitude call:
“The fact that the noisy parts (subset 4_5_6) of LUMPUR RADAR) produce a relatively low LR (27.85) can be explained easily by the extreme level of background noise present in speech segments nos. 5 and 6 of this test audio.”
Künzel still concludes it was Zaharie Shah despite the background noise.
It really is worthwhile reading the whole report from Künzel and the methods employed and the results obtained.
Erratum: In my last post 07:07:21 should read 07:04:21. Sorry for the typo.
@DrB: MH370 was not “typical”. I believe a V/S descent to FL200 at around 2500 fpm would not have been unsafe, and the descent would have been less discernable to passengers for the reasons I mentioned above, if a descent did indeed occur.
@Richard Godfrey,
My point was simply that there appears to be either a different microphone used or the call was made from a different location. That may or may not imply something about which person was speaking. I don’t think Kunzel did a comparative spectral analysis of the background noise, although he did note level differences.
You said: “Künzel still concludes it was Zaharie Shah despite the background noise.”
My reading of Kunzel’s report suggests he analyzed, and found a Liklihood Ratio (LR), for the last 3 recordings as a group. I didn’t see any evidence that the last 3 recordings were analyzed individually. Kunzel concluded that it was likely that the Captain spoke those last three times, although the LR was only 1.47 for the Captain and 0.88/0.66 for the FO. The standard for a high probability result is about 2.5, so the results are a bit shaky. I wonder if the low score could be due to only 2 of those 3 messages being spoken by the Captain, in addition to the high background noise.
@DrB
I read the report differently. In table 2, the Likelihood Ratio for the LUMPUR RADAR group is high for Zaharie and for the subset of interest is still 27.84. In table 3, the Identification Score for the LUMPUR RADAR group is still high at 2.69 and for the subset of interest the 1.47 is marginal, but still above 1.
For Fariq, the Identification Score for the subset of interest is 0.88 and below 1.
It is more likely, that despite the noise and the brevity of the sound bites, the speaker of the LUMPUR RADAR messages is Zaharie, including the subset of interest. This is what Prof. Künzel concludes.
@Richard Goidfrey,
To clarify, the numbers I listed above are the “Identification Scores” from Künzel’s Table 3 (not the LRs, which are given in Table 2).
Bobby:
I do not disagree with your MH371 BFO STDEV math or Richard’s. But I don’t think the data below 10,000 feet should be used for this type of analysis. The samples you and Richard used (07:17:54-07:19:08) were all below 10,000 feet I believe. Surely, that is lower than MH370 at 18:40. And I do not see any reason to assert that one must have data spanning over a minute for valid results. That would be desirable, but not absolutely required IMO.
I checked the BFO data for the MH371 descent into KL using available data between ~36k-10K ft to limit the samples to those that were less likely to be contaminated with boundary layer turbulence. Here are the results.
Time period Delta T Altitude Decent rate #samples STDEV
07:04:12 to 07:04:17 5 sec ~36100 ft ~2000 ft/min 16 1.7 Hz
07:17:59 to 07:18:05 6 sec ~10000 ft ~2000 ft/min 30 1.7 Hz
All of the above were T1200 channel data. Looks like the STDEV in NAV can be fairly low, at least some of the time.
Whether 46 samples is enough for a valid estimate or not, I do not believe we can rely on data below 10,000 feet for this type of analysis.
@ALSM
The time period between samples in your results above is well below the typical time period for oscillator random walk to be a factor (which begins at sample periods of ~100 seconds for oscillators of the type used in the AES). I think the 1.7Hz STDEV is almost pure measurment noise which is consistent with an earlier estimate you made.
Dennis: The time between samples in the two groups I used is the same as (most of) the samples in the 07:17:54-07:19:08 period. I agree it is basically measurement noise. That is exactly the point. The STDEV shows no indication that the VNAV controlled descent is any more noisy than level flight, as long as you are not descending through low level turbulence, which is much more likely below 10,000 feet.
@Richard Godfrey
With respect to my assertion that Fariq made the 2nd FL350 call, I don’t blame you for siding with a voice expert of Dr. Künzel’s caliber.
I simply point out there are several unusual aspects regarding that particular call. The fact that it was made at all was noted as unusual in the Final Report. It also lacked the characteristic “aaa” that Zaharie’s speech pattern had for all the other calls. I conclusively demonstrated that his “aaa” has the same spectral frequency pattern from the YouTube clip with his speech segments in it. The 2nd FL350 call did not have that pattern. All the other airborne calls did.
The microphone was quite obviously different than the call before or after (the last call). It has a cleaner (less noise) display when the speaker made no sound. It’s possible the speaker was on headset for the previous call, switched to a hand microphone for the 2nd call, then used the headset for the last call…but in my aviation experience I judge that unlikely.
The pacing of the 2nd FL350 speaker is slightly different for the callsign “370”. It’s subtle, but it’s there. The voice is a different quality just from a spectral frequency analysis.
All of this is important because it strongly suggests that if normal SOPs were followed, particularly on a training flight, Fariq had NO reason to make that call unless Zaharie was not in the cockpit. I say this for 2 reasons. First: if Fariq could not recall if the FL350 check in had not been made, he should have simply asked Zaharie if it was done. Zaharie, by tradition and assigned role, would have again made the call–or confirmed it was made.
Second: Under the training circumstances, it is unlikely that Fariq would have simply keyed the mike and made the check-in that was accomplished 6 minutes earlier.
That 2nd FL350 call was made just as MH370 approached the coastline. It is easy for me to envision young Fariq unexpectedly alone in the cockpit, not recalling if the required report had been made. So just to ensure it had been, he made it as a mark of a good airman.
It is an unpleasant scenario to consider if Zaharie left the cockpit to prepare for the diversion he knew would be coming, but I suggest the 2nd FL350, spoken by Fariq, offers a clue.
@airlandseaman,
The BFO standard deviation you measured over 5-6 seconds during the MH371 descent is probably another good estimator of the BFO read noise. The 1.7 Hz (+/- 0.26 Hz) you got is statistically equal to the ~1.75 Hz estimated previously on the ground and in level flight, even over longer timescales of minutes. I conclude from this that the aircraft motion contribution to the BFO read noise is essentially negligible when the aircraft is stationary or in level flight or during a VNAV descent over a brief 5 second period.
That is not the case in a VNAV descent for time scales longer than 5 seconds, say about 1 minute. It seems that the ROD in a VNAV descent is indeed fairly stable over a time span of ~5 seconds. That’s why you see BFO variation matching the read noise when you consider short time spans of BFO data. However, when you consider longer time spans, the BFO variation increases substantially. For instance, using the MH371 BFOs from 07:03:22 – 07:04:21, that is 1 minute of data while passing through 36,200 feet altitude, the standard deviation is 6.3 Hz.
Thus, while there may be larger turbulence and therefore higher ROD variations below 10,000 feet, as you said, even at 36,000 feet the MH371 BFO standard deviation is about 4X larger over 1 minute than it is over 5 seconds. That demonstrates my point, that the 1-minute long 18:40 BFOs show a variation that is consistent with level flight and inconsistent (by a large factor) with a VNAV descent.
As DennisW points out, one minute is too short for the extra variability to be caused by oscillator drift. Therefore it seems the extra variability is most likely due to aircraft ROD variations.
Victor argues that the ROD variations during a V/S descent may be small enough not to noticeably degrade the observed BFO variation compared to level flight. He may be right, and the effect is certainly real and significant, but we don’t at this point have ROD or BFO data during a V/S descent at ~2,500 fpm to confirm it.
BOBBY: My point is that the available MH371 data is insuficent evidence to concude anything about what happened at 1840. It certainly does not prove anything about the AP mode, much less the cause of the change in BFO trend (descent vs. turn).
re: Co-pilot Mobile Phone Re-switched On Implication
Assuming:
A. Mobile phone was switched off before diversion.
B. Mobile phone was re-switched on (by co-pilot) after diversion.
This means co-pilot was still concious after diversion.
Further assuming:
C. This was a pilot (alone) triggered control take-over.
Question:
1. Why did pilot choose “lock-out” option instead of “knock-down” option?
2. Provided (one accepts) that “lock-out” is [potentially] an inferior/impractical option, one wonders whether assumption C above is correct or not?
Suggested “Lock-out” risk/disadvantages:
– co-pilot might not agree to leave cockpit (even if suggested).
– in fact, no really justifiable reason to ask co-pilot to leave cockpit (unless voluntarily himself).
– frequent radio communications before diversion provide relatively little opportunity to (convince one to) leave cockpit (even voluntarily).
– cabin usually relatively ‘busy’ at this stage (say shortly after seat-belt light off).
– co-pilot “lock-out” might provoke cabin alert/disturbance (too early).
– co-pilot/crew might trigger cockpit door password access alarm (if any? — say during [last] radio communication).
[“Lock-out” is also not possible (logically) if co-pilot has the habit of leaving mobile phone in cockpit after boarding.]
@mash
I presume the lock out theory is the more likely one based on how simple it is, particularly in a culture where captain is king. The captain merely needs to tell the FO to go to the cabin [insert reason], and it is done.
In my Zaharie-visits-the-MEC scenario (the reason FO Fariq is by himself and makes the 2nd FL350 call), once Zaharie starts pulling circuit breakers, there will be EICAS messages popping up in the cockpit display. Upon return to the cockpit, Fariq reports the messages, and Z sends him on a mission to the cabin to “investigate.”
I also surmise that if it is true Z left the cockpit, that leaves a young Fariq alone in a B777 cockpit, and it could be too tempting not to capture that moment with a selfie. He would have pulled out his phone and powered it back on the take the image. Afterwords, when Zaharie returns to the cockpit and orders Fariq out, the mobile phone is left above the glareshield next to the window. I’ve done something similar myself.
I realize that this is all simply conjecture and could never be proven right now. Only recovering the CVR/FDR (and having recoverable data on it) will help us understand the series of events that night.
@Don Thompson
RE: “I’m mystified that the ODC staff were not aware of how Flight Explorer derived its data, the mechanism of updates in various operating regions, and what was extrapolated vs confirmed track. Flight following is an explicit requirement by the regulator.”
I agree; it does seem odd that the MAS ODC staff were not initially aware the information presented by Flight Explorer was ‘projected’ rather than ‘actual’. I wonder if the software version they were using at the time made the difference between the two obvious, or did the ODC staff have to go looking deeper into the system? Even if it wasn’t obvious, surely they should have taken a closer look after the initial enquiries from ATC, instead of blindly accepting the displayed position as real.
Flight ‘following’ is not the same as flight ‘tracking’, although they are obviously related. In some jurisdictions, the flight dispatcher shares with the aircraft commander the legal responsibility for operational control aspects of the flight. To that end, the dispatcher ‘follows’ the flight from beginning to end; the minimum requirement is to monitor the flight time remaining, destination and alternate weather trends, enroute weather, NOTAMS, etc and pass the relevant details to the aircraft. In the past, there was no regulatory requirement for the dispatcher to ‘track’ the aircraft’s position in real time, however, flight tracking systems (eg Flight Explorer) have been introduced to improve the flight following process and provide greater situational awareness. Such systems will become mandatory later this year, with the implementation of ICAO’s normal aircraft tracking initiative.
@airlandseaman
You stated “The samples you and Richard used (07:17:54-07:19:08) were all below 10,000 feet I believe.”
… and …
“Time period Delta T Altitude Decent rate #samples STDEV
07:04:12 to 07:04:17 5 sec ~36100 ft ~2000 ft/min 16 1.7 Hz
07:17:59 to 07:18:05 6 sec ~10000 ft ~2000 ft/min 30 1.7 Hz”
The ACARS report at 07:04:04 UTC states that the Altitude was 36,234 feet and the ROC was -753 fpm.
The ACARS report at 07:19:04 UTC states that the Altitude was 8,699 feet and the ROC was -2,055 fpm.
I do not believe the descent rate for your first selection between 07:04:12 to 07:04:17 was ~2000 fpm. That is why I did not select this timeframe.
The timeframe I selected from 07:17:54-07:19:08 was at ~2000 fpm according to the ACARS data and the Altitude at the start of the timeframe was ~11,000 feet and at the end of the timeframe was ~8,600 feet.
There is a standard holding point at 7,000 feet on the STAR NIPAR THREE BRAVO ARRIVAL, which MH371 had not yet reached. At 07:19:04 UTC MH371 was still at 2.818°N 102.031°E at 8,699 feet descending at -2,005 fpm on flight route W533. In another 50 seconds, MH371 would reach 7,000 feet and level off for the final approach. The winds were light and the weather was good. There was no precipitation in the area.
I do not agree that the 164 BFO samples in this 73 second timeframe and a SD = 4.33 “is insuficent evidence to concude anything about what happened at 1840.” In my view the MH371 data shows that MH370 was not in a LNAV turn or a VNAV descent at 18:40 UTC.
I agree the case for a V/S decent is not proven or refuted, without further relevant BFO data.
@Andrew
@Don Thompson
@Victor Iannello
The SIR makes the function of Sabre Flight Explorer clear on p.326;
‘Note: Flight Explorer is a computer-based system which is also known as “Flight-Following System” to track aircraft based on input of the aircraft’s Flight Plan data into the computer. The Flight Plan data generates the flight profile and position of the aircraft and updates every 30 minutes. However, the system does not provide real-time tracking.‘
For a sense of how well MAS ODC staff understood the functionality of Flight Explorer point (3) on the same page is salient;
‘ To understand how the Flight Explorer works, the [MOT Annex 13] Team requested for a copy of the Flight Explorer User Manual and was informed that there was none in the office. Later, a copy of the Flight Explorer User Manual was provided to the Team.‘
So, the MAS ODC couldn’t lay their hands on the user manual for a piece of key, if not core, technology.
Add the bad information coming from the MAS ODC to the fact that the KL ATSC Duty Watch Supervisor (aka Rip van Absent) was, if not asleep, most assuredly not supervising his controllers for a good three hours as events played out, it’s actually surprising that the Distress Phase was declared as early as 2232 UTC.
So, Victor, when you said earlier, ‘There seems to have been a lot of errors.‘, when you get the ball rolling with the 19 minute delay (as opposed to the permissible 5 minutes) in HCM ACC querying the location of MH370 after it failed to report in, yes, there were; multiple and manifest errors by multiple agencies.
@Victor
Re: Skipping-on-Water Crash (FS9)
Accidental flight sim case, FS9 was flying circles around Penang due to loss of autopilot due to fuel exhaustion.
Upon hitting water surface in level gliding flight (~5% bank) the aircraft first bounces/skips off the water surface (possibly losing a damaged flaperon), then back goes up in the air, and finally comes back down for a more nose-in final crash.
I only mention it because double-impact crash might help explain the conflicting debris evidence.
@TBill
What is FS9 and what do you mean be “Re: Skipping-on-Water Crash” please?
I thought we had established from the BFO satellite data and the Boeing end of flight simulations that the last seconds of MH370 were a descent at greater than 15,000 fpm with a single catastrophic impact. This is in contradiction to a double impact, that you appear to be suggesting. What supporting evidence do you have for your theory, or what conflicting evidence do you see the debris provides?
@Richard
OK I made a 30-sec video which I was not happy with, but here it is:
https://www.youtube.com/edit?o=U&video_id=YDn9B0fFqRU
FS9= MicroSoft Flight Sim Ver. 9 with PSS777 add-on model, which is the software that ZS used in his simulator cases of interest.
>>I agree with you that the next search probably should focus near Arc7 +/- 25-nm from say 18 to 25 South.
>>However, in doing a recent simulation, I inadvertently ran out of fuel and created a double crash landing that seemed to possibly explain the debris. Speculative only.
@TBill: I think what you are proposing is a steep descent at 00:19, a manual recovery, a controlled glide, followed by a botched ditching. If there were no pilot inputs, I don’t see how the final BFOs could have led to anything but a high energy impact.
@Victor
Yes this example would be for active pilot pulling out of a descent, which active pilot was always my inclination, but I have to hope it is near Arc7 as the first priority, because I do not know where the flight went nor why, except to say SIO.
Because of active pilot hypothesis, I am in agreement with 18-25 South search area. My other hypothesis is deep trench, and we have some in the 22 South area.
@Mick wrote “The SIR makes the function of Sabre Flight Explorer clear on p.326;”
I’ll disagree with that comment and the paragraph to which it refers.
Flight Explorer did, and does, provide real-time tracking. The issue is the interval of the real-time updates, the updates are applied as they are received. In certain parts of the world Flight Explorer uses high frequency ANSP data feeds derived from secondary surveillance. In the case of MAS B777-2H6ER aircraft in SE Asia, it relied on the position reports from the aircraft itself. The update was immediately displayed by Flight Explorer, the update interval was simply too long to permit a timely intervention.
One must ask why was MAS ODC not aware of the overdue position report that -MRO was expected to send at 17:37.
@Andrew, DCA does set out requirements for the supervision of flights, and that suitably qualified persons are employed to perform that supervision.
As you note, the first phase of ICAO’s GADSS initiative is due to come into operation in November requiring aircraft location to be reported at 15 minute intervals, or less.
@Don Thompson
Don, we clearly have different understandings of the term ‘real time‘. I do not consider a system that updates once every 30 minutes to offer ‘real time‘ tracking.
@Richard Godfrey,
You said: “I thought we had established from the BFO satellite data and the Boeing end of flight simulations that the last seconds of MH370 were a descent at greater than 15,000 fpm with a single catastrophic impact.”
You left something out. Don’t include me in your “we” unless you add the caveat that it was assumed that there was no active piloting then. I agree that the BFOs indicate a high ROD at 00:19:37, and I agree that the debris suggests a “single catastrophic impact”. However, I don’t agree that the crash must have occurred within seconds of the high ROD. There could have been a significant time delay between the two events if the aircraft were piloted in the interim.
Mick,
I was being a little provocative, and taking issue with the text of the SIR, but even a long interval update can be deemed real-time if it is immediately processed and presented. Continously updated, and presented in real-time, would be the preferred solution.
@airlandseaman:
Previously you provided these MH371 calculations:
“Time period Delta T Altitude Decent rate #samples STDEV
07:04:12 to 07:04:17 5 sec ~36100 ft ~2000 ft/min 16 1.7 Hz
07:17:59 to 07:18:05 6 sec ~10000 ft ~2000 ft/min 30 1.7 Hz”
Your data are not always representative of ~5-6 second samples of MH371 BFOs. For instance, from 07:03:58 – 07:04:04, which is the BFO string near 36,000 feet immediately preceding your first data set by only 14 seconds, the BFO standard deviation is 5.6 Hz. Thus, it appears that the ROD in a VNAV descent in the stratosphere is sometimes fairly constant over 5 seconds, and sometimes it varies significantly even during 5 seconds. This reinforces the conclusion that the 18:40 BFOs (over 60 seconds) are inconsistent with the VNAV descent made in MH371.
@Don Thompson
RE: “DCA does set out requirements for the supervision of flights, and that suitably qualified persons are employed to perform that supervision.”
Yes, of course; however, the regulatory requirement for ‘flight following’ does not necessarily imply a requirement for real-time flight tracking (yet!). Nevertheless, I would expect the ODC staff to have been trained how to use the systems they were given to do their job.
@Edward Baker
Perhaps placing the phone at the RIGHT place in the plane (i.e. the tip-top of the cockpit [not pot-pit]) is the reason why ONLY co-pilot’s phone was detected/connected; assuming al least all crew member’s phone numbers have been checked.
But I wonder whether restarting the phone at the RIGHT moment is another factor, say when the plane was approaching Penang near enough, which someone in the cockpit would be better able to judge …
[Not necessarily technically/technologically correct — but restarting the phone is what I usually do occasionally to re-establish connectivity (successfully), say after coming out of a long tunnel.]
@Don Thompson
Further to my previous comment: “…I would expect the ODC staff to have been trained how to use the systems they were given to do their job.”:
Clearly they weren’t!
@mash
Unfortunately, I have never heard any official statement relative to the extent of the examination of the Penang base station “tower dump”.
Bobby:I’m well aware that not all BFO bursts had a low stdev. In fact, I included the example you cited above in a private email to several other IG people 2 days ago. But that was my point. The MH371 data is not always consistant, thus not depositive. It does not prove anything wrt what was happening at 18:40. In particular, it does not offer any solid indication of the mode of AP descent, if one was occurring. It is certainly a very long stretch to conclude that V/S mode must have been used to make passengers and crew not notice the diversion underway.
@airlandseaman
“It (MH371 data) does not prove anything wrt what was happening at 18:40.”
I disagree.
We do not know when the descent started, you are relying on assumption.
I selected reliable ACARS data, where I have a start time and end time, a start altitude and an end altitude. I am not making any assumptions in calculating the ROC = -2,055 fpm. The timeframe I selected from 07:17:54-07:19:08 was at ~2000 fpm according to the ACARS data and the Altitude at the start of the timeframe was ~11,000 feet and at the end of the timeframe was ~8,600 feet. I do not agree that the 164 BFO samples in this 73 second timeframe and a SD = 4.33 is insuficent evidence to concude anything about what happened at 1840. In my view the MH371 data shows that MH370 was not in a LNAV turn or a VNAV descent at 18:40 UTC.
The proposal from Victor of a V/S descent at 18:40 UTC at -2,600 fpm is not perfectly matched by any of the MH371 data, but the closest is the data between 07:14:04 and 07:19:04 UTC.
@DrB
“However, I don’t agree that the crash must have occurred within seconds of the high ROD. There could have been a significant time delay between the two events if the aircraft were piloted in the interim.”
Re: The case for a longer glide/Lack of IFE logon
Has anyone yet suggested the pilot could have switched off “L BUS TIE” to ISLN to cut the SDU after the 00:19 BFO’s? Does that work? Or else pilot turned off APU (not sure pilot can fly plane very well without it).
However, I have to say the alternate config quick dive/no glide tends to look like a good theory at this point.
If the existing data from the MH371 logs are insufficient to come to any definitive conclusions, perhaps more data would help ?
For example:
(a) Did 9M-MRO have a QAR ?
(b) If so, was it downloaded during the maintenance period between MH371 and MH370 ? (perhaps even the FDR may have also been downloaded ?)
(c) Has the flight crew of MH371 been interviewed, and if not, would doing so, even now, be useful ? I am pretty sure they would remember quite a lot about their last flight in that aircraft.
Richard: We will have to agree to disagree on this matter. I stand by my comments and calculations. I disagree with your conclusion, and your statement that we do not know when the descent started. It is pretty obvious when the 2000 ft/min descent started (circa 07:02). At 07:04:08, the ROD was not 753 ft/min, as you calculated. That value was the average rate over the period from 06:59 to 07:04, which included about 3 minutes at 40,000 feet, and 2 minutes descending at 2000 ft/min. By 07:04:08, the rate was well established at 2000 ft/min.
The data you chose to use (lower in the descent) matches your conclusion. However, if one considers all the data in that MH371 descent (40,000 ft to 8000 ft), the only conclusion one can reach is that we simply cannot draw any dispositive conclusions applicable to MH370 at 18:40. The STDEV varies. It was high at the start (probably due to the initial deceleration), then lower, then higher again at the bottom of the descent (where boundary level turbulence is more prevalent).
I will emphasize that I don’t disagree with any of the STDEV math you, Victor and Bobby have reported. I get the same results. Our disagreement is over the choice of data to consider. I prefer to look at ALL of the descent data in drawing a conclusion.
@airlandseaman
You stated “I prefer to look at ALL of the descent data in drawing a conclusion.”
… implying that I did not!
What a cheek!
Yet another assumption.
Richard: I intended no such implication. I assumed nothing in that regard. I simply meant that you chose to use only the lower data for your analysis.
@airlandseaman
I have explained multiple times that the “lower” data (lower in altitude, not in quality) were the only secure data, that a closer analysis allows. For you to claim, that the data allows no conclusions is illogical.
You stated originally “I prefer to look at ALL of the descent data in drawing a conclusion.” Implying others do not. It might get come a big surprise to you, but I also looked at ALL the data.
That was a cheap comment, against those who have a different opinion to you.
You state now “I intended no such implication. I assumed nothing in that regard.”
I am sorry, but you should think about the implications of your comments, before you shoot from the hip.
@Victor
Below is a link to a new paper entitled “MH370 Floating Debris Simulator”:
https://www.dropbox.com/s/b6kr4waxvc4ai7s/MH370%20Floating%20Debris%20Simulator.pdf?dl=0
@everyone:
I’ve been absent due to health issues. Apologies to those whose questions I wasn’t able to answer anymore before disappearing. I will try to search them and get back to them with answers when I’m getting better.
I’ve followed this blog and its predecessors since day 1, but unfortunately haven’t been able to closely follow any new developments in the past few weeks. Subject to Victor’s consent, I’ll hope to be here from time to time depending on my condition. I hope you’ll bear with me, if I missed something. I’ll try to stay up to date as best as I can.
PS: I’m glad you are all still here.
> mash:
> Assuming:
> A. Mobile phone was switched off before diversion.
> B. Mobile phone was re-switched on (by co-pilot) after diversion.
Couldn’t it have been on all the time ?
> 1. Why did pilot choose “lock-out” option instead of “knock-down” option?
100 things could go wrong with the latter, no ?
> no really justifiable reason to ask co-pilot to leave cockpit
The TV documentaries I’ve seen suggest the pilot could have asked the co-pilot to fetch him some coffee – although this scenario doesn’t strike me as very realistic, given that’s a task more suitable for the cabin crew than the co-pilot.
Off the top of my hat, I’d come up with a different strategy:
Pilot to co-pilot: [i]”I think, I’ve heard a strange noise coming from the starboard engine.”[/i] After checking the instrument readings: [i]”Can you have a look if you see something unusual.”[/i]
I’m sure a perpetrator with time for elaborate planning can find some pretext at least as good.
> Niu Yunu:
> DEAR MH370? Isn’t that a bizarre ACARS message?
I wondered about that, too.
DEAR MH370? Isn’t that a bizarre ACARS message?
– Culturally very common way of writing in English by folks from the Far East. It is a more personal way of writing and may have been adopted for the messages just by habit. There is nothing bizarre about that.
@everyone (Victor, Andrew, Don Thompson, Mick Gilbert, sk999, Nederland, … )
Do the ACARS revelations as presented in Victor’s last 2 blog posts shed new light on our search for a smoking gun concerning the “lost VHF” message (if you remember my observations) ?
Since you also seem to know more about MH371 now, has it been established in the meanwhile, whether or not VHF deselection – which Andrew qualified as unnecessary and unusual – is part of MAS preflight checklists for flights to China ?
Andrew had established 4 scenarios, in which VHF is unavailable for ACARS transmission:
(1) data mode was deselected on VHF-C
(2) default radio mode set to VOICE
(3) manual VHF deselection
(4) no usable frequencies in the VHF scan mask
Andrew later explained why we should ignore n°4. So we are left with the first 3 explanations for the “lost VHF” message. Andrew’s perspective on them:
(1) innocent explanation possible: “crew used VHF-C for voice transmission on another frequency (eg company ops, engineering, etc) and subsequently forgot to re-select data mode” ¹
(2) “Could be suspicious – there should be no need for pilots to change the default radio mode to voice.” ²
(3) “If it was a ‘permanent’ SOP, I would expect to see it mentioned in one of the MAS manuals, probably the Route Manual or the OPS-A Manual, however, It is not mentioned in either (or the FCOM for that matter). It may have been some temporary thing that was promulgated by another medium that has not been shared, such as a Flight Operations Notice (or whatever MAS calls them).” ²
Looking at the full ACARS message log, Victor posted above, I notice:
15:54:29 Lost SATCOM
15:54:31 Frequency Auto-tune, “SWITCH VHF3 TO VOICE (131.550) IF SATCOM SERVICABLE”
15:54:41 Established SATCOM
15:54:53 Lost VHF
• Does that allow us to narrow down the reason for the loss of VHF ?
• I don’t remember that we ever talked about lost SATCOM. This is the first time I see it. I recall that 0LV155453S was in sk999’s list, but not 0LS155441V. Why is that ?
• Doesn’t the ACARS log excerpt above fit quite well with this B777 pilot’s statement ?
“It is MAS procedure to switch ACARS VHF and HF selection off but this is only for flights to China as the service provider for MAS does not cover China. Some if not all pilots switch them all off for a while and then later switch SATCOM back on to force the system into SATCOM mode.”
In case none of the above allows any conclusions to be drawn, how can we make progress on this issue ?
Andrew suggested this:
“Do we know for sure that it was done on MH371? The list of media advisory messages that @sk999 posted only shows one instance where the VHF link was ‘lost’. That occurred at 15:55 on 7 March, when the ill-fated MH370 was on the ground in KUL. We do know that only the SATCOM datalink was available for MH371’s flight from PEK to KUL on 7 March, but that could be the result of the VHF scan mask not allowing the system to use the VHF frequencies in China. I suspect you’d have to go back and look at the messages for MH370 on 6 March (UTC) to see if the VHF link was ‘lost’ on that earlier flight. If such a message exists, it might indicate that deselecting VHF was a regular occurrence on flights to PEK.”
Don Thompson suggested “looking into hobbyist ACARS tracker logs with examples of MAS traffic over VHF”. Could someone familiar with these sites have a look ?
If we can determine the “lost VHF” message on MH370 to be something abnormal, we would be a giant step closer to having solved the MH370 mystery definitively: I would consider the combination of [abnormal VHF deselection on the ground] + [(abnormal) SATCOM deselection in flight] definite proof of pijacking, because if both events are highly abnormal and similar to each other, they must be connected, and there was no emergency on the ground. The first event (if proven to be abnormal) foreshadows the second event. It’s somehow analogous to how the flight simulation into the SIO foreshadowed the real event.
Therefore I would be so glad if we could settle this issue one way or the other. I would welcome any information on that front, whether confirming or denying the hypothesis. What do we need to get at the bottom of this?
–––
¹ “I’m not sure that VHF ENABLE had to be deselected to generate the media advisory message. As I understand it, the message only indicates there was a change in the mode (ie VHF was ‘lost’) and that only SATCOM was available (ie VHF unavailable). I believe that could also occur if DATA mode was deselected on VHF C while the crew used it for voice transmission on another frequency (eg company ops, engineering, etc). If the crew subsequently forgot to re-select DATA mode, then VHF ACARS would remain unavailable. Perhaps Don can confirm if that’s correct or not.”
http://mh370.radiantphysics.com/2018/03/18/mh370-search-update-mar-18-2018/#comment-14213
I don’t think, we have heard Don on this yet.
² http://mh370.radiantphysics.com/2018/03/18/mh370-search-update-mar-18-2018/#comment-14247
@Victor
David Griffin kindly provided some feedback to my latest paper in a private email.
The paper was presented in the following comment:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18782
Here are the points David raised together with my answers:
1) DG: All undrogued GDP drifters start their lives in the drogued state, then become undrogued about halfway through the deployment.
RG: For the undrogued drifters used in this test, the timeframe of interest was from the area near the 7th Arc to near the east coast of Africa. Out of the 7 drifters used, 5 had lost their drogue before the start at the 7th Arc, the remaining 2 lost their drogue whilst still in the area of the 7th Arc or shortly afterwards and were at least 80% undrogued for the transoceanic journey.
https://www.dropbox.com/s/iswokhhwg6h1r70/Calibration%20Summary.png?dl=0
2) DG: You describe the velocity in terms of speed and direction. Can I assume that the simulator does all its calculations using averaged Cartesian components?
RG: Yes, that is correct.
3) DG: For only one of the 7 example buoys is the simulated track of longer duration than the real track. This suggests that the average model velocity vector does not match the vector-average of the observations.
RG: The model is based on the average historic data from the 280 undrogued drifters used. The model does not take into account when a particular drifter encounters unusual weather along its track. For example, the drifter 84148 crossed the path of Tropical Cyclone Anja with winds up to 165 km/h between 14th-17th November 2009. This results in drifter 84148 being faster than the historic average.
As I mentioned before, I have updated my path generation tool to WGS84 coordinates. Currently I’m looking into CTT paths as a function of FFB (assuming this value possibly had changed after the 18:25 log-on).
Working on the new path calculations I ran into some basic questions I’m quite certain others here have already found the answer to:
I’m trying to understand MRC and LRC speed schedule in detail. In the following document it is suggested that MRC M number depends on weight:
https://www.icao.int/Meetings/EnvironmentalWorkshops/Documents/ICAO-TransportCanada-2006/Anderson_ops.pdf
In the next document it is suggested that MRC M number is constant:
http://www.boeing.com/commercial/aeromagazine/articles/qtr_02_10/pdfs/AERO_FuelConsSeries.pdf
Is there precise information available for the B777?
How can I find the “optimal altitude” for MRC and LRC (Pressure altitude for a given weight and speed schedule that produces the maximum air miles per unit of fuel);
Would the use of “optimal altitude” imply climbing when fuel is burned?
@Niels: Some short responses to your questions:
1) The LRC speed is a function of weight and altitude. The “optimum” altitude is the one that maximizes mileage at LRC speed.
2) The MRC speed is ECON with CI=0. In general, the ECON speed is a function of weight, altitude, and tailwind/headwind. For the case of CI=0, the ECON speed is a function of weight and altitude. Again, the “optimum” altitude is the one that maximizes mileage.
3) The LRC speed and fuel flow tables are readily available in the FCOM for a B777-200ER with RR engines. However, a pilot would typically choose ECON over LRC for a number of reasons, including the value of CI can be chosen based on a particular airline’s fuel costs vs. time costs, and the choice of speed includes the effect of wind.
4) The tables of ECON speed versus CI, weight, altitude, and wind are to my knowledge not available for 9M-MRO. @DrB and @sk999 have done some work on this for use in their models.
5) The optimum altitude increases as fuel as burned. To remain at the optimum altitude would require step climbs during the flight.
@VictorI
Thank you, Victor, that’s helpful. Of course it is difficult to know in the case of MH370 if a “typical” choice was made, so perhaps I’ll start with checking the LRC speed profile if there is some information (tables) available in the public domain. Possibly also the constant M in combination with step-climbs is worth checking.
@Niels: I’ve mostly looked at LRC and constant Mach number at constant altitude after 19:41, but certainly ECON at various values of CI with and without step climbs could be considered.
@DennisW
(copied from SIR page 20)
“1.1.5 Detection of Hand Phone Signal
A Telco service provider in an interview with the RMP confirmed a signal “hit” occurred at 0152:27 MYT on 08 March 2014, coming from the mobile phone tower (LBS Location Base station) at Bandar Baru Farlim Penang. The signal “hit” however did not record any communication except to confirm that it was in the ON mode signal related to the “hit”. The phone number xxxxxxx was later traced to that registered under the FO. This was supported by the RMP’s report. ” …
Given that the plane was flying so fast, say more than 10K per minute, I think the distance and the time the plane was inside the coverage area might be less than 1K and 0.1 minute respectively. Therefore, it is very easy to check the hit list say +/-1 minute of 0152.27 MYT against the available crew/passenger phone list.
And it is possible that a call (if attempted) be unsuccessful because of the limited time available.
@mash: The problem I have with the statement A Telco service provider in an interview with the RMP confirmed a signal “hit” occurred at 0152:27 MYT on 08 March 2014 is how incomplete that inquiry appears to be. You would expect that the Malaysia investigators would supply the cell phone numbers of all passengers and crew and ask all the telco providers to do a thorough search and provide the records in an electronic format. That information would not lend itself to be transmitted in an interview.
@mash
The FO phone registration near Penang is an annoying incident in many ways. Hishamuddin Hussein initially claimed to have no knowledge of the incident, and called it improbable. This suggests that the service provider was not specifically tasked to perform any base station forensics. If the service provider was not tasked by Malaysian authorities, how did it know the FO’s phone number, and what motivated even checking for it?
No reference has ever been made relative to the extent of the base station investigation i.e. were any other PAX or crew member phones even considered? If the FO’s phone registered, it seems likely to me that other registrations from the plane occured.
Basically very sloppy reporting at least, and very sloppy forensics at worst.
@Peter Norton asked “If we can determine the “lost VHF” message on MH370 to be something abnormal”
My conclusion is that disabling VHF data was not something abnormal. The instruction was issued by MAS ODC, and during the previous MH371 service from Beijing VHF was disabled (e.g. ACARS Media Advisory at 0107:40Z, below).
S61A MH0371 0 ES010740S
While it’s apparent that MAS contracted with SITA as their provider for aeronautical messaging service, and other data processing services, MAS chose to avoid using VHF data comms in areas not served primarily by SITA’s VHF datalink service.
MAS ‘home turf’ is predominantly served by AeroThai’s VHF datalink radio network that is part of the ARINC GLOBALink datalink service. Message transit over this datalink service provider incurs additional cost that, in aggregate, renders it more expensive than SATCOM.
Here’s an excerpt from a discussion with Hawaiian Airlines’ principal avionics engineer:
“In fact, satcom ACARS is cheaper than VHF ACARS in the far East. In that region, destinations we fly to such as New Zealand, Australia, China and Japan, are charging 45 cents per kilobit for VHF ACARS. We’re talking $1,000 per megabyte. What we’ve done is turned off VHF ACARS in the Far East. On our A330s, for example, in Japan and China we send everything over satcom,”
Japan’s AVICOM, China’s ADCC, Australia AirServices, and NZ all operate under the ARINC GLOBALink umbrella. Don’t get distracted with other facets of the Hawaiian Airlines story, that report deals with a trial service.
This expanded Traffic Log shows that logic used in processing MAS ACARS messages reverted to force messages via VHF. That’s is interesting, but it doesn’t indicate a malevolent intent.
Concerning “DEAR MH370. PLS ACK TEST MSG. RGDS/OC“. I do regard it as odd, “PLS ACKNOWLEDGE MSG AND REPORT POSITION IMMEDIATELY” would have been more appropriate to the situation, considering three FMC derived position updates were then overdue at 1838:51Z.
@Niels,
You seem to be in the same place that I was in January 2017 regarding Boeing fuel flow and air speed models. At that time, I had many of the same questions as you have now. I first tried empirical models using the few available Boeing fuel flow tables, but as you know, these do not include MRC or any other ECON Cost Index. After some investigation, I decided that for a fuel flow and speed model to be useful it needed to have an accuracy of 1% are thereabouts. It did not seem possible to do this without a complex model involving aerodynamic drag curves. Because the fuel flow is a big part of the operating cost, Boeing desires to keep that data proprietary (particularly near MRC) and only releases, even to its customers, minimal fuel flow information, primarily non-cruise modes and LRC, which is not normally flown by the airlines since they use a cost index in ECON mode. In order to maintain its competitive advantage Boeing retains that data and does not distribute it. Fortunately, there is one drag plot for a Boeing 777–200 ER in a book. Using that drag plot I was able to generate a digital table of drag versus weight and airspeed for a Boeing 777–200 ER in level, constant-speed flight. Using that table and an aerodynamic model I was able to find the drag under the conditions corresponding to the published Boeing fuel flow tables. That allowed me to drive a thrust-specific fuel flow equation for the engines. With those two pieces of information one can derive fuel mileage and therefore find the speed (MRC) which corresponds to the maximum fuel mileage. Using very many computed fuel mileage curves as a function of speed for a given altitude and weight I was able to derive equations for various values of cost index including cost index equals zero, which corresponds to MRC. This fuel modeling took me about 1,800 hours to complete between January and November 2017.
My fuel flow, air speed, and endurance model can be found here:
https://drive.google.com/file/d/1Wt9DOU0Z53W7NERzSsK2sxcyrojmN7Sq/view
It has the equations for estimating the fuel flow and the air speed for many standard speed modes. Using that model you can find the predicted fuel flow and airspeed for MRC under a various weight, altitude, and temperature conditions. Comparisons with other 9M–MRO flights indicate an agreement in predicted and measured fuel flows under typical level cruise conditions of about 1%. The predicted MRC airspeed is accurate to perhaps 1 to 2%.
@Peter Norton
I agree with @Don – the full traffic log indicates that VHC C was selected to voice in response to a message from MAS operations. I don’t believe there’s anything sinister in that – it’s probably a cost issue, as @Don discussed. I do wonder why the airline didn’t set up their ACARS to automatically use SATCOM instead of VHF in areas where SITA didn’t provide coverage. I guess there’s more to that part of the story.
@Andrew:
You are saying we are dealing with scenario 1 of 4, correct ?
(1) VHF-C was switched from data mode to voice mode
questions:
• Do you mean that was done in order to …
(1a) … talk to “company ops and they subsequently forgot to re-select data mode” as you suggested before, or
(1b) … disable VHF-C data mode due to abovementioned cost issues (by Don)
?
• If you say it’s (1a) and Don says it’s (1b), how do we find common ground? Logically, it cannot be both at the same time.
• If (1b): If it was MAS SOP to disable VHF to/from China, then why is the manual “SWITCH VHF3 TO VOICE” message required ?
• “why the airline didn’t set up their ACARS to automatically use SATCOM instead of VHF in areas where SITA didn’t provide coverage. I guess there’s more to that part of the story.” What do you have in mind in that regard ?
• How does that all explain lost SATCOM a few seconds prior ?
@Peter Norton
RE: “You are saying we are dealing with scenario 1 of 4, correct ?
(1) VHF-C was switched from data mode to voice mode
Yes, the crew was instructed to do so by MAS ops, presumably due to cost issues.
RE: “If (1b): If it was MAS SOP to disable VHF to/from China, then why is the manual “SWITCH VHF3 TO VOICE” message required ?”
I don’t know. Perhaps it was something new that hadn’t yet made it into the formal SOPs.
RE: “why the airline didn’t set up their ACARS to automatically use SATCOM instead of VHF in areas where SITA didn’t provide coverage. I guess there’s more to that part of the story.” What do you have in mind in that regard ?
Again, I don’t know. If avoiding excessive VHF ACARS charges was something new, then perhaps engineering hadn’t yet modified the ACARS software.
RE: “How does that all explain lost SATCOM a few seconds prior ?”
It’s not uncommon to intermittently ‘lose’ SATCOM for a few seconds on the ground and in-flight. In this case, the SATCOM link was re-established 12 seconds later.
@Niels
I have a difficult time getting concerned with fuel consumption details. The route between IGARI and 19:41 is far from characterized in terms of how the aircraft was flown, and precisely where the FMT occurred. These unknowns have a far greater impact on the latitude of fuel exhaustion than a fine grained consideration of the physics of fuel flow rates.
@Victor or @Andrew
Question Re: SDU/IFE reboot at 18:25 and 00:19
What EIACS (screen) messages do we think the system shows in the cockpit when SDU is powered back on? IFE reboot? I know you said when LEFT BUS is off we get a SDU disconnect message.
@Andrew: thanks for your reply
If I interpret it correctly, you believe that
– “VHF-C was selected to voice in response to a message from MAS operations”
– this message being “SWITCH VHF3 TO VOICE”
– and that the crew did not “subsequently forget to re-select data mode”
– so this was not a mistake, but they rather made that switch on purpose as “instructed by MAS ops, presumably due to cost issues”.
Is that correct ?
Your points are well taken. 2 things bother me, however:
(A) If VHF was to be disabled (for ACARS) on every flight in that region, I find the message peculiar.
If VHF was not disabled on previous MH370 flights, that would obviously merit investigation.
(B) “It’s not uncommon to intermittently ‘lose’ SATCOM for a few seconds on the ground and in-flight.”
According to the log
– it happened only once on MH370
– and not once on MH371
And what is more, SATCOM was lost only 24 seconds before VHF loss and a mere 2(!) seconds before the VHF-instructions were received.
15:54:29 Lost SATCOM
15:54:31 Frequency Auto-tune, “SWITCH VHF3 TO VOICE (131.550) IF SATCOM SERVICABLE”
15:54:41 Established SATCOM
15:54:53 Lost VHF
Are you suggesting this is pure coincidence,
i.e. the events are unrelated and not connected to each other ?
@Peter Norton: The wording of the message at 15:54:31 is also a bit odd. Why would MAS operations advise to switch VF3 to voice on an ACARS frequency (131.550 MHz)? I think company operations are usually at the upper end of the VHF frequency allocation (136.000 – 136.975 MHz). Now, it’s possible that the 131.550 MHz was referring to the present (ACARS) frequency, but if so, the message is very poorly worded. I suppose that changing the frequency from DATA to 131.550 would also allow somebody in the cockpit to audibly monitor ACARS traffic between the ground station and other aircraft, but I don’t see how that would be useful in most circumstances.
@DrB
Many thanks for sharing your FF/AS/Endurance model, as well as for giving some perspective on the difficulties involved in the proper modeling with the limited data available. The problem is much more complicated than it looks at first sight, even if one would “only” want to know the different speed profiles. For weight vs. time calculations a major problem seems to be in our understanding of the crucial period between 18:28 and 19:41. However, what I initially will try (to derive speed profiles for after 19:41) is to calculate “backwards” assuming fuel exhaustion happened some minutes before 00:19.
It will take me some time to study your models; I hope I can sometimes bother you with (probably rather basic) questions.
@DennisW
I just see your posting: yes we agree on some of the issues, however as I just mentioned to Bobby, with some assumptions I think the models can be useful and applied for the period after 19:41.
@Don Thompson:
I’ve read your article. Thank you for the interesting commercial background.
You suggested “looking into hobbyist ACARS tracker logs with examples of MAS traffic over VHF”. My search came up emtpy. Do you know any such sites?
And are there any chances to retrieve ACARS logs for previous MH370 flights, given how long it took to even get the full log for the fateful flight ?
@Andrew:
“VHF-C was switched from data mode to voice mode
as instructed by MAS ops, presumably due to cost issues.”
Is that even a working solution? I’ve read that after x minutes without transmission from the pilots, control of VHF-C is given back to ACARS and the frequency switched back to the ACARS frequency. I am not sure, if that applies to a B777.
@Victor Iannello:
I agree, this is seems peculiar, particularly when adding the fact that of all messages, these 2 messages (established SATCOM / lost VHF) are precisely the ones which were erroneously repeated, as you pointed out:
“The facts surrounding the ACARS traffic log in the SIR (also Appendix 1.9A) are more suspicious. For Page 1 of the log, the filter parameters did not limit the messages to only SATCOM messages. In fact, at 15:54:31, there is a message submitted by MAS ODC over the VHF link that requests personnel on the aircraft to re-configure the center VHF radio so that future messages would be exchanged via SATCOM. However, starting with Page 2 […] VHF messages, if any occurred, were excluded. The change in filter parameters after Page 1 is unexplained. Two messages received by MAS ODC at 15:41:41 and 15:54:53 appearing at the bottom of Page 1 are repeated at the top of Page 2. The change in filter parameters and the repeated messages are clear evidence that the traffic log in the SIR is actually two reports that were pieced together and presented as a single report.”
(BTW, this is a typo. 15:41:41 should read 15:54:41)
> @Victor Iannello:
> The wording of the message at 15:54:31 is also a bit odd. Why would MAS
> operations advise to switch VF3 to voice on an ACARS frequency (131.550 MHz)?
> I think company operations are usually at the upper end of the VHF frequency
> allocation (136.000 – 136.975 MHz). Now, it’s possible that the 131.550 MHz was
> referring to the present (ACARS) frequency, but if so, the message is very poorly
> worded. I suppose that changing the frequency from DATA to 131.550 would also
> allow somebody in the cockpit to audibly monitor ACARS traffic between the ground
> station and other aircraft, but I don’t see how that would be useful in most
> circumstances.
I think it depends on whether VHF-C was switched to voice mode in order to “[talk to] company ops on another frequency and the crew subsequently forgot to re-select data mode” by mistake (as Andrew suggested in April) or to prevent ACARS VHF transmission for financial reasons as suggested above. It can only be one or the other (since it cannot have happened both by mistake and on purpose at the same time). Are we sure it is the latter?
@Peter Norton: We now know that VHF was configured for DATA mode at 15:54:29, prior to switching to voice, so it would seem that the message at 15:54:31 was intended to inhibit ACARS over VHF, presumable for cost reasons.
Thank you for finding that typo. I’ll fix it.
@Peter Norton
RE: “If I interpret it correctly, you believe that
– “VHF-C was selected to voice in response to a message from MAS operations”
– this message being “SWITCH VHF3 TO VOICE”
– and that the crew did not “subsequently forget to re-select data mode”
– so this was not a mistake, but they rather made that switch on purpose as “instructed by MAS ops, presumably due to cost issues”.
Is that correct ?”
Yes.
RE: “(A) If VHF was to be disabled (for ACARS) on every flight in that region, I find the message peculiar.
If VHF was not disabled on previous MH370 flights, that would obviously merit investigation.”
Yes, it would be interesting to find out if VHF ACARS was disabled on other flights. I understand that VHF ACARS was not available on the MH371 flight from PEK-KUL immediately before MH370, but I don’t know how that came about.
RE: “Are you suggesting this is pure coincidence,
i.e. the events are unrelated and not connected to each other ?”
Yes. If you have a look at the downlink messages coming from the aircraft, you’ll see that none of the messages before 15:56:08 show the Flight ID ‘MH0370’. That suggests the crew were still doing their pre-flight preparation and hadn’t entered the Flight ID in the FMC until some time between 15:54:53 and 15:56:08. The ‘lost SATCOM’ message might be related to something the crew did during the pre-flight (eg ADIRU cycled, Master Comms reset). Perhaps @Don has some ideas.
RE:“Is that even a working solution? I’ve read that after x minutes without transmission from the pilots, control of VHF-C is given back to ACARS and the frequency switched back to the ACARS frequency. I am not sure, if that applies to a B777.”
That is not correct for the B777.
RE: The frequency 131.550.
I don’t see the point in directing the crew to select an ACARS frequency in voice mode. The MAS route manual shows that the company frequency at KUL was 130.15.
@DennisW
@Victor Iannello
Just a guess (could be a completely wrong interpretation).
Perhaps it was a search report/result “interview” (between RMP and the service provider), say as follows:
1. MAS gave a list of numbers to RMP.
2. RMP asked the service providers to check the numbers (without names).
3. One service provider reported a hit. [“hit” different meaning here]
4. RMP/MAS “traced” that it was FO’s number.
@mash
The whole cell phone registration report is cloudy IMO. Who authorized it? What was the extent of the forensics? Blah, blah, blah,… It really is a perfect example of Malaysian obscuration of the diversion.
@Peter Norton
Concerning data/voice mode for the VHF radio.
When used for data, the VHF radio is under the control of the Data Comms System, hosted on AIMS. The radio is configured via the ACARS and VHF Manager pages on the MFDs.
I perhaps assume too much, but I’d expect that MAS 777 flight crew would understand that ACARS frequencies, selection of available datalinks, and defining whether the selected VHF radio defaults to voice or data mode, are all managed via the ACARS Manager and VHF Manager pages.
FWIW, I interpret the sequence of ACARS Media Advisories as:
1) 0EV125037VS – aircraft start up, both datalinks available, VHF is the established ACARS medium.
2) 0LS155441V – SATCOM has been unchecked in ACARS Manager, VHF remains (momentarily) the active medium, VHF is then unchecked in ACARS Manager.
3) 0ES155446VS – SATCOM checked in ACARS Manager, SATCOM is the established medium. VHF radio remains available for data use but VHF remains unchecked in ACARS Manager.
4) 0LV155453S – SATCOM is established ACARS medium. the VHF radio is no longer available to ACARS. VHF Manager screen used to change radio default mode from DATA to VOICE.
4a) The crew is working pre-flight procedures & the flight ID is entered on the FMS. The AES renews its Log On via POR so as to include the Flight Identifier in the GES Log On. Log On remains Class 1, LGA only.
5) 0ES155607S – SATCOM is established ACARS medium.
5a) ADIRU aligns and AES renews its Log On as Class 3 via HGA.
6) 0ES155759S – SATCOM is established ACARS medium.
6a) AES executes a region handover between POR and IOR.
7) 0ES160015S – SATCOM is established ACARS medium.
and then…
8) ES 161101S – SATCOM established as ACARS medium.
The final Media Advisory, timed on the aircraft at 161101, doesn’t correspond to any SATCOM datalink event. Possibly, it denotes a brief uncheck/re-check of the SATCOM flag on the ACARS Manager screen.
The Stratos GES Log, including the earlier ZBAA-WMKK service, shows that the Media Advisories sent by -MRO reflected only the SATCOM datalink active and available during that service.
@Don Thompson said: Possibly, it denotes a brief uncheck/re-check of the SATCOM flag on the ACARS Manager screen.
That’s an interesting possibility. It might have been a check to ensure that with the SATCOM datalink deselected, there was still SATCOM connectivity, as displayed by messages on the EICAS.
@Victor
Ocean Infinity have completed the first week of their search for the submarine San Juan off the coast of Argentina.
Seabed Constructor first docked at Comodoro Rivadavia in Argentina before travelling to the search area.
https://www.dropbox.com/s/80ke1e0978qc4b9/San%20Juan%20Argentina%20Track.pdf?dl=0
@Richard Godfrey: Have there been any ROV excursions?
@Victor
“That’s an interesting possibility. It might have been a check to ensure that with the SATCOM datalink deselected, there was still SATCOM connectivity, as displayed by messages on the EICAS.”
Please be more specific on the EICAS messages you expect. I think you previously said there is an “SDU off” (paraphrasing) message when the Left Bus is off.
In the Alt Config at end of flight, does the pilot see “SDU off” message (upon loss of Right engine)? What does pilot see upon APU restart? If he sees SDU come back on-line, maybe that’s a mistake and he cuts it back off with L Bus Tie ISLN.
@TBill: When the SATCOM is unpowered, the following EICAS messages are generated:
SATCOM
SATCOM DATALINK
SATVOICE LOST
I believe that if the SATCOM is powered and the SATCOM datalink is deselected in the ACARS Manager screen, only the SATCOM DATALINK message is displayed.
@Victor
There have been no ROV deployments.
OI are operating in water between 700m and 1200m deep.
A ROV deployment would take between 47 mins and 80 mins just to go down and back up at 0.5 m/sec.
There have been 29 stops averaging 6 mins each, where the longest stop was 25 mins.
@Victor
…OK that is interesting. One possible rationale for 18:25 reboot is that we (the outside observers) cannot gleen much new position information, assuming the pilot realizes the approx. radar coverage. If ACARS unexpectedly dumped a dataset at reboot, it would not tell us too much, although we’d sure like to know more exactly the 18:25 position.
If we could get data for another flight, we’d like to see ZS’s 22-Feb_2014 trip to Beijing (9M-MRO). That could have given ZS some feel for fuel consumption differences of the engines and if MH370 was a planned event, we are expecting a hypothetical perpetrator may experiment with certain settings changes as a pre-test. Of course, we like to see all Beijing MH370 flight fuel loads.
>>from FI, are they saying Left IDG must be held permanently OFF once it is turned off (per Alt Config)?
“The IDG drive can be disconnected from the engine by pushing the respective DRIVE DISCONNECT switch. The IDG cannot be reconnected by the flight crew. High drive temperature causes the IDG to disconnect automatically.”
An updated table comparing predicted latitudes at the 7th Arc is available HERE.
Several recent results have been added.
HERE is an updated version of the “Observational Data” for MH370. I added two items at the end for the reasons why no IFE messages occurred circa 00:21.
@DrB: I believe I fixed your links.
@TBill asked: from FI, are they saying Left IDG must be held permanently OFF once it is turned off (per Alt Config)?
No. There are two separate switches. To electrically isolate the left IDG, the L GEN CTRL switch is set to OFF, and can be turned back ON during the flight. On the other hand, the L DRIVE DISC once pressed cannot be reset during the flight, and requires maintenance on the ground.
@Victor Iannello,
Thanks, Victor. You are a gem! The work you do maintaining this blog is much appreciated.
@Victor Iannello,
FYI, both of the links now go to the same document (Latitudes). The Observational Data is here: https://drive.google.com/file/d/1Ggxjn_QH6E0Whznpt4MkXEXKR5yDdL7b/view?usp=sharing
@Don Thompson, @Victor Iannello, @Andrew:
15:54:29 Lost SATCOM
15:54:31 Frequency Auto-tune, “SWITCH VHF3 TO VOICE (131.550) IF SATCOM SERVICABLE”
15:54:41 Established SATCOM
15:54:53 Lost VHF
I have difficulty relating to the VHF-instructions coming in only 2 seconds after SATCOM loss just by coincidence. Is it possible that the message was auto-generated as a response to the SATCOM loss?
Otherwise, if the message was sent manually, but still in response to the lost SATCOM, is it possible that the timing stated in the traffic log is not precise (e.g. different events logged by different, asynchronous clocks), given that 2 seconds would not suffice for a manual response message ?
Speaking of which, why isn’t 0LS155441V logged at 15:54:41 but 15:54:29? Is this the transmission delay or the aforementioned imprecise logging or something else ?
@Andrew, @Victor Iannello, @Don Thompson:
Do I read you correctly, that you all agree the 15:54:53 Lost VHF was a reaction to the 15:54:31 “SWITCH VHF3 TO VOICE” message ? (as opposed to my theory that LOST VHF might have been an early first step in the multi-step process of shutting down coms later in the flight, perhaps to make absolutely sure ACARS won’t fall back to VHF once SATCOM is disabled)
@Don Thompson:
Thanks for your step-by-step interpretation of the media advisory sequence.
1) RE: 0LS155441V – SATCOM has been unchecked in ACARS Manager
WHY? Do you see a reason/necessity for that?
Maybe this? –> “It is MAS procedure to switch ACARS VHF and HF selection off […] for a while and then later switch SATCOM back on to force the system into SATCOM mode.” (see above)
But wouldn’t the same result be achieved without unchecking SATCOM in the ACARS manager? This seems unnecessary.
Andrew suggested “the ‘lost SATCOM’ message might be related to something the crew did during the pre-flight (eg ADIRU cycled, Master Comms reset). Perhaps @Don has some ideas.” Any ideas?
2) Sorry for returning to my comment above, I’m not sure if you saw it:
> You suggested “looking into hobbyist ACARS tracker logs with examples of MAS traffic over VHF”.
> My search came up empty. Do you know any such sites?
> And are there any chances to retrieve ACARS logs for previous MH370 flights,
> given how long it took to even get the full log for the fateful flight ?
3) Victor and Andrew agree that it is odd to select an ACARS frequency in voice mode. Can you make sense of it?
Andrew: “I don’t see the point in directing the crew to select an ACARS frequency in voice mode. The MAS route manual shows that the company frequency at KUL was 130.15.”
Victor: “The wording of the message at 15:54:31 is also a bit odd. Why would MAS operations advise to switch VF3 to voice on an ACARS frequency (131.550 MHz)? I think company operations are usually at the upper end of the VHF frequency allocation (136.000 – 136.975 MHz). Now, it’s possible that the 131.550 MHz was referring to the present (ACARS) frequency, but if so, the message is very poorly worded. I suppose that changing the frequency from DATA to 131.550 would also allow somebody in the cockpit to audibly monitor ACARS traffic between the ground station and other aircraft, but I don’t see how that would be useful in most circumstances.”
I must be missing something, but if the intention of switching VHF3 from data to voice mode was merely to prevent ACARS from using VHF, I don’t see why the frequency is of importance. Any frequency would do in this case, no ?
@Dr B. A couple of minor points on your useful table of predicted latitudes:
1. item 13. You say, “experimentally determined by DGA” where in fact this was by CFD though using data from flotation experimentation.
2. items 17 & 21. “Debris used”. The CSIRO used replicas of ‘Roy’ and a flap fairing, plus a 777 flaperon modified to the recovered flaperon’s shape supplemented by earlier work with several flaperon wooden replicas.
In the sense of your column I would suggest 3.
@Peter Norton
RE: “Do I read you correctly, that you all agree the 15:54:53 Lost VHF was a reaction to the 15:54:31 “SWITCH VHF3 TO VOICE” message ?”
I can’t speak for others, but that’s what I believe. FWIW, the MH370 SIR states (p.108):
“On the event flight, as instructed by Ground Operations via text message shown on the MFD (shown as ‘Switch VHF3 to Voice’), the flight crew would have selected voice on the Center VHF resulting in SATCOM being used for the data transmissions…This switching from VHF to SATCOM for the data transmissions is normal practice in MAS for commercial reasons.“
re: Voice Recognition – Repeated Call
SIR (page 38):
“1.5.12 Voice Recognition of the Radio Transmissions between MH370 and Air Traffic Control
The radio transmissions made between MH370 and the air traffic control were studied. The Team used pilot friends, family members, and an expert report of objective analysis of the radio transmissions in the voice recognition of the transmissions made between MH370 and air traffic control. ” …
As early as about a few months after the incident, it was reported that the pilot’s friend and the pilot’s son could recognise the pilot’s voice ‘subjectively’. IMO I don’t think they would have agreed with the finding (or with each other) if there were any doubts. [Just like the first time in 2014, especially under the circumstances at that time.]
pilot’s son:
https://www.telegraph.co.uk/news/worldnews/australiaandthepacific/australia/10921910/MH370-latest-Pilot-spoke-final-words-from-cockpit-says-wife.html
pilot’s friend:
http://youtu.be/SmOtyjsJ4hQ
Peter Norton said: I must be missing something, but if the intention of switching VHF3 from data to voice mode was merely to prevent ACARS from using VHF, I don’t see why the frequency is of importance. Any frequency would do in this case, no ?
Sure, but if you are going to change it from DATA to a frequency, it would make sense to choose a voice frequency that might actually be useful, such as the company frequency.
@mash,
Unless I missed something those friend/family identifications you referenced were for the last “Good night” radio call. As far as I know, no one is disputing that the Captain made it. Edward Baker believes the previous radio call (repeating FL350) was made by the FO. The audio expert did not analyze the last 3 calls individually, but only as a group. Dr. Kunzel found that the group of the last 3 calls was more likely said by the Captain than by the FO, but the result was not highly certain and doesn’t exclude the possibility that the second of the 3 calls was made by FO. I think Edward is attempting to get Dr. Kunzel to address that possibility.
@Victor
Regarding the Left Bus turn off at IGARI, do you have an opinion on electrical isolation vs. actual disengaging of the IDG motor? I presume latter saves some fuel and possibly provides more thrust.
@David,
Many thanks for your suggestions related to the Observational Data. You are always helpful in getting things right.
On your first item, I was going by this quote from Nesterov’s paper which said: “This theoretical value is in a good agreement with the experimental data for the flaperon (3.29 %) estimated by the DGA in the hydrodynamic engineering test facility centre in Toulouse (Daniel, 2016).”
In addition, Griffin/CSIRO said: “Our earlier field testing of replicas of the flaperon was unable to confirm numerical predictions by the Direction Generale de L’Armement (DGA) that the flaperon drifted left of the wind. Field testing of a genuine Boeing 777 flaperon cut down to match photographs of 9M-MRO’s flaperon has now largely confirmed the DGA predictions, at least with respect to drift angle. “
Why do you think the statement “experimentally determined by DGA” is incorrect, and who is “CFD”?
On the second item, I would agree with three (not 1) debris items for entry #21, the 2017 paper by Griffin of CSIRO. Their analyses included three items: the flaperon, generic low-windage items, and generic high-windage items.
Entry #17 is the previous CSIRO paper in 2016. In it, they studied three types of replicas (the flaperon, “Roy” panel, and the track fairing). So I would agree that 3 is correct for entry #17 also.
I’ll put out a revised version when the dust settles.
@Dr B. “CFD” = Computational Fluid Dynamics, being the “numerical predictions” of David Griffin’s comment. He has pointed out that the French research was CFD,“in silico”’, adding that, “There is no mention of waves, which I thought was a critical omission, hence our resolve to do field testing. I think their model has correctly simulated the asymmetry of the hydrodynamic forces acting on the asymmetric below-water profile, but not found the correct equilibrium speed through the water.”
At p151-186 below you will find Annex 9, where the DGA hyrodynamics centre did flotation tests (including settling rate), physical measurement and response to added weight, including tilt. Computer shapes were drawn up and such measures of hydrostatic stability as metacentric height were derived.
Monsieur Pengram, also from the hydrodynamics centre, took these data (as described in Annex 10, pages 187-217), modified the shapes for the computer to suit and computed what would be the various lift, drag and moment coefficients before using these in balancing aerodynamic and hydrodynamic forces at different Reynolds numbers. I do not think he assessed the effect of dynamic perturbations and such as flipping over obviously was not evident. That might be initiated by a wavelet?
Thence he determined the ‘static’ equilibrium water speeds and drift angles from airspeeds, “in silico”, so I think that saying the outcome was CSD/computer generated using data from flotation experimentation is fair.
Both Annexes though are in French.
http://mh370.mot.gov.my/Appendix1.12A-2-Item1Flaperon-2-Appendices.pdf
On the CSIRO debris items, Entry 21, he took his coefficient from, “a value commonly used for small buoyant items such as seat cushions” (13th April report, p.15), so I do not think that counts as use of a debris item. On the other hand he did not utilise “Roy” for his Entry 21 work as you say so my revised suggestion is 2 for that, the 3 for Entry 17 being unchanged.
@All:
I’ve been in dialog with CSIRO’s David Griffin about drift modeling. In particular, I’ve asked him to simulate the drift paths of relevant drifters that were near the 7th arc to see how well his model does. In particular, he’s looked at an undrogued drifter 101703 that was near 23.6S latitude. He’s updated his thoughts on MH370 to include these results.
Drifter 101703 spent considerable time meandering near the 7th arc before it was transported across the Indian Ocean. As a result, its arrival time near Reunion Island was similar to the flaperon (July 2015). On the other hand, the vast majority (but not all) of the virtual particles in the drift model arrive much earlier.
I combined three of David’s videos into a single video that compares the drifter trajectory with the virtual particles.
My takeaway is there are probably a significant number of locations along the 7th arc and north of 25S where the transport of debris across the Indian Ocean is delayed. This, combined with the possibility that there was a significant delay between grounding and discovery of debris, leads to me to apply some caution in the interpretation of the previous drift results.
@Victor
…from the video you’d think we could find a piece of this thing if we got close the correct location. Also maybe inside Arc7 is better for finding something.
@All
Victor suggested I modify my drift simulator to show the tracks for a cluster of start points around a particular start latitude on the 7th Arc, rather than a single point.
Here is an example, using 9 start points around 20°S 104.5152°E on the 7th Arc, for trajectories lasting 480 days or until land is reached. The central track is marked with bold dots, the 8 surrounding tracks are marked with fine dots. The colour red is for a fast speed, orange a medium speed, yellow a slow speed.
The start points are shown in the table linked below:
https://www.dropbox.com/s/ym7vwg7cw8tyq8f/Cluster%20Analysis%2020S%20104.5152E.png?dl=0
The results are shown in the link below:
https://www.dropbox.com/s/h2n6bmhcm7c8ucu/Cluster%2020S%20480d%20Map.png?dl=0
The results show a considerable dispersion both spatially and temporally. Cluster item 7 (see table linked above) passes the closest to Reunion reaching 20.4473°S 55.4673°E after 376 days. I plan to repeat the simulation for other starting points along the 7th Arc.
Richard: Please explain the meaning of the color and scale for the results map.
Richard: Disregard previous question. Found the answer.
@Victor
You stated “My takeaway is there are probably a significant number of locations along the 7th arc and north of 25S where the transport of debris across the Indian Ocean is delayed. This, combined with the possibility that there was a significant delay between grounding and discovery of debris, leads to me to apply some caution in the interpretation of the previous drift results.”
There were a number of drifters caught up in gyres and delayed on their journey, in the MH370 timeframe.. Similarly, MH370 floating debris could be delayed.
I have analysed the weather between the 7th Arc and Reunion for the timeframe between 8th March 2014 and 29th July 2015 on a daily basis. There were 15 tropical cyclones which may have affected the path of floating debris during its transoceanic journey. Sometimes there were 2 storms at the same time. Each storm lasts a few days, but if you are in the wrong place at the wrong time, then it can have a significant effect. The storms only occur between November and April, but a trajectory starting in March 2014 and ending in July 2015 gets hit twice in March and April, both in 2014 and 2015.
I have decided to try and build a more refined wind direction and wind speed model into my simulator using the GDAS data for the time period, at least for the Reunion timeframe and from 106°E to 56°E and 14°S to 34°S.
@DrB
According to “6. Conclusions” on page 7 of Prof Dr Kunzel’s report:
“6. Conclusions
…
(1) Inside the five sets of recordings there was always a single speaker and
(2) …
”
[difficult to cut and paste the whole section from online format.]
But I cannot find mentioned “exclusion” part (at least not in section 6. above).
By the way, when I read the SIR, I also find that there are several places in the report studying/describing the hows and whys of the last set(s) of radio communication and the fact that it is against the company’s rules to leave the cockpit before reaching cruise level.
[For example, SIR (page 219):
“Evidence from the KL ATSC’s voice recording indicated clearly (in interviews with the Captain’s colleagues, friends and son), that the voice recorded was that of the Captain after the aircraft took off.”]
@mash,
Regarding whether or not the FO made one of the radio calls after take-off, Edward Baker is waiting for Prof. Dr. Künzel to respond to his query.
The complete quote from the report you showed is this:
“Results obtained with both approaches, and supported by auditory phonetic analysis by a trained listener (speech scientist), provide strong evidence for the assumptions that
(1) inside the five sets of recordings there was always a single speaker . . .”.
That conclusion seems to say that he heard only one speaker in the Group #5 series of the last 3 radio calls [i.e., the set called “Lumpur Radar (Area)”], and therefore he was comfortable with that “assumption”. However, there is no evidence of any analysis presented in the results tables for cases with more than one speaker. Perhaps he did analyze various combinations of speakers (but didn’t include these results in his report), or perhaps he simply listened to the recordings and was comfortable with what he had been told to assume – that it was standard practice at MAS for one person to make all the calls when in flight – and therefore he made that assumption in all his analyses.
Anyway, this is a point worth pursuing, and we await a clarification from Prof. Dr. Künzel.
@All,
I have been in communication with Dr. David Griffin of CSIRO regarding his drift modeling.
There are multiple links in the text below, and if problems with the links arise I will break this post into several parts.
In response to my query, Dr. Griffin has kindly provided a concise summary of the drift modeling methodology his group has used, and this has been incorporated in a revised version of my table of predicted latitudes on the 7th Arc. He also notes that their analyses have included the finding dates for all 25 debris items. He writes:
“We have taken the arrival times of all of them into account, in some way.
E.g., http://www.marine.csiro.au/~griffin/MH370/br15_pwent2d/beachdates_nonflap_25_08_25.gif
shows that the model’s median beaching date falls near the midrange of observed beaching dates if the crash was south of 26S. 35S becomes a ‘best fit’ only if we require the model’s 10th percentile beaching date to align with the onset of observed beaching. In other words, I don’t think it’s a simple job to tabulate the recommended crash site without being clear about the decision process. E.g. the Pleiades images are a separate issue entirely, that could or could not be factored in. At present, the debate is more about how much debris washed up before anyone reported any of it.
But how can we decide how long debris on beaches was overlooked? 6 or 12 months? Either is a large fraction of the total travel time. Assuming additional beach dates ~8 mo earlier means some items were 8 mo less in water than we’ve been assuming, so the best fit crash site will move north by ~8 months of ocean travel. According to our ocean model (see beach dates plot above) the best fit shifts from 35S to anywhere north of 25S, excepting 23-22S. If you want to assume 12 months of debris-overlooking, 23-22S is then also in scope. OI’s search has already covered the possibility of ~5 months debris-ignoring, assuming our model is correct.
However, note also that the flaperon is independent of the debris-ignoring issue.
http://www.marine.csiro.au/~griffin/MH370/br15_pwent2d/pfromto_1_flaperon_25_08_2510_20_99.gif
says the flaperon is very consistent with 36-30S, grudgingly consistent with 30-26, and inconsistent with 26S-8S.
http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dpf10_20_99_arc7_3608_25/20140901.html
shows we can make the flaperon arrive at Reunion by increasing the drift angle relative to wind (to bring the trajectory south) but this will have the flaperon likely to arrive up to 10 months earlier than it did. So rather than saying other flaperons were unnoticed, you would have to argue that (as well as fudging the drift angle) the flaperon took, by chance, a very slow route (the model has renegades still at 100E in Sep 2014).
Lesson: the next time a plane hits the water there should be more effort put towards watching beaches, like Australia did in late 2014. Kudos to Blaine for taking this job on himself.”
My interpretation of his results is that the flaperon hits Reunion at the right time (July 2015) if you use the empirically determined drift angle for the flaperon when starting from near 35S, but when starting from north of 25S you miss Reunion when using that drift angle. If you fiddle the angle to hit Reunion then you don’t get the right arrival time. He is currently exploring this further.
My revised table of predicted 7th Arc latitudes, incorporating Dr. Griffin’s model description, is at the same link HERE.
I searched 22 797 NOAA buyos and found common debris origin for eleven MH370 debris +/- 2° in a square 4-5 S and 80-81E. It looks like that debris had to travel through equatorial currents.
https://image.ibb.co/mYMNbe/SIO_11_debris_4_S_80_E.jpg
@mash,
You said: “. . . and the fact that it is against the company’s rules to leave the cockpit before reaching cruise level.”
Cruise level FL350 had been reached at the first FL350 call at 17:01. At the time of the last ACARS report at 17:07, the FL was still 350. So that rule was not an issue between 17:01 and 17:19 when the ATC handoff did not occur, or at any time in between, such as at 17:07:56 when the second FL350 call was made. It would not have been against the airline rules for one pilot to be away from the flight deck at 17:07 for the second call or at 17:19 for the third call or shortly thereafter.
@All
I previously published the results from my MH370 floating debris simulator based on the data from 280 undrogued GDP drifter buoys, for a cluster of start points around 20°S on the 7th Arc above:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18849
Here are the results, using 9 start points around 22°S 103.5638°E on the 7th Arc, for trajectories lasting 600 days or until land is reached.
The central track is marked with bold dots, the 8 surrounding tracks are marked with fine dots. The colour red is for a fast speed (above 0.970 knots), orange a medium speed (0.582 to 0.970 knots), yellow a slow speed (0.194 to 0.582 knots) and white is almost stopped (below 0.194 knots).
The bifurcation around Madagascar, from a start point centred on 22°S near the 7th Arc, is clearly demonstrated in this simulation.
The start points are shown in the table linked below:
https://www.dropbox.com/s/h986mfq3z2noy7t/Cluster%20Analysis%2022%C2%B0S%20103.5638%C2%B0E.png?dl=0
The results are shown in the link below:
https://www.dropbox.com/s/7pjf4iw00yqpsw1/Cluster%2022%C2%B0S%20600%20days.pdf?dl=0
Cluster item 7 (see table linked above) passes the closest to Reunion reaching 21.2996°S 55.8308°E, just 2 km off the south-east coast after 390 days.
I plan to repeat the simulation for other starting points along the 7th Arc.
@mash said: By the way, when I read the SIR, I also find that there are several places in the report studying/describing the hows and whys of the last set(s) of radio communication and the fact that it is against the company’s rules to leave the cockpit before reaching cruise level.
The top-of-climb was FL350. Both the first and second altitude calls occurred after reaching cruise level.
@DrB
@Victor Iannello
Assuming the first call was made as soon as the plane reached FL350, then both the pilot and the co-pilot should be in the cockpit at that moment (the time of the first call). So it is unlikely that the co-pilot did not notice the first call. Similarly, there is a substantial difference between breaking rules or not breaking rules; and therefore to ensure whether one leaves the cockpit before or after FL350. Hence whether leaving before or after, if done, it is still very difficult to explain why the co-pilot did not know what he had done; and the need to make that second call again himself.
And more or less the same argument for the pilot, if he did really leave the cockpit after the co-pilot coming back, or vice versa (?!) …
@Dr B. Flaperon drift.
“If you fiddle the angle to hit Reunion then you don’t get the right arrival time. He (David Griffin) is currently exploring this further.”
In parallel he and I have discussed this and other aspects of his drift studies as per the below. To my mind the accuracy of flaperon drift estimates could be problematic, as at its pages 4 and 5.
https://www.dropbox.com/s/0sddw1drbo4m9dp/Some%20comments%20on%20CSIRO%20studies%20of%20MH370%20flotsam%20drift%2C%20rev%201.docx?dl=0
In a previous comment, John said:
I searched 22 797 NOAA buyos and found common debris origin for eleven MH370 debris +/- 2° in a square 4-5 S and 80-81E. It looks like that debris had to travel through equatorial currents.
On Facebook group VeritasMH370, Dan Richter said:
I searched 22 797 NOAA buyos and found common debris area for eleven MH370 debris +/- 2° in a square 4-5 S and 80-81E. It looks like that debris had to travel through equatorial currents.
Unless I quickly get an explanation as to why there are identical posts from two different names, @John and @Dan Richter are both banned.
@David: My takeaway from your work is that the undrogued drifters follow trajectories similar to non-flaperon debris, and the drift behavior of the flaperon is far from settled science. So, what does this lead you believe about the range of possible impact points along the 7th arc?
re: Hits and Misses (Again)
No of hits = 1
No of misses = X (unknown – not yet disclosed?!)
Cannot wait any longer; so make assumption X is a ‘big’ number, as ‘big’ as no. of crew & passengers minus one (i.e. 239 – 1) [theoretically].
And make even bolder conjecture: Co-pilot was in the cockpit trying to make a call when the plane was over Penang. [No need to choose between option “lock-out” or “knock-down” anymore.]
Why? Probability – one hit and many many misses; the only possible reason is that the co-pilot has an ‘unfair advantage’. And the only possible scenario imaginable is that he was in the cockpit.
[Recall that co-pilot is the least likely person to switch on phone in this flight (analogy: learner driver in a driving test).]
@Victor. From just his reports the flaperon drift estimates looks pretty hairy to me. David Griffin is more confident. I think that stems from having witnessed its behaviour, the written word and the trials actually documented in the reports not imparting the whole scene (stray currents etc).
But based just on the reports and his confidence in the modelling more generally I find his Fig.3 mentioned in his 10th August update the most comprehensible and authoritative, ie that to do with beachings, specifically, in East Africa. (Both Dr B and I have referred to it and Dr B has described it above).
As a mind clearer I drew up the below from it, having plotted the actual recoveries of the 18 ‘likely’, highly likely’ etc items recovered in ‘East Africa’ (as DG has defined it, ie including Madagascar) and assuming there would be no more, given the time gap from the last. The 18 included the vortex generator, yet to be assessed
That outcome is striking and indicates that crash sites further up north and down south are very unlikely. To my mind to warrant searching there first needs a refutation of the message from that modelling but I note that Professor Pattriatchy independently seems to have come to the same broad conclusion.
Most others of us have neither the skills (me anyway) or the info to do that.
So I cannot foresee a good case being made for a search in those places without a source of error in those estimations becoming evident. While searches there are not ruled out as possibilities, an unlikely possibility does not a good case make. Meanwhile I share the view that the aircraft is along the arc already searched, missed or out wider. Fig 3 provides a clue as to the most likely sites, but IMO it is unlikely that a ‘cost effective’ site for searching these will be identified now. Maybe future data and modelling refinements will do that, as it did with finding HMAS Sydney.
https://www.dropbox.com/s/ujjs4svzhsqznfp/Model%20vs%20actual%20debris%20arrivals%20east%20African%20coast%20%283a%29.jpg?dl=0
@Victor. Last sentence should read, “Fig 3 provides a clue as to the most likely latitudes, but IMO it is unlikely that a ‘cost effective’ proposal for searching these further will be identified now.”
@David
You stated “To my mind the accuracy of flaperon drift estimates could be problematic”.
… and …
You stated “From just his reports the flaperon drift estimates looks pretty hairy to me. David Griffin is more confident.”
How can you draw any conclusions from possibly problematic and pretty hairy analysis?
Surely, you solve the possible problems and clarify the open issues first, before coming to a conclusion.
But no, you already conclude that MH370 did not end up north of 25°S (“crash sites further up north and down south are very unlikely”) and instead is along the 7th Arc already searched but missed or out wider (“the aircraft is along the arc already searched, missed or out wider”).
@David: David Griffin’s experimental work shows that an undrogued drifter drifts in a similar manner to replicas of MH370 debris. We also have drifter 101703 that crossed the 7th arc in March 2014 and meandered for months before it was transported across the Indian Ocean. David G’s drift model did not predict that would be a high probability path. That result tells me to treat his drift modeling results with due caution.
@Richard Godfrey. My conclusion was based on East Africa beachings, his Fig 3, and was based on his advice that there were more trials of ‘Roy’ than have been published. He indicated that they supported the general conclusion as to low-windage items drift being similar to the unbuoyed drogues.
My conclusion did not draw from flaperon drift at all. For the reasons I have put the estimates of that could be unreliable, IMO.
That will persist until more conclusive flaperon results are evident. However I do not see how those doubts undo the low windage result.
@Victor. The CSIRO Fig 3 outcome should be subject to continuing scrutiny but right now from that a more northern crash is improbable. Does one drifter overturn that?
Along those lines I note David Griffin’s 17th September entry below which says in part, “The track of drifter 101703 is well within the envelope of 120 model counterparts (all initially within 5nm of the drifter and thus equally representative) – but that envelope is very wide.”
http://www.marine.csiro.au/~griffin/MH370/index.htm
@Victor
@Richard
…I might see a new (old) flight path BEBIM to 78S67 (NZPG) at 400 knots. Equivalent to DOTEN to NZPG orig sim path, except BEBIM is the focal point. Struggling with winds at FL250 but I think it has potential maybe FL300.
Two End Point Options:
Plan-A ends @30S and pilot makes extreme long glide to BR using what I call “Alt Config with intentionality”…or Plan-B the pilot realized he could not make his Plan-A target, and diverts to Zenith Plateau deep trench @22S.
It is perhaps philosophically similar idea to Victor’s latest path: slows near Cocos/BEBIM, then takes a path south.
@David
My database of 280 undrogued drifters shows that there are MH370 end points north of 25°S that fit the data.
My latest findings show that the most likely MH370 End Point is around 24°S, which does not conflict with the CSIRO Fig. 3.
@David: Overturn his model, no. Lead us to treat his model results with due caution (the words I used), Without a doubt. It would be unwise to place too much emphasis on a single drifter. It would also be unwise to ignore it. Look at how much attention is placed on the flaperon despite the uncertainty of how that part actual behaved as it crossed the Indian Ocean.
The vast majority of the calculated tracks for drifter 101703 significantly diverge from the actual track. Few meander near the 7th arc. That means his model predicts the meandering would be a low probability event, but certainly possible. I fully acknowledge that with a sample size of N=1, it’s difficult to draw conclusions with certainty. However, I would also not conclude that his model’s predictions are a resounding success.
Unless there is something very unique about drifter 101703, its path tells me there is significant area along the 7th arc north of 25S where there would be a substantial delay before debris is transported across the Indian Ocean. If you look at other drifters in that vicinity, they also experience a delay before transport across the 7th arc.
I’d like to see how well David G’s model does against the other undrogued drifters. I believe it’s the best way to validate his drift model, and probably should have been done a long time ago.
@TBill: With pilot inputs after 19:41, there are an infinite number of possible scenarios. Unless there is evidence that helps us to narrow down the list of possibilities, I don’t see how latching onto a particular scenario is going to get us anywhere.
@Victor Iannello,
@David,
@Richard Godfrey,
Regarding the debris drift predictions, the ensemble of low-windage items is superior to any single item. A single item can tell us what is possible. An ensemble can tell us what is likely. For that reason, I don’t put much weight on using either the flaperon or a single undrogued drifter for predicting starting latitude. The flaperon is also a particularly bad example to model because of its instability in the water and its large drift angle. The low-windage debris, on the other hand, seem to be much more predictable, and they also allow more accurate predictions because of their similar drift characteristics to undrogued drifters.
Griffin’s plot of non-flaperon beaching dates is, in my opinion, an excellent way to present MH370 drift model predictions. I suggest other drift modelers produce the same type of plot to allow comparisons. Justifying a search farther north requires either a plausible explanation of why Griffin’s predictions there are wrong (as David noted) or, at the very least, a credible and independently done model demonstrating feasibility.
@Victor
Here is the actual plot of GDP Buoy 101703, showing that it started on 8th March 2014 at 23.416°S 103.006°E on the 7th Arc. It finally left the 7th Arc on 12th December 2014 have travelled around between 22°S and 26°S.
https://www.dropbox.com/s/u2vexgy2u50tojo/GDP%20Buoy%20101703%20Data.png?dl=0
By contrast, here is plot from my software simulation, based on the data from 280 undrogued drifters, showing a start in March at 25.0°S on the 7th Arc. It also shows a complex path between 24°S and 27°S before leaving the 7th Arc for Africa.
https://www.dropbox.com/s/j3ttcbo98kq2dvi/Drift%20Map%2025.0000S%20100.8756E%20687d.png?dl=0
@DrB: Based on what we know today, I think it is indisputable that the best way to validate David G’s model is to compare predictions with actual undrogued drifter data. We have one comparison with drifter 101703. I’d like to see other comparisons with other drifters using David G’s model and others models. Those models that predict the undrogued drifter data most accurately are most likely the best able to predict the path of MH370 debris, unless there is rationale for why the undrogued drifters behave differently than MH370 debris.
@DrB
You stated “I suggest other drift modelers produce the same type of plot to allow comparisons.”
I don’t know what other drift modellers are doing, but I have already prepared a comparison to the CSIRO Fig. 3 and am just waiting on a bit more data before publishing my initial findings probably tomorrow. I have already published the results for the cluster analysis at 20°S and 22°S and in the process of preparing the results for 24°S, which shows a good match to the CSIRO Fig. 3.
Meanwhile, I endorse Victor’s request “I’d like to see how well David G’s model does against the other undrogued drifters. I believe it’s the best way to validate his drift model, and probably should have been done a long time ago.” I have already published my results testing my simulation software against a set of undrogued drifters at the following link:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18782
re: Last Non-standard Call – Possible Alternate Explanation
Inspired by @Victor Iannello said what @Edward Baker claims:
“He makes another interesting observation regarding the final exchange where MH370 is handed off to HCM. He claims the typical radio procedure would be to receive the new frequency (in this case, 120.9 MHz), dial it into the radio as the standby frequency, read back what frequency was dialed in as part of the confirmation of the handoff, listen for the final transmission, hit the switch which swaps the active and standby frequencies, and, make the call on the new frequency.”
But this is an interesting “handoff” case:
(all [MYT] time below)
@0043:31 estimate of MH370 for waypoint IGALI: 0122
@0119:26 KL ATC transfer call (2 min 34 secs before estimate)
@0119:30 Pilot non-standard reply
@0120:31 Plane actually over IGALI
@0120:36 Mode S transponder off
I suspect that the pilot’s response/intention at that point is:
a) The transfer call at that time is not appropriate, because the plane is not over IGALI.
b) He also intends to contact HCM at the appropriate time/place: i.e. when the plane is over IGALI.
So the dilemma here is: if he repeats the frequency then he is not doing what he is saying (according to the “typical radio procedure” above anyway); hence the ‘simplified’ response “Good Night”. [In a way, it is sort of a compromise … KL ATC, ‘disciple’, HCM ATC, and possibly the ‘uncertain’ (“repeated call”) situation.]
@mash: First, it’s IGARI, not IGALI.
It’s not the decision of a pilot to decide when the appropriate time should be for a hand-off. Once MH370 received the ATC instruction to change frequency, the call to HCM should have been soon after, and not wait for a waypoint to be crossed.
@Victor
“With pilot inputs after 19:41, there are an infinite number of possible scenarios.”
…but there is only one scenario on the pilot’s home flight simulator. That puts one big X on the map, logic might dictate, at least for the intentional flight camp.
@TBill: I studied that possibility when I considered great circle paths. It’s the “45S” path shown in Figures 2 and 4, and crossing the 7th arc at 28.3S latitude. That area of the 7th arc was searched at a width of +/- 25 NM.
@mash
@Victor
Please note that repeating the message about maintaining FL350 is inconsistent with a message confirming the transfer to another ATC. In the first case, the sender tries to strictly follow the procedures, but in the latter case he bypasses these procedures.
I haven’t said much about the double FL350 report. But since it keeps coming up, I now want to offer my 2 Cents.
As a pilot with 50 years of flying experience in the US ATC system, I see nothing here to discuss. It is not evidence of anything that could be proven important, even if it is a bit unusual. There are no ATC rules that say a flight crew can’t repeat an altitude or any other position report. Maybe they just could not remember if the report had already been made, and they wanted to make sure the report was received. Who knows?
@Victor Iannello
SIR page 303:
“ii) The Transfer of Control Point (TCP) for flights on route R208 IGARI M765 BITOD L637 TSN…ZBAA is IGARI. Aircraft operating on this route shall be transferred by KL ACC to HCM ACC when the Radar Controller observes on the radar display that the aircraft is over IGARI or when the aircraft reports over IGARI. ”
Still the pilot has to consider:
(notice the “or …” part above too)
a) How to respond to the KL ATC?
b) When [say within margin – “soon after”] to contact HCM ATC?
(perhaps different reactions when transfer call is (judged) 1 min earlier vs 5 min earlier. [SIR keep repeating 3 min, don’t know why?])
@All
I previously published the results from my MH370 floating debris simulator based on the data from 280 undrogued GDP drifter buoys, for a cluster of start points around 20°S and 22°S on the 7th Arc.
Here are the results, using 9 start points around 24°S 102.4762°E on the 7th Arc, for trajectories lasting 728 days or until land is reached.
The track from the central point is marked with bold dots, the 8 surrounding tracks are marked with fine dots. The colour red is for a fast speed (above 0.970 knots), orange a medium speed (0.582 to 0.970 knots), yellow a slow speed (0.194 to 0.582 knots) and white is almost stopped (below 0.194 knots).
The bifurcation around Madagascar, from a start point centred on 24°S near the 7th Arc, is clearly demonstrated in this simulation.
The start points are shown in the table linked below:
https://www.dropbox.com/s/9g1dtmbiekwc44h/Cluster%20Analysis%2024%C2%B0S%20102.4762%C2%B0E.png?dl=0
The results are shown in the link below:
https://www.dropbox.com/s/ytkigbas9lklrjw/Cluster%2024%C2%B0S%20728%20days.pdf?dl=0
Cluster item 8 (see table linked above) passes the closest to Reunion reaching 19.6323°S 55.8007°E, 144 km off the north-east coast after 435 days.
The simulation from 24°S starts to align with the key MH370 floating debris discoveries:
The earliest discovery of the Flaperon at Saint-André, Reunion, discovered after 508 days, where the arrival off the coast is simulated after 435 days (-73 days).
The most southerly discovery of “Roy” at Mossel Bay, South Africa, first discovered after 655 days, where the arrival off the coast is simulated after 732 days (+77 days).
The most easterly discovery of the Cabin Divider at Var-Brûlé Beach, Rodrigues after 753 days, where the arrival off the coast is simulated after 530 days (-223 days).
The most northerly discovery of the Outboard Flap at Pemba, Tanzania, discovered after 835 days, where the arrival off the coast is simulated after 519 days (-316 days).
I plan to repeat the simulation for other starting points along the 7th Arc.
@DrB, @Victor, @David,
Here is a link to a comparison between the CSIRO Fig. 3 and my drift data:
https://www.dropbox.com/s/kspjp9oy3rduapb/CSIRO%20vs%20RG%20Comparison%20Latitude%20vs%20Time.pdf?dl=0
I have used the data from my drift analyses previously published here:
http://mh370.radiantphysics.com/2018/07/20/godfrey-drift-model-says-mh370-might-have-crashed-further-north-on-arc/
… and here:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18782
… as well as the data from my cluster analyses fro 20°S, 22°S and 24°S previously published here:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18849
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18864
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18892
@airlandseaman: The implications of that repeated message can be important. If it was spocken by the same person, it would be at least strange and could be interpreted as a message to the ground for s.th. we don’t know. It is even more interesting if it was spoken by two different persons because then we could interpret it so that one pilot had left the cockpit and the one in the cockpit wasn’t sure if the message from ATC had been answered yet. Only one pilot in the cockpit is very interesting because that was the opportunity to lock the door, or for the one who left to go down into the EE Bay.
@DrB
@Victor Iannello
@mash said: “Assuming the first call was made as soon as the plane reached FL350, …”
Sorry making a wrong assumption above; in fact, according to SIR (page 5)
”
MH370 climbing to FL350 at 1650:11 UTC [0050:11 MYT]
MH370 maintaining FL350 at 1701:17 UTC [0101:17 MYT]
”
Because I always thought that “maintaining” somehow meant “just reached” (or “climbing to” as above). And I still don’t quite understand the purpose (of the time ‘flexibility’) of the call.
@mash: You might be confusing the calls. At 1701:21, there was a call stating that MH370 was maintaining FL350. That same call was repeated at 1707:56.
@mash
@Victor
At 16:50 ATC cleared MH370 to climb to FL350. According to FR24 MH370 reached FL350 between 17:01 and 17:02. It seems to me that the first report on maintaining the FL350 was reflex reaction immediately after reaching the cruising level.
The IG’s view:
https://www.airlineratings.com/news/mh370-left-flawed-investigation/
@Richard Godfrey,
Thanks for posting the comparison of some of your recent drift calculations with Griffin’s results.
If I am interpreting your comparison plot correctly, your predictions for 20/22/24 degrees south latitude show the flaperon arriving early by about 2-4 months. For the low-windage items arriving in the vicinity of Africa, your model predicts cluster median arrivals from 1-7 months before the FIRST piece was recorded there, and roughly 5-11 months before the median arrival date.
Am I missing something here? Or should I conclude, as seems apparent, that your GDP model does not match the arrival times for those 20/22/24S latitudes for either the flaperon or for the low-windage debris items?
Another missing item from the investigation seems to be an interview or account by the UAE343 pilots who were about 30 NM behind 9M-MRO flying NW on N571. What did they see? The aircraft should have been plainly visible to them. Did they recall seeing it? Did it turn left or right or do a lateral offset maneuver? What about relative altitude?
I am unaware of any information being published concerning 9M-MRO sightings by nearby aircraft. Does anyone recall this being addressed?
As has been discussed previously, another missing item from the investigation is a search for mobile phone connections possibly made by the passengers’ phones after take-off. Since the investigation did not obtain the mobile numbers from the NOK, they may not have had a complete list. At present there is no indication this search for tower connections was ever done for passengers’ phones.
@mash
@Victor
After reflection, I do not see any rational reason based on good aviation practice, which could require reporting the maintainig of the FL350 after more than 6 minutes since its achievement. The flight was under radar control, and ATC did not issue a clear command requiring a report on maintaining the designated level. If necessary, ATC could ask for the current FL
@DrB
I agree with you UAE343 is a very key nearby flight. I would *assume* the JIT had complete flight path details on EK343 (we which do not have because the FR24 and FlightAware data do not go that far out). One thing we could ask for is KLATC commuications with UAE343.
If UAE343 was on N571 and 6 minutes behind, then an OFFSET puts it 3 minutes behind and 15-nm to the left. Closing the offset at IGOGU gives us a head on crash, figuratively speaking.
Another important flight could be EK425 from Perth, as I estimate it would be about at 22 South at Arc7 at about the crash time.
@DrB
At present there is no indication this search for tower connections was ever done for passengers’ phones.
True, and it is very annoying (to me). The lack of transparency is bothersome relative to this specific question, as well as many others. There is little doubt in my mind that the Malaysian government is engaged in a smoke screen.
@DrB
You are correct that I am stating that the arrival times for start points at 20°S and 22°S near the 7th Arc appear to be too early. I therefore discount these latitudes as unlikely.
However, the median arrival time for the start point at 24°S near the 7th Arc is 26 days before the first discovery in Africa, this is a good fit in my view. There was surely a time gap between arrival and beaching as well as between beaching and discovery. For both processes to have taken 26 days is not unreasonable.
The range of arrival times from the cluster analysis for a start point at 24°S is from 175 days before to 65 days after the first discovery. This only shows what is possible, not what is probable.
You miss the point that the cluster analysis from 24°S shows tracks passing close to all the key MH370 floating debris discoveries:
1. The earliest discovery of the Flaperon at Saint-André, Reunion.
2. The most southerly discovery of “Roy” at Mossel Bay, South Africa.
3. The most easterly discovery of the Cabin Divider at Var-Brûlé Beach, Rodrigues.
4. The most northerly discovery of the Outboard Flap at Pemba, Tanzania.
@David @IG
“The IG’s view:
https://www.airlineratings.com/news/mh370-left-flawed-investigation/”
I fully support the IG view on this matter. The SIR report is increasingly disappointing to me.
The picture on top of the Geof Thomas article is uncanny for me, I have a similar flight sim picture in a new essay. For me the picture represents MH370 scooting away from Arc7 on 8-March-2014 with clouds and sun angle about the way it may have looked (cloud cover may have been more solid though, is my thought).
There’s a new article from the Herald Sun that highlights the captain’s risky social media posts, and in particular his obsession with a young model from Penang.
This is not totally new information, as Folder 4 of the RMP Report devotes a section to the Facebook page of the model. Frankly, I think there is other, stronger evidence that suggests the captain was likely responsible for the disappearance. Nonetheless, this article will surely get a lot of attention.
For those that can’t get past the paywall, here is a link to the new article in the Herald Sun.
@Richard,
You said: “However, the median arrival time for the start point at 24°S near the 7th Arc is 26 days before the first discovery in Africa, this is a good fit in my view. ”
Comparing the MEDIAN predicted arrival time to the FIRST discovery is an apples to oranges comparison. Comparing apples to apples results in a poorer match.
You said: “You miss the point that the cluster analysis from 24°S shows tracks passing close to all the key MH370 floating debris discoveries:”.
That geographical match is interesting but the timing is off. I suspect that more southerly starting latitudes may also put debris over a similar geographic range but with a better agreement in arrival dates. It will be interesting to see if your cluster analyses at more southerly latitudes show this effect. David Griffin’s predictions seem to show the debris moving up along the arc before turning rather sharply to the west near 24S. If I am understanding this correctly, the distribution of landfall locations won’t change must for starting points south of 24S, only the beaching dates. The farther south you start, the longer is the travel time and the later is the beaching date.
@TBill,
You said: “If UAE343 was on N571 and 6 minutes behind, then an OFFSET puts it 3 minutes behind and 15-nm to the left. Closing the offset at IGOGU gives us a head on crash, figuratively speaking.”
I did a rough estimate that if MH370 turned left from the right offset path to cross perpendicular to N571 at IGOGU, the separation would be about 9 NM if UAE343 was also traveling at M0.84, but about 15 NM if its airspeed was close to MRC at M0.78 (which is more likely). So 9M-MRO with its navigation lights on should have been visible to UAE343 from MEKAR through IGOGU, although it probably would never have gotten close enough to raise concern by the UAE pilots. If the nav lights were off, then they would not have seen it at all visually. Since 9M-MRO’s TCAS was not operating, there would have been no anti-collision warning in UAE343 even if the separation was small.
@Nederland
Trying to remember your flight path, I might be coverging on a similar solution to yours.
@Greg
Thanks for the (actual “FL350”) clarification.
@DrB
@Victor Iannello
Apologies once again for the silly (comprehension/communication) mistakes.
Hi,
About the repetiton of the level 350:
From the flight plan received by ACARS (see log pages 65 to 67), the planned altitude at IGARI (p. 66) was level 330, with a change to 350 after BITIS. Could the repetition of the level 350 was due to the difference between actual and planned altitude?
@PaxLambda said:
“Could the repetition of the level 350 was due to the difference between actual and planned altitude ?”
One of the things that is an issue, that I think needs to be addressed, is that the “complete” recording of ALL calls on the relevant ATC frequencies has not been released. Knowing what other traffic the ATCO was “working” – in multiple sectors – on multiple frequencies – up to the IGARI hand-off, would give us a “feel” for the ATCO’s workload, and perhaps an “insight” into whether or not MH370 may have had, or perceived, a valid reason for making the double FL350 calls.
@PaxLambda: Before the “push back and start”, MH370 requested and received airway clearance for FL350. The two altitude calls were made at that flight level. There should have been no confusion.
re: Cabin-initiated & Cockpit-initiated Diversion [Impossibility/Insolvability]
While “timeline-analysis” might indicate that a cabin-initiated diversion is very unlikely due to time limitation (i.e. between last call and start of U-turn recorded by SSR); a “time-of-attack-analysis” might also suggest that a cockpit-initiated diversion is very ‘unlikely’ [strange/peculiar] due to time selection (i.e. time between last call and transponder off).
Specifically, it makes one wonder why the transponder was not switched off immediatly after the last call, or at least when the plane was over IGARI [supposing the pilot is really a very very meticulous person]; but some five odd seconds later (altogether some 1 min 5 sec after the last call). Why wait, why wait so ‘long’ – what did he do during that ‘idle’ time? What a ‘wasted’ opportunity [earlier than expected transfer call]!
[Analogy: One (psychologically) definitely wants to start an escape as soon as possible; or a race at the same time as others – not some 5 secs later.]
@mash: If the captain intended to wait until being seen on radar as passing IGARI before going dark, I don’t think 5 seconds would be considered “long”, especially since the radar sweep time is around 4 seconds. In fact, I’d call it fairly well correlated.
@Victor
I have a new (old) flight path vision, suggesting MH370 flew the path shown on pilot’s home flight simulator, at slow speed 400 knots. That puts MH370 at 30 South and Arc7, already searched, but a long glide is possibly suggested by the analysis.
Did MH370 Fly a “Low and Slow” version of the Pilot‘s Flight Sim Path?
However, holy mackerel, @Nederland had previously suggested nearly the exact same flight path (BEBIM to NOBEY via 171 South Heading) ending at 30 South, thus the current paper is essentially the converging independently on the exact same prior idea. Richard Godfrey of course also has a prior pin at 30 South reflecting a somewhat similar flight path to Wilkins Field in Antarctica.
@Victor Iannello
Suppose there are 2 options immediately after the “handoff” call, namely:
1a. Switch off transponder
1b. Contact HCM
and still the same 2 options when over IGARI 1 min later, namely:
2a. Switch off transponder
2b. Contact HCM
As explained previously, the reason for choosing “2b” instead of “1b” might be the plane is not over IGARI yet (a ‘legitimate’ one – under the circumstances, perhaps); but there seems to be no such ‘obvious’ reasons for choosing “2a” instead of “1a” – though I agree that it’s still possible [such as a “meticulously” calculated (tradeoff) advantage, say].
Besides, contrary to ‘popular’ opinion, I think one would do the most standard things if one were to ‘escape’ afterwards; therefore, the “non-standard” things indicate something abnormal is going on – since an ‘abnormal’ someone would “pretend that everything’s normal” …
@Victor
@mash
As I continue to pursue the accident scenario, I need to explain why there was no call to HCM before the transponder went off. We have about a minute to account for when it should have been able to transmit the ‘check in’ call to HCM. Here are a few reasons that may normally delay a call—-
1/ The HCM frequency was busy with other transmissions. Has anyone heard any unedited HCM ATC transcripts? I don’t believe we have had access to any. Even unedited Malaysian ATC transcripts would be good so we can get a feel of how busy the frequency was.
2/ The Pilots had miss-dialled the next frequency. It’s common to miss-hear a 3 instead of the 2. However, this is unlikely as both pilots were on home territory and would have known the frequency even without being told.
3/ An interruption from the cabin crew on the interphone.
4/ Cabin crew were bringing in their meals.
5/ Distraction caused by the pilots being mid-conversation(training flight), and then had forgot to check in.
6/ Distraction caused by one pilot taking a loo break, then forgetting to check in.
Are there any other reasons anyone can think of?
@Tim. In accident scenario, I think the emergency would have to start unfolding around or very shortly after the last call if transponder off corresponds to switching of either individual circuits or entire circuits. You would need a little time between problem emerging, seriousness becoming apparent/checklists being run and actions being taken.
@Paul Smithson,
My favourite scenario is a sudden rupture of the O2 bottle that takes out the electrics, mainly on the left side of the MEC. The first checklist the pilots had to do is the rapid decompression checklist.
@All: It is being reported by CNN that a US diplomat in Madagascar was found dead in his home on Friday. Here is the story as reported in the WSJ:
A U.S. foreign service officer died while serving in Madagascar, the State Department said Monday, adding that a suspect is in custody and investigations by local authorities and U.S. officials are under way.
The unnamed diplomat was found dead at home during the overnight hours Friday, State Department spokeswoman Heather Nauert said in a statement. No cause of death was provided.
Ms. Nauert expressed sympathy to the officer’s family and to “the U.S. Embassy Antananarivo community.”
Ms. Nauert said U.S. investigators have opened an investigation, as have Malagasy authorities. She declined to provide further information out of respect for family members and because of the continuing investigation. Local officials couldn’t be reached.
Madagascar, one of the world’s poorest nations despite reserves of nickel, cobalt, gold, uranium and other minerals, has struggled with high crime rates and political instability. A 2009 coup prompted an exodus of foreign investors. Some streets of the dilapidated capital, Antananarivo, have dedicated tourist police stationed to guard foreigners who could be targeted by thieves.
The island is bracing for elections in November. Thousands protested earlier this year after a legal amendment that would have prevented the popular former president from standing for office.
Readers here might recall two previous blog posts related to a Malaysian diplomat that was killed in Madagascar before he was scheduled to collect MH370 debris to send to Malaysia.
Slain Diplomat Was Helping Get MH370 Debris to Malaysia
New Pieces Possibly from MH370; Blaine Gibson Threatened in Madagascar
Local, Malagasy, media reports that the gendarmes & police have a suspect ‘bang to rights’, caught at the scene. Such efficiency!
@Don Thompson: I suspect the US is pursuing this case very vigorously, and we’ll soon learn more.
“Nature – International Journal of Science” is ranked as one of the most cited scientific journals in the world. @DrB reminded me of a paper published in Nature on 11th April 2017 entitled “General characteristics of relative dispersion in the ocean” co-authored by a contact of mine, Gug (Guglielmo) Lacorata and featuring MH370.
https://www.nature.com/articles/srep46291
I was given a private copy of the paper back in February 2017, following a substantial email exchange with Dr. Guglielmo Lacorata of the Institute of Atmospheric and Climate Sciences (ISAC) in Lecce, Italy. This is the reason for the kind acknowledgement in the paper of my contribution on MH370, although I disagreed with the paper on several points (backtracking, only 4 drifters used, only 1 drifter crosses 7th Arc, only 1 drifter in March timeframe, etc.), which was subject of multiple emails with Guglielmo (or Gug, as he prefers).
The paper describes the different dispersion regimes in the world’s oceans and finds there is a common thread for all 11 major ocean areas. Of interest in the case of MH370, is the dispersion in the Equatorial Indian Current System (EICS), which is governed by three dispersion regimes: exponential growth for 1–30 km; Richardson law for 30–300 km; and diffusive growth beyond 300 km. The paper describes the back tracking of 4 GDP Buoys from Reunion to the 7th Arc. The paper concludes that the dispersion for a transoceanic journey between Reunion and the 7th Arc (after an average point-point distance of 3,419 km and an average actual track distance of 11,950 km) is around 500 km.
David Griffin of CSIRO recently published the track of GDP Drifter 101703 compared to the CSIRO simulation. Victor kindly compiled the results into a single video. Victor requested more comparisons with actual GDP Drifters saying “I think it is indisputable that the best way to validate David G’s model is to compare predictions with actual undrogued drifter data.”
A comparison of the actual track of GDP Drifter 101703 after 502 days with the track of my simulator after 561 days (10.5% slower) and the CSIRO simulator after 502 days is shown at the following link:
https://www.dropbox.com/s/t3tpyli3fkuv8be/101703%20CSIRO%20Simulator%20Comparsion%20502%20Days%2076%20Particles%20Dispersion.pdf?dl=0
The dispersion shown by the CSIRO simulator is extremely wide and only 7 out of the 120 particles are found in the 500 km radius circle centred on the end point of GDP Drifter 101703. It behoves David Griffin to explain why the dispersion shown in his model is so large. As Victor says “those models that predict the undrogued drifter data most accurately are most likely the best able to predict the path of MH370 debris, unless there is rationale for why the undrogued drifters behave differently than MH370 debris.”
@Tim
7/ A 4D alien ‘walked through’ the 3D cockpit door, perhaps in the middle of the frequency change process.
[Just like a 2D creature cannot ‘see’ the shape/pattern of a 3D world; neither can one solve a 3D jigsaw puzzle by assuming/requiring it to be 2D only. ]
@Tim
The ZS simulator data makes any accident scenario ridiculous. As well as many of the events of the diversion.
@DennisW,
The SIR says there was nothing other than normal gaming activity on the simulator. The reported simulator route showing a line into the middle of the IO released to the Australians by the FBI, must be seen as an attempt to frame Z and close this investigation. Morally that is wrong.
@Tim said: The reported simulator route showing a line into the middle of the IO released to the Australians by the FBI, must be seen as an attempt to frame Z and close this investigation. Morally that is wrong.
I strongly disagree. The ATSB’s reporting of the simulator data was factual. It would have been morally wrong to hide the facts surrounding the case.
I have received a lot of criticism for my analysis of the simulator data, and drawing the conclusions that I have. One person said that I have deliberately manipulated the analysis as part of a masterful scheme of mathematical deception. (That person has been pruned from this blog.) In fact, the paper that I co-authored with Yves Guillaume in November 2016 that concluded that the sim data was from a single flight, with manual manipulation between saved data files, has stood the test of time, and proven to be correct.
If you believe that those that have linked the disappearance of MH370 to the simulator data are “morally wrong”, then you are contributing to the wrong blog.
@All: In a previous comment, in making observations of CSIRO’s modeling of the drifter 103703, I said:
As a result, its arrival time near Reunion Island was similar to the flaperon (July 2015). On the other hand, the vast majority (but not all) of the virtual particles in the drift model arrive much earlier.
Two private discussions lead me to correct this statement. The CSIRO results show a large dispersion of virtual particles. However, for a significant fraction of particles, the CSIRO does properly predict there was a delay before transport across the Indian Ocean, with some virtual particles near the “middle of the pack” approximately following the trajectory of the drifter.
Comparisons of the CSIRO drift model with the trajectories of other drifters would be helpful, although the large dispersion of virtual particles makes it difficult to validate the drift model by using the paths of single drifters.
@Tim
Zaharie Shah’s home flight simulator was featured by Zaharie Shah in YouTube. There can be no doubt it existed.
The data recovered has undergone the expert scrutiny of Victor Iannello and Yves Guillaume, who discovered and published a list of facts demonstrating a consistency in the data in several points, some of which were missed by the official enquiry.
In consideration of all the other evidence against Zaharie Shah, motive, means, opportunity, politics, behaviour, obsession with models, Facebook posts, etc., how you can conclude he was framed is beyond belief.
Re: Pilot’s Recovered Simulator Data
In my proposed new (old) flight path posted above (24-Sept 6:18PM), I argue that the BTO/BFO math suggests MH370 may actually be located along the path of the pilot’s simulator flight. It turns out @Nederland had previously derived the same mathematical conclusion back in 2017.
@Victor Iannello said:
“@mash: If the captain intended to wait until being seen on radar as passing IGARI before going dark, I don’t think 5 seconds would be considered “long”, especially since the radar sweep time is around 4 seconds. In fact, I’d call it fairly well correlated.”
[Counter-argument attempt – not sure whether maths completely correct or not.]
Assuming 3 secs is the cutoff point for switching off the transponder (for an IGARI attack):
a) Uniform probability distribution (0-5 sec: [—.—.—|—.—])
Pr(hit before 3rd sec) = 40%
Pr(hit after 3rd sec) = 60%
b) Triangle probability distribution (say 1s = 4.5 sq, 2s = 3.5 sq, …)
Pr(hit before 3rd sec) = 6 sq/12.5 sq = 48%
Pr(hit after 3rd sec) = 6.5 sq/12.5 sq = 52%
Conclusion:
Cannot determine any correlation between IGARI attack (by pilot) and transponder off time.
[Hint: Transponder definitely was not switched off before 1st second, according to stated “sweep time” above.]
@Victor,
I’m sorry, I don’t mean to cast doubt on all the fine analysis on this blog. But as someone who doesn’t understand the first thing about hidden simulator data files, I have to rely on the official reports. The last report in the SIR says the RMP found nothing suspicious on Z’s computer. So what is the truth? For me, a suspicious route on the computer is the only piece of evidence implicating Zaharie. The rest can be accounted for by a decompression accident.
Is it possible the computer data you were given had been tampered/manipulated in any way?
@Tim: It is impossible to say with certainty that ANY of the evidence has not been manipulated. So should we ignore all of the evidence we have? Or just the evidence that implicates the captain?
With what we know today, it is beyond me why anybody would blindly trust the statements made in Malaysian reports and by Malaysian officials. If you want evidence that the Malaysians have been less than forthcoming, just re-read the contents of the current blog article and the one preceding it. That’s just the tip of the iceberg.
@Victor
I am constantly amazed by the variation of conclusions made by well-educated people when presented with the same data. I see it relative to:
1> MH370
2> Vaccinations
3> Global warming
4> Bret Kavanaugh
5> metoo movement
6> Kaepernick
There is no end to it. My guess is that @Tim is still living with his parents. The biggest differentiator, IMO, is generational.
@Victor:
I just stumbled upon the press conference at the end of July.
There is a sentence uttered by Kok Soo Chon, which caught my eye:
“we are not of the opinion it could be an event committed by the pilot”
I see that you also referenced it in 2 of your postings:
http://mh370.radiantphysics.com/2018/07/20/godfrey-drift-model-says-mh370-might-have-crashed-further-north-on-arc/#comment-17197
http://mh370.radiantphysics.com/2018/07/30/mh370-safety-report-raises-many-questions/#comment-17278
I find this statement quite bizarre.
If they said, they don’t believe it was the pilot, ok. But the wording seems to exclude the mere possibility of an “event committed by the pilot”. But then again, it may just be a language issue ?
@DennisW: I’m not sure it’s generational, but getting a consensus on anything related to MH370 is not easy.
@Peter Norton: The investigators were trying to deflect blame away from the captain without using language that could later prove to be wrong. You see the non-sensical result.
https://www.indiatoday.in/india/story/air-india-pilots-new-york-landing-instrument-failure-1342373-2018-09-17 Interesting story of multiple instrument failures on a 777-300. @Andrew, can you enlighten us if radio altimiters and auto land function are left Ac bus dependent?
@Paul Smithson
No – There are three sets of autoland components, each consisting of an ILS, radio altimeter, autopilot/flight director computer and ILS antenna switch; hence the term ‘triple channel’ autoland. During an autoland operation, the electrical system divides into three separate channels when the aircraft descends below 1,500 ft, to ensure that each set of autoland components is powered by a separate source. The loss of any one channel reduces the system redundancy and increases the required landing minima, but does not prevent an autoland. The loss of two channels will cause the loss of autoland capability. It has been speculated that the Air India aircraft had dispatched with an acceptable electrical defect and that the loss of autoland and other systems was caused by a problem with the autoland bus isolation during the ILS approach.
Thanks
@Andrew, is there *any* aspect of standard descent and instrument approach routine that is dependent on Left AC bus powered?
Re: MH370 Accident Cause
>Prime minister Razak said apparent deliberate action until end of his term in office
>More recent gov’t says possible 3rd party hijack in verbal description of SIR
>SIR report concludes no known cause of the accident.
I have been trying trying to get Wikipedia article to say apparent deliberate action, but they counter that officially there is no known cause.
@TBill
If you ignore the conspiracy theorists, the vast majority are still sitting on the fence regarding the cause of the MH370 tragedy. You will find a few people like me who are convinced that ZS is the perpetrator, you will find a few people like @Tim who think ZS is innocent, but the majority are undecided.
Soimilarly, the vast majority are still sitting on the fence regarding where to search next. You will find a few people like me who are convinced we should search further north, you will find a few people like David Griffin who are convinced we should search further out from the 7th Arc, but the majority are undecided.
Clearly, in both cases, the case has not been convincing for the majority.
@Richard
Good summary…
@all
It is Google’s 20th anniversary. That timing makes sense to me, because for TWA-800 in 1996, I recall using other internet search engines (probably Yahoo or Dogpile) to find out that fuel tank explosion was what the experts were thinking (vs. missle theory).
@Tim,
I myself am a noob in terms of aviation and related.
But the last thing I would do is “rely on official reports”.
Regardless of the issue.
I will do my own research and listen to experts, try to gauge and balance both sides before I make up my mind.
“Officials” are far too biased and have specific agendas, so that’s a definitive “no” in my book.
Just sayin’
@AuthurC,
I think you’re right, official reports are not always correct. Otherwise you wouldn’t have 2 different reports of the same accident, thinking of EgyptAir 990 & the SilkAir B737.
Let’s hope we don’t end up with 2 different reports on MH370!
@paul smithson
RE: ‘…is there *any* aspect of standard descent and instrument approach routine that is dependent on Left AC bus powered?’
None that spring to mind. Essential systems are duplicated (sometimes triplicated), with different power supplies.
@Andrew thanks again.
@Victor. From the below, talking about AF447 wreckage recovery, “One of the problems that the team faced was that currents had pulled the undersea wreckage 24 miles away from the surface wreckage.”
Quite a distance.
https://www.thesun.co.uk/news/7361344/mh370-flight-plane-may-never-be-found/
@David
“One of the problems that the team faced was that currents had pulled the undersea wreckage 24 miles away from the surface wreckage.”
That is a misleading journalistic statement. Going through the details of the metron search shows that the aircraft was found very close to the last known position (which we don’t have for MH370). Take a look at figure 1 and figure 2 of the linked Metron document.
https://www.bea.aero/uploads/tx_elyextendttnews/metron.search.analysis_01.pdf
@all
Speaking of the Metron document, take a look at Table 3. The distance of the historical incident wreckage from the last known position is consistent with the previous decisions on the width. Searching wider is simply a bad idea.
@DennisW: Unless manually glided after fuel exhaustion, which I think we can’t completely dismiss.
@Victor,
Yes, I agree. That scenario makes very little sense. I am comfortable with the mainstream thought on the issues. I just wish I could find a way to make the cost of a renewed search sensible. I am not able to do it.
@Dennis W. “That is a misleading journalistic statement.”
Thank you for your reference.
Yes it was, as was, “They also had….an accurate final position for the aircraft”, which should have read, “accurate last known position”.
The 24 miles of the article might come from Fig 31, page 65 of the below, which depicts the nearest wreckage as being around this from the crash site.
https://www.bea.aero/docspa/2009/f-cp090601.en/pdf/f-cp090601.en.pdf
The furthest wreckage there is about 160 NM, wreckage recovery being over 20 days. The spread of 120˚ is way beyond MH370’s. It contained high windage items but I wouldn’t have thought that would affect spread much. The MH370 spread was more like 30˚.
More recovery, weather and current detail from the AF447 experience might disclose what the drift rate was for the MH370 items of low windage.
I see they did reverse drift studies though with no immediate success and there were several Bayesian updates over the several phases of the search, entailing feeding in the outcomes from earlier phases.
One other point of interest at p.77 of the above is that a 7 m piece of fuselage with 11 windows was found 2 km from the main wreckage field.
CORRECTION
I apologise for misrepresenting David Griffin’s views.
What I should have said in my previous comment is:
“@TBill
If you ignore the conspiracy theorists, the vast majority are still sitting on the fence regarding the cause of the MH370 tragedy. You will find a few people like me who are convinced that ZS is the perpetrator, you will find a few people like @Tim who think ZS is innocent, but the majority are undecided.
Similarly, the vast majority are still sitting on the fence regarding where to search next. You will find a few people like me who are convinced we should search further north, you will find a few people like Charitha Pattiaratchi who are convinced we should search further out from the 7th Arc, but the majority are undecided.
Clearly, in both cases, the case has not been convincing for the majority.”
David Griffin’s view is north of 25°S is “not consistent with either drift modelling or the satcom data and flight path simulations” and searching farther off-arc between 25°S and 39.6°S is “inconsistent with physical and satellite (Burst Frequency Offset) evidence that the aircraft made an uncontrolled descent, rather than gliding a long way off-arc”.
The confusion arose because Charitha Pattiaratchi stated that oceanographers believe that MH370 is not north of 25°S and is farther off the 7th Arc. I had erroneously assumed that Charitha Pattiaratchi was including David Griffin in his general statement about what oceanographers believe.
An observation on the requested clearance, with the WMKK ATC ACD service.
During the dialogue with WMKK ACD at 1626:22UTC, the crew requested clearance to FL350. However, the FMS plan was to attain only FL330 during departure within the Kuala Lumpur FIR.
The operational flight plan, signed off by the MAS dispatcher & commander, showed that the FMS plan for altitude at TOC was FL290 with a delayed first step-climb to reach IGARI at FL330. A later step climb was planned at TSN (Tansonnhat VOR) so as to reach FL350.
Why would the commander request a higher altitude than detailed in the FMS plan?
The flight plan sububmission, and sign-off, was timed at 1505UTC, which corresponds with related ground-ground message traffic with f:wz and Flight Explorer. The SIR records that the flight was released at 1515UTC, coincident with the FO signing in for duty. Certainly, any discussion of last minute details for the flight plan were completed before the arrival of the FO.
And… yet more evidence of political interference in Malaysia Airlines’s business.
Najib Razak visited Washington DC in Sept 2017 where an MoU was signed as an intent to purchase eight B787 and more B737 airliners. Sep 13th, an inauspicious date.
Barely five weeks later, 17th Oct 2017, Malaysia Airlines lost another Group MD. The third to exit from that role in 30 months.
This week, Reuters reports that Malaysia Airlines’ MoU has lapsed without execution.
Najib’s failed gambit to ingratiate himself with the incoming US administration.
@Don Thompson
RE: ”An observation on the requested clearance, with the WMKK ATC ACD service.”
Aircraft departing Kuala Lumpur are often held down at lower than optimum flight levels during the initial part of the cruise, due to aircraft operating through the adjacent Singapore FIR. My guess is that the computer-generated OFP for MH370 reflected the expected levels after departure. Nevertheless, it is normal practice for the crew to request their desired level from ACD, even if that’s higher than the level shown on the OFP. ATC then allocates the highest level available, up to the requested level. In this case, FL350 would have been much closer to the optimum level for the aircraft weight.
@Don Thompson, @Andrew: I think the early airway clearance to FL350 and the clearance to fly direct to IGARI are both consistent with light traffic in the sector. FL290 is significantly lower than the optimum cruise altitude.
@DennisW
It *might* make sense to first search +/-25nm from Arc7 say 20-25 South, even though glide beyond Arc7 is starting to look like a definite possibility.
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18784
@Peter Norton asked in above link:
<<>>
https://www.express.co.uk/news/world/470172/Malaysia-Airlines-MH370-Who-cut-off-pilot-s-final-call
According to link above:
<<>>
So at least an “insider” gave the above ‘opinion’ …
[Wonder what is the result of THE ‘investigation’?!]
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18784
@Peter Norton asked in above link:
”
Couldn’t it have been on all the time ?
”
https://www.express.co.uk/news/world/470172/Malaysia-Airlines-MH370-Who-cut-off-pilot-s-final-call
According to link above:
”
Another source told the newspaper that Fariq’s phone had been “detached” before take off.
The insider said: “This is usually the result of the phone being switched off. At one point, however, when the aeroplane was airborne, between waypoint Igari and the spot near Penang [just before it went missing from radar] the line was ‘reattached’.
“A ‘reattachment’ does not necessarily mean that a call was made. It can also be the result of the phone being switched on again.”
”
So at least an “insider” gave the above ‘opinion’ …
[Wonder what is the result of THE ‘investigation’?!]
@TBill
You stated “even though glide beyond Arc7 is starting to look like a definite possibility.”
On what basis, do you think a glide beyond the 7th Arc is possible?
1. The BFO data shows a steep descent.
2. The investigation of the Flaps show they were not extended.
@Richard
I hope I am wrong, but I currently view MH370 as an intentional pijacking designed for plausible deniability, including hiding the crash location.
I totally agree with you that it looks like a flaps-up, high speed glide type of ditch, that would break up the aircraft. That potentially goes a long way towards explaining the flaperon and other debris.
The steep descent is a challenge for me to explain, but I am getting better at understanding end of flight.
It could be intentional to get down to FL200 from FL400 to accomodate 1-engine flight in the Alt Config, or to get below the clouds, and/or possibly intentional to make it look like end of flight. Possibly the pilot let the autopilot go off, and descent started, and he turned on APU and autopilot back on and configured the aircraft for straight glide after the fuel exhaustion. If APU was still running at fuel exhaustion, that ought to trim up for when APU was lost. Maybe it was a mistake he accidently rebooted SDU, not sure, but he would have noticed the EICAS messages, and turned off L BUS TIE to silence any more of that.
Potentially seems to me, could have been quite a lot of remaining fuel in the left tank if say the Left IDG was off. And having taken the Beijing flight in 9M-MRO in February 2014, ZS could have known the relative fuel consumption of the engines.
@David
Figure 31 is on page 68 of your linked reference. Also on page 68 just below Figure 31 is the text below.
The aeroplane wreckage was found about 6.5 NM on the radial 019 from the last known position, slightly to the left of the planned route.
The 24 mile number in your “journalistic” reference is simply bogus.
@TBill
Your last post assumes that ZS knew about the Inmarsat data, as do all postulated scenarios involving a rapid descent followed by a long glide to “hide” the aircraft.
IMO, ZS was clueless relative to the BFO and BTO data. The most likely conclusion by far, given the data we have, is that the aircraft is close to the 7th arc.
@Andrew, Victor
I did check the other requests recorded in the SIR/Appdx 1.18A. There was no particular correlation in altitude requests for similar distant destinations (ICN, KIX) to the east, whereas the westward departures for IST and AMS were requesting lower final altitudes.
Nothing glaringly untoward.
@TBill
I agree with @DennisW, that ZS did not know about the Inmarsat satellite data.
There was no need to enter a steep dive and then pull out for a long glide, just to trick the unknown observers of the Inmarsat satellite data.
ZS already went to extraordinary lengths to hide the aircraft, proven by the fact that it has not been found so far.
@DennisW
I am not sure what ZS knew about Inmarsat, but I think he could have realized there was a logon to the network, and he might have been at least concerned that sat calls or pings could possibly give away some location info.
@TBill,
@Richard,
In the alternate electrical configuration, the SDU reboot was caused by the R engine flame-out at 00:17:30. The L engine would continue to produce thrust for roughly 8-9 minutes before it was fuel exhausted (assuming the aircraft did not crash beforehand).
Several factors enter into this figure. First, The L engine was about 2% more fuel efficient in cruise than the R engine. That means that, with no other imbalance in their loads, there would have been about 2%, or roughly 8 minutes of fuel remaining at the nominal cruise fuel flow. Two other factors affect the duration of L engine operating time after R engine fuel exhaustion. First, there was more fuel in the left tank than in the right tank at 17:07 because of the lower L engine fuel flow in the flight up to that time. This amounts to about 0.7% of the remaining fuel. Second, in the alternate electrical configuration, the L engine had no electrical load and the R engine had an increased electrical load on its IDG. A very rough guess at this effect is about 1% of extra fuel available to the L engine at 00:17. So we have three “credits” to the L engine compared to the R engine at 00:17 : (1) 2% for cruising efficiency, (2) 0.7% for initial fuel imbalance, and (3) roughly 1% for reduced electrical load. The total is 3.7%. When the R engine quit, the L engine thrust would be increased to maximum continuous thrust, causing fuel flow to be about 72% higher than the cruise fuel flow. Thus, the extra L engine run time after 00:17 is roughly 3.7% / 1.72 = 2.15% of equivalent cruise fuel flow, or about 8.6 minutes. Therefore, I would expect the L engine to run until about 00:26 unless it impacted the water before then.
The implication of this calculation is that the APU could also continue to run until 00:26 and probably a minute longer considering the fuel in the line from the L tank to the APU.
When the R engine flamed out circa 00:17:30, the aircraft must have begun descending very shortly thereafter because level flight is impossible at cruising altitude with one engine INOP. So, a descent must occur regardless of whether or not the aircraft was being piloted then. In INOP the maximum altitude for level flight at Maximum Continuous Thrust near the Zero Fuel Weight is FL270. That implies a large drop in altitude must occur in a relatively short period of time (pilot or no pilot). If controlled in an efficient manner, the descent rate can be traded for airspeed so the aircraft does not stall. Or it can enter the stall regime periodically, as in phugoidal oscillations.
The potential energy of the aircraft at cruising altitude is sufficient to allow very extended glides (> 100 NM even without considering the still-available single-engine thrust) if piloted the entire time after R engine flame-out. The ROD in this situation is nowhere near the high ROD implied by the 00:19:37 BFO. Accepting that high ROD as being valid, any gliding done after 00:19:37 would be limited to a much shorter range because the gliding would either be inefficient on the average (i.e., for phugoidal oscillations) or, even in the best piloted case, would begin at a much-reduced altitude.
The distance traveled during the descent is > 25 NM in many of the Boeing simulations, but the end point separation from the 7th arc is a fraction of that because of the high turn rates they predict. On that last point I am unconvinced that much lower turn rates are impossible. At present, we don’t understand why in the Boeing simulations for the alternate electrical configuration the right roll rate slows and a left roll begins some 160 seconds after R engine loss of thrust. It may be correct, but I have not yet heard a reasonable theory to explain this.
Here’s a new scenario: What if the pilot turned on the L IDG after 00:17:30? Even if the APU also auto-started, suppose the L IDG was powering the L Main AC Bus instead of the APU at 00:19. Then the pilot could fly at FL270 in INOP for roughly another 8 minutes and go a long distance. The only inconsistent evidence we have is the lack of IFE messages at 00:21, but that is pretty strong. I don’t know how to prevent those messages from being sent out by taking actions in the flight deck (other than shutting off the entire L Main AC Bus and thus depowering the SDU before 00:21).
@DrB: In your new scenario, what caused the increasingly high rate of descent as inferred from the two final BFO values at 00:19?
@DrB
Let me think on it. But I certainly appreciate brainstorming on end-of-flight options.
According to my sources: See also the ATSB’s “MH370 – Definition of Underwater Search Areas” of 3 December 2015: “In the case of MH370, due to the individual engine efficiency, it is likely that the right engine flamed-out first followed by the left engine. Given the amount of fuel uplifted in KL and historic fuel burn data for each engine, it is estimated that the left engine could have continued to run for up to 15 minutes after the right engine flamed-out.”
>>So 15-16 minutes remaining fuel was apparently ATSB thinking…I am not clear on why you say 8.6-minutes, but could be difference in throttle position? I would ask what is the worst case scenario? If Left IDG was off, how many minutes fuel woudl remain in the Left tank?
Of course worse case of all, more fuel was pumped over there from the Right tank, but I shall not go there.
@DrB
…PS- the Left IDG cannot be restarted, only by the ground crew, right? But the Left IDG could have been isolated electronically with Left IDG still running, or possibly the back-up gen.
@TBill
But I certainly appreciate brainstorming on end-of-flight options.
I regard such brainstorming as a waste of time at this point. The only conclusion that is actionable(1) is that the plane entered the water close to the 7th arc. Personally, if I was funding a new search (or was asked for an opinion by the funding party) I would strongly suggest +/- 10nm as a search width.
(1)By actionable I mean reasonable to be acted upon by a person of sound judgement.
@DennisW
Well I am looking to future after MH370 is not found close to Arc7, and I see one answer being the Simulator Path may have been flown, beyond Arc7. I am also hoping you are not the funding DI.
@Dennis W. “Figure 31 is on page 68 of your linked reference.”
Where I live it hasn’t yet drifted off page 65.
Your 6.5 NM was between the submerged wreckage and the last known position. The journalist’s 24 NM was that ostensibly between the submerged and floating wreckage, not that it makes much sense with the floating wreckage so dispersed.
I found that figure interesting nevertheless.
Their attempts at drift analysis were exhaustive.
http://archimer.ifremer.fr/doc/00027/13777/10915.pdf
TBill says: I am not sure what ZS knew about Inmarsat, but I think he could have realized there was a logon to the network, and he might have been at least concerned that sat calls or pings could possibly give away some location info.
I don’t think Z knew about Inmarsat pings nor Log-Ons, but not so long ago (before 2008?), you have to know where a called satphone approximatly was situated to be able to make a call to it (different prefix codes for different areas).
We don’t know what Z knew about phone calls via satellite: he could think that the plane being called implies the caller have to know grossly where he was.
So, hiding the plane implied not sink it in SIO near max range limit, and preferably after some distance under the clouds (everybody has seen aircraft in flight in clear sky on satellite pictures).
A fake flameout (including a steep descent to gain speed) followed by a motor flight (with all electrical off and just the RAT) and/or a glide would be the perfect sherry on the cake for a final gambit.
However, I think a search near the 7e arc between 25 and 20, and perhaps also at the south end beyond the already searched area, would be the best to do first, just to close this possibility at a relativly low cost.
@David
Where I live it hasn’t yet drifted off page 65.
Interesting. I reloaded your link and indeed it was on page 65 as was the 6.5nm text I copy-pasted. I swear it was on page 68 yesterday – double checked before posting. Something weird is going on with Adobe. Do you think they are involved with this somehow?
As you know, Silicon Valley is the heart of libtard culture.
@Dennis W. I have looked up libtard. Page numbers there are what they ought to be?
A relief. Not the $A drop reducing pages here as I feared.
@Victor Iannello,
The only ways I know to get 15,000 fpm ROD at 00:19:37 are for the aircraft to do it with the autopilot in secondary mode or for a pilot to put the aircraft in a steep dive.
@DrB: “Autopilot in secondary mode” has no meaning. If autopilot is engaged, then the flight control mode is normal. Perhaps you mean flight control mode as secondary?
I thought you were considering a pilot was drifting down to FL290 and flying with an engine out. In that case, there would be no reason for an increasingly steep descent.
But don’t feel obligated to explain. The scenario is already quite contorted and also doesn’t explain the missing IFE log-on.
@TBill,
Yes, I meant the left IDG isolation switch could be set to reconnect the IDG to the left Main AC Bus. I did not mean to mechanically disconnect the IDG from the engine power take-off.
I don’t know what assumptions were made to get ATSB’s 15 minute number. That would require the total (L and R Main Buses) electrical load on the right engine to be about 4% of cruise fuel flow. I was first guessing 1%. I don’t think it is as large as 4%. You should be able to get a pretty good guess at this percentage by simply taking the ratio of APU fuel flow to one engine cruise fuel flow. Running the APU when both the main engines are operating increases fuel flow by about 170 kg/hr (FL350 and 180 tonnes). The APU itself only burns 60 kg/hr to run both buses. That extra fuel consumption of 110 kg/hr is caused by the higher drag when the APU air inlet doors are open. This causes higher fuel flow to the main engines to increase thrust to compensate for the higher drag. The number we need to use here is the 60 kg/hr to generate electricity for the two main AC buses. That is 1.9 % of the typical cruise fuel flow to one engine. So I would say the 1.9% is a more accurate estimate than my first guess of 1%. Using 1.9% I figure the extra L engine run time to be about 11 minutes. It’s also possible my fuel flow estimate for maximum continuous thrust is a bit off. It might be smaller than 72% higher than cruise, and this would increase the extra run time to be higher than 11 minutes. Maybe someone had an authoritative number for maximum continuous thrust fuel flow.
On the other hand, maybe ATSB computed the extra run time using the typical cruise fuel flow instead of maximum continuous thrust fuel flow. They would be an easy mistake to make if ATSB computed the 15 minute figure from a fuel quantity given to them by Boeing.
@Victor Iannello,
I was not suggesting a pilot “drifting down” to FL270, but rather a pilot who recovers the aircraft from an initially steep descent caused by loss of the right engine thrust and subsequent reversion of the flight control system to secondary mode, such recovery occurring after significant loss of altitude, perhaps to roughly FL200.
I said up front the lack of the IFE messages circa 00:21 argues against this particular scenario.
There is a potentially a difference in the elapsed time from loss of R engine to SDU log-on, depending on whether the L AC Bus was then powered up by the APU or by the L IDG. The L IDG could in principle be supplying power to the L Main Bus within a couple of seconds of the R IDG shutdown, whereas it takes a minute for the APU to auto-start. AFAIK the Boeing alternate configuration simulations all assumed the APU auto-started.
@DrB: Sorry, but I am still unsure of what you mean. I think you are implying that in the alternate configuration, after the loss of the right engine, the pilot applied no inputs and allowed the plane to enter into a banked descent, turned on the left IDG, waited a minute, and then recovered from the steep descent by leveling the wings and holding altitude, and then flying for some distance with an engine out. That is a strange sequence that I suppose is remotely possible, but as you said, also the lack of the IFE log-on remains unexplained.
There is a class of scenarios in which the pilot is conscious at the end of the flight, allows or causes a steep descent, recovers, and then flies for some distance either with an engine out, a glide with no engines, or some combination. Once you deem this class of scenario as plausible, it becomes nearly impossible to define a width to search or a reasonable range of latitudes.
re: KL ATC’s Handoff Call Recorded Frequency
http://youtu.be/lRXYb-eO1ew
Perhaps only a small thing – but just for the record:
Was it frequency “120.9” or “120.99” in the recording above?
[almost to the end of the recording – last 10 secs]
@David
Sorry to be a bore here, but the page numbering anomaly mentioned earlier was really bugging me. Turns out that the pdf reader in Chrome (not sure about any other OS) starts numbering pages in the document starting with the first page. In your linked reference the embedded page numbers began on page 3. Duh. I was mistakenly using the page number displayed by the pdf reader, not the numbers emedded in the document.
@DrB @Victor
The question I had, if the pilot turned on the APU manually after fuel exhaustion instead of waiting 1-minute for autostart, does the APU come on immediately or is there still the 1-minute wait time?
@TBill: The APU autostart begins almost immediately after the transfer busses lose power. The one-minute delay in supplying electrical power is related to the time to spin to idle, ignite, and spin to full speed. That delay is unavoidable.
@mash,
Re…Was it frequency “120.9” or “120.99” in the recording above?
[almost to the end of the recording – last 10 secs]
No, the controller says “niner” the standard phraseology for the number nine. So all very normal.
@Victor Iannello,
You said: “There is a class of scenarios in which the pilot is conscious at the end of the flight, allows or causes a steep descent, recovers, and then flies for some distance either with an engine out, a glide with no engines, or some combination. Once you deem this class of scenario as plausible, it becomes nearly impossible to define a width to search or a reasonable range of latitudes.”
No, it’s not impossible to assign a search width. The maximum distance from the 7th Arc is equal to the maximum range in single-engine flight until L engine fuel exhaustion plus the maximum glide distance after that (starting at the maximum altitude possible at L engine fuel exhaustion). It’s a long distance, but it’s straightforward to define and to determine.
In addition, in some cases it’s not impossible to define a reasonable range or set of search latitudes. Assuming a course was set circa 19:41 using either the MCP or a single waypoint, then the most probable courses close to being fully consistent with the satellite data are already known (see my Figure 11 in my 7 March 2018 paper). The two best fits for great circles end at 35.5S and 27.1S, the one best fit for CMH is at 30.3S, the one best fit for CTH is at 31.8S, and the only CMT fit (and the best overall) is at 31.6S. Thus, in this case there are about four likely latitude zones to be searched, and these are easily prioritized.
Allowing multiple course changes is problematic in developing a search strategy because then the latitude range is primarily limited by aircraft range and endurance. So far there are no studies that might assist in prioritizing latitudes within that very large range, assuming multiple courses occurred after the FMT.
@Dennis W. I can understand it nagging at you.
The trouble is that my PDF page numbering has that at page 67/223 not the page 68 you have. Libtards still?
@David
🙂
@DrB: Yes, of course we can define a maximum search width. I was referring to the difficulty in defining a width and range of latitudes that is manageable to search and has a reasonable chance of finding the debris field. (I think you know this because we’ve discussed this here at length.) Your “likely” latitudes are not likely at all. Perhaps by some subjective ranking criteria they are arguably more likely than others, but at this point, I don’t think that really helps us.
@Victor Iannello,
You said: “Your “likely” latitudes are not likely at all. Perhaps by some subjective ranking criteria they are arguably more likely than others, but at this point, I don’t think that really helps us.”
The ranking I referred to for various latitudes is not subjective. It is objective and based on the degree of consistency with the satellite data. I would also disagree with your conclusion that such prioritization “ doesn’t really help us.” Indeed, that is the primary purpose of this discussion – to prioritize the remaining potential search areas that not too long ago (and still today in my opinion) were all considered unlikely.
That purpose also motivates my speculations about end-of-flight scenarios. Perhaps our current understanding of what could happen then is incomplete. Maybe someone can figure out how the aircraft could end up more than 25 NM from the 7th Arc. Or maybe not. Maybe someone can figure out why most of the drift models are wrong about why north of 25S doesn’t seem to match the debris arrival dates. Or maybe not. But if either of those things happen, then there would be some reason to be optimistic about a particular search area. At the moment all proposed areas have some type of contrary evidence.
@DrB
With all due respect, that is just not good enough. I don’t see any government stepping up to fund an underwater search without a compelling narrative as to why there is a high liklihood of success searching a relatively small area (say 5,000km^2). Even then, it is no slam dunk on the funding.
@Dr B, Victor. Difficulties with a new search case.
Victor you said earlier of a piloted end, “…. it becomes nearly impossible to define a width to search or a reasonable range of latitudes.”
DrB has responded. So do I but with a different outcome.
My recent conclusion on the JW site (which I will summarise here should you wish) was that none of the most recent simulations, Normal or Alternative was consistent with the final BFOs for various reasons including their timing. If that be so, the only reasonable conclusion to be drawn, other than these simulations being unrepresentative of unpiloted, is that a pilot caused that descent.
Another reason for some doubt about the investigators’ current assumption that there was no pilot, is that the analysis based on that, complemented by drift analysis, has resulted in an unsuccessful search to date.
Supposing there was a pilot, a persuasive case for a new search would need to assume that:
• The pilot did not alter course enroute to the south since otherwise this would be a multiplier of possible end latitudes.
• He did not extend the aircraft’s range by step climbing (unless future analysis includes a further-south crash possibility, despite the improbability of that drawn from current drift analysis.)
Supposing also that he used engine power and/or glided beyond the 7th arc log-on then a constraining search arc based on an assumed course could well be needed to keep search area to cost/area effective.
Moreover, if as per that engine power the Alternative configuration were applied to a piloted aircraft as Dr B has done, to have any effect a manual start of the APU would need to precede the auto-start at right engine run-down yet be no more than a minute before that run-down. Otherwise there would be no log-on due to the SDU having been de- and re-powered, unless some other reason becomes apparent as to why the SDU became inoperative then operative. (Mind, were another identified or the BFO interpretations found flawed, that could undo the underpinning of all end-of-flight analysis to date, including here).
(One other reservation as to left engine running during a steep descent is whether it would retain fuel suction with a small quantity in its tank).
Likewise, it would have to be assumed that the pilot did not switch off the APU until at least two minutes after its start commenced but did so shortly after that (no IFE connection) or else the crash intervened.
Within the above constraints there would appear to be no appreciable difference in outcome between manual APU start/stop intervention and an auto-start, except that a manual stop a little over two minutes after an auto-start would explain the IFE non-connection.
Similar assumptions about the APU would apply to the Normal configuration after auto-start except that APU fuel starvation might account for its failure before IFE connection.
With the unpiloted assumption put aside all these points and assumptions weaken the case for a future search but IMO none more than the glide possibility. However that can be sidelined by one plausible extra assumption, which is that the pilot pitched the aircraft nose down with an obvious purpose; and for which there are precedents. The plausibility of that does depend on a scenario that the pilot was suicidal at that point.
Finally, as to the place of drift modelling generally supporting one crash site’s latitude probability over another, Richard Godfrey’s work implicitly points out that in principle there is a similar variation of probability with longitude. A flight of say 120 NM due east, a little over 2˚ in longitude, would have an effect similar in character to the equivalent north/south crash site change of 2˚. In short, if a piloted end without a continuing steep descent were assumed, the outcome of much previous drift analysis would become inapplicable to search latitude probabilities.
Indeed this raises questions as to what the current search width’s effect should have on current drift analysis conclusions. To my knowledge the CSIRO and WA University work do not address longitude as a discrete variable.
A summary of the above is that:
• It is realistic to assume there was a pilot and that he crashed the aircraft close to the 7th arc, though that does depend on some assumptions.
• If piloted powered flight and a glide are assumed there is not just a problem with search areas but with the supplementary drift analysis that might be advisable to support them.
• I will ask Dr David Griffin about the longitude effect of the past 25 to 40 NM search width on latitude based beaching probabilities in case that should be of further interest though I think he is overseas at the moment.
@Dennis W. That is what we have to work with IMO. As it stands, it does not look promising.
@DrB said: The ranking I referred to for various latitudes is not subjective. It is objective and based on the degree of consistency with the satellite data.
Of course it’s subjective. It’s based on assumptions and an objective function that you defined.
My point is that there is little use in ranking end points (by whatever criteria) that each have low probability. Until evidence or insights surface that increases our confidence in one of those end points, the search effort is dead. I think you agree.
David said: As it stands, it does not look promising.
I won’t quibble with some of the details of your comment. I do agree with your last sentence.
@Victor
I do agree with your last sentence.
Me also, however I view not finding wreckage dissappointing, but not a huge loss. I think the CVR will contain no information, and the FDR will show that the plane was flown to where it was found (not an aircraft failure). Of course, details of the flight path, especially between IGARI and the FMT, will be of interest to the analysts here.
What I don’t think the wreckage will provide is the answer to why the plane was diverted. Since I am not a fan of the suicide motive, I am sure that answer lies elsewhere and may eventually surface.
@DrB
I feel your path-definition approach has merit but two or three things:
(1) I feel you may need to incorporate the 23:14 sat call into the BFO fittings. Possibly you could have as-measured and adjusted value (options) for that.
(2) Also consider selected active pilot paths. I am not sure how to do that, but one path I would assign priority is Flight Sim ending point (Nederland’s 2017 path which I now agree with)
(3) I am confused why the orig IG path to 37.5 South does not rank better. The match to BFO thru 2241 was perfect. Maybe that is due to not starting FMT at IGOGU but then that would just mean the starting point alters the results, which would suggest again the need for optional FMT starting points.
@TBill,
To better than about 5-10 Hz, the 23:14 phone call BFOs don’t add any new information compared to the combination of the 22:41 and the 00:11 BFOs. If you want to do better than that, then you have to have some means to calibrate the relative channel frequency offset for the channel used for the two phone calls, for which we have no other data as far as I know. I chose to use the 23:14 BFOs to find the relative channel offset (by interpolation between 22:41 and 00:11), which I then applied to the 18:40 BFOs since those are not as predictable.
Modeling “active” pilot paths, by which I think you mean those with late course changes, is fraught with difficulty. I do think that using some “known” positions from either the flight simulations or existing waypoints/airports may be useful, as several people have already done.
My analyses of post-19:41 routes make no assumptions about starting points. The satellite data can be fit very well near 37.5-39S with a constant ground speed and pretty well at a constant TAS, but neither of those can be flown in a B777.
@DennisW,
I agree that no government would fund a search at the present, given the low probability of success, but there are several other possibilities for funding that might accept higher risk.
You said: “What I don’t think the wreckage will provide is the answer to why the plane was diverted.” I agree that is likely to be the case. In my opinion the FDR was shut off at about 19:00, so it may actually tell us very little. We probably won’t learn anything new from the wreckage, and, even if some body parts are recovered, they may not shed any light on how or when that person died.
@Victor Iannello,
My objective Function (a figure of merit) is simply the combination of BTO and BFO errors. It is objective, not subjective.
I agree that all assumptions are subjective in nature, and all efforts to date to predict probability of end point suffer as a result of the unknowability of those subjective assumptions.
You said: “My point is that there is little use in ranking end points (by whatever criteria) that each have low probability. Until evidence or insights surface that increases our confidence in one of those end points, the search effort is dead. I think you agree.”
Not exactly. I would agree the likelihood of a government-funded search now is small. Crowd funding or entrepreneur funding might be possible.
In my opinion, what is needed is a solid reason for why the previous searches failed. That reason does not necessarily have to discern relative probabilities among the remaining possible search areas. There are other ways to do that, such as flight path modeling and drift modeling. If the newly identified reason for the previous search failures was also discerning among the remaining search areas, so much the better, but that is a big ask.
@DrB: It is the mathematical objective function combined with assumptions (such as automated flight over a time period) that makes the criteria subjective. That makes the ranking of end points subjective, which is my point. I’ve seen a lot of solutions based on solid math and physics (with unprovable assumptions) that are both precise and wrong.
Let’s be realistic. Crowd-funded searches will never raise the millions of dollars that are needed, an entrepreneur needs a large reward to get a return for the venture risk, and altruistic searchers like Paul Allen only search if there is a defined radius from a last known point so the costs are reasonable and there is some hope of success. If the probability of success is low, nobody will search. Period. And by most counts, the risk of failure is higher this time than the last OI search.
Perhaps we’ll learn more from the drift models of Richard G. and David G. That might be our last hope.
Something ‘unintelligible’ written in August, but perhaps relevant for the recent “MH370 ‘last minute’ pilot swap/switch” discussion:
[last edited 28/08/2018]
re: Hidden Mission Purpose
If the secondary mission purpose is to hide the primary mission purpose, then perhaps the best tertiary mission purpose is to find a scapegoat. In that case Capt Zaharie would be the best candidate. Have you noticed, perhaps not by chance alone, that he met so perfectly all the motivational and experiential requirements for this ‘job’? (diverse reasons – political, marital, judicial, psychological, intellectual, real and simulated flight record, etc.)
[plus coincidental ‘inexperienced’ copilot ‘cooperation’.]
@mash
So what is the primary mission?
@Victor Iannello,
You said: “Perhaps we’ll learn more from the drift models of Richard G. and David G. That might be our last hope.”
If a consensus magically appeared that the most likely latitude was north of 25S based on the drift models, then a search there close to the 7th Arc would be indicated. In that case we get “the big ask” because now we understand why the previous searches failed, and we have a new area indicated.
However, CSIRO has held steadfast that those latitudes are fairly strongly excluded. A consensus on the northern area would require CSIRO to change their opinion or, at the very least, to admit that other models consistent with northern latitudes could be correct.
We also need to make it clear that interpreting drift model predictions requires a subjective assumption of the average time difference between arrival and discovery. If you assume that is as large as 5-6 months, then Richard Godfrey’s 22 S predictions can match the non-flaperon discovery dates. If you assume a 1-2 month arrival-to-discovery interval, then nothing north of 28 S works.
So every prediction method, including drift modeling, requires making subjective assumptions. In the case of the drift modeling, the uncertainty in actual arrival dates, even given precise discovery dates, may be large enough that a clear-cut case for being N or S of 25 S may not be possible.
@DrB
You stated “If you assume a 1-2 month arrival-to-discovery interval, then nothing north of 28 S works.”
I disagree.
David Griffin’s results show that the curve for 50% debris items reaching Africa aligns with 24S (August update Fig. 3).
My results for 22S show a 55 day arrival prior to discovery.
@DennisW
It’s a secret, of course (almost by definition). Say something unconventional or revolutionary, something ‘valuable’, something unexpectable (hence a scapegoat provided) …
@Richard Godfrey,
You said: “David Griffin’s results show that the curve for 50% debris items reaching Africa aligns with 24S (August update Fig. 3).
My results for 22S show a 55 day arrival prior to discovery.”
I will say (again) that that comparison you are making is a median prediction date to the earliest discovery date of one item. Comparing apples to apples, the difference in median prediction date to median discovery date is larger, about 4-5 months. I do not call that a good alignment.
Victor Iannello says: …/… Perhaps we’ll learn more from the drift models of Richard G. and David G. That might be our last hope.
So, you don’t think that radar data which may exist (Military? Thaï? Singapour? Indo? … ?) or data from police inquiry, could bring anything new?
Or your statement is for a new search with only already known data?
@DrB
We are talking at cross purposes. I was referring to your statement about reaching consensus on the drift analysis.
You are referring to the difference between the median discovery data and median arrival date. On your point all I can say is there is a wide range in the temporal dispersion and points up to 20°S are within the range of possibilities.
@All,
I have a new, simple FMT model which uses BEDAX and BEBIM to match the 18:40 BFOs and the 181.2 degree CMT Route.
It is available HERE.
@DrB
OK good so that makes three of us going from BEBIM (you Nederland and me)… I wonder if that waypoint would be in MAS data base? I believe I had to add it manually to the PSS777 sim model that Z might have used. But Z could have used a 3rd party program for the flight sim path.
I like the fact you fit the BEDAX to BEBIM as I was looking at that too.
So Nederland and I feel a slow down and heading change at BEBIM to approx. 170 deg or towards NZPG. But we are converging, sort of…
@DrB: Thank you for the new paper. I think it is an improvement over the last one.
You said that you think it may be possible for the plane to fly a relatively straight path while experiencing phugoid oscillations. I think it is very unlikely the descent rates inferred by the final BFO values could have occurred with level wings and no pilot inputs. With no pilot inputs, the measured descent rates are consistent with an impact soon after. That means that either there were pilot inputs, or the plane crossed the 7th arc north of 25S latitude.
@TBill
“Consensus” is a relatively new term when applied to scientific/technical endeavors. It has found favor in the theories of anthropogenic global warming, the big bang, and evolution. All of which are wrong, IMO.
@DennisW,
You make a good point. The “consensus” manufactured by the media for anthropogenic global warming is vastly overblown, as a political tactic. There is some correlation of recent global temperature data with CO2, but that doesn’t make CO2 the cause. Much larger temperature effects have occurred in the not-too-distant past (but long before the CO2 rise), and all of them are IMO most probably explainable by changes in solar activity. AFAIK, all of the projections made by AGW proponents have turned out to be wrong. Duncan Steel has also written a book on why the Earth’s precession causes an average solar insolation variation which is larger than every other term used in the AGW models. BTW, I studied planetary temperatures and atmospheres for my PhD dissertation. Some planetary scientists chose to study AGW simply in order to get easy research funding from their current government.
The other two theories you mentioned are mostly apolitical, and serve a scientific purpose. They may be wrong, but they are the best we’ve got at the present. Undoubtedly all three will be turned upside down by “future enlightenment”. The same may occur for MH370. We may all be surprised when some day new information appears which turns our current MH370 theories upside down.
@Victor
You stated “That means that either there were pilot inputs, or the plane crossed the 7th arc north of 25S latitude.”
That is a profound summary.
@Richard Godfrey,
Or both.
Or (very unlikely) something happened to the 00:19 BFOs which we don’t understand at present.
@All,
A new search within 25 NM of the 7th Arc north of 25 S may fail, or it may succeed. We won’t know until it is done, or until we get new strong evidence for a different location.
There are two arguments for possible failure:
(1) Based on the drift models, the delay between median debris arrival date and median discovery date must be assumed to be 5-8 months for locations from 25 S – 22 S. That’s a very long time difference. More southerly latitudes require much shorter delays (1-5 months), and in my view they seem more likely.
(2) The assumption that there were no pilot inputs after 00:19 may turn out to be wrong, and the aircraft may be > 25 NM away from the 7th Arc, including north of 25 S.
@Victor Iannello,
I agree that the likelihood of piloted flight after 00:19 is significantly greater than the likelihood that the aircraft flew a relatively straight phugoid on its own. That’s why I listed the pilot inputs first, then, in order of decreasing likelihood, the straight phugoid and lastly the “skewed” BFOs.
“Or (very unlikely) something happened to the 00:19 BFOs which we don’t understand at present….”
+1 from me. It seems to me much simpler and more likely for the BFO to be wrong and a multitude of prior assumptions to be correct than vice versa. Avoiding the “beyond 7th arc” conclusion simply on grounds of search economy is a matter of convenience and feasibility, not solution probability.
@DennisW
Did I say “consensus” anywhere re: MH370?
A few of us are converging on BEBIM, and I would like to suggest consideration of active paths from BEBIM as a token active path for more detailed study.
@TBill
No. It was implied. I took the opportunity to pontificate. Sorry.
We have all seen this movie a few times.
@Someone who knows the answer
One speculative question:
What would happen if the MH370 started programmable VNAV idle power descent before the left engine stopped?
It connects with other questions coming back to me persistently:
What was the purpose of ZS simulator exercise?
Could it be to determine the appropriate TOD (to reach 4000ft) at the border of the range?
If it’s too stupid, please forget it.
@Greg
It is a good question, if the simulator work gives any clues about end-of-fight intent, both re: (1) flight path and (2) maneuvers or glide. I was going to another look at the sim end points to see if I can see anything from my viewpoint. I am not familiar with the approach you mention…I was thinking find a way to get autopilot on when the left engine cuts out and then trim up the aircraft in the no engine mode at about 1100 ft/min decent (and then leave the cockpit).
@TBill,
BEBIM was a standard aviation waypoint in 2014 and remains so today. That means it should have been available in 9M-MRO’s database.
@Greg, @TBill,
The data showing a route to fuel exhaustion in the SIO on the ZS home simulator was last modified on 3rd February 2014. The next day, on 4th February 2014 ZS flew KUL-JED-KUL with a B777 (not 9M-MRO) with a similar fuel load as shown on the simulator. Was this flight the first dry run for ZS’s plan? Was the fuel usage, endurance and range noted?
One of the drives on ZS home simulator MK22 was formatted on 20th February 2014 at 03:37. The data on ZS home simulator was deleted from drive MK23 on 20th February 2014 at 11:30. The simulator on ZS home simulator was uninstalled from drive MK25 on 20th February 2014. The next day, on 21st February 2014 ZS flew KUL-PEK-KUL with 9M-MRO. This was his last flight with 9M-MRO prior to 7th March 2014. Was this flight the second dry run for ZS’s plan but this time with the target aircraft 9M-MRO? Was the fuel usage of the left and right engines noted on 9M-MRO and the endurance and range to fuel exhaustion estimated?
On 26th/28th February 2014 ZS flew KUL-MEL-KUL and visits his daughter.
On 7th March 2014 ZS disappears with 9M-MRO to the SIO, having a well prepared plan, having said goodbye to his daughter and thinking all the home simulator data has been irrecoverably deleted.
@Richard
“…thinking all the home simulator data has been irrecoverably deleted.”
Yes that might be our only hope, that we actually do have an important clue. A clue that says, perhaps the simulator flight path is as important as Arc7 as a search basis operating line. Previously I have given total search priority to Arc7.
@TBill
To my knowledge, this is the first time the RMP report flight schedule for ZS has been compared to the RMP report flight schedule for 9M-MRO.
re: Logon Characterisation – Unintended vs Intended
What if the 1st (difficult-to-explain) logon was unintended and the 2nd (easy-to-explain) logon was intended. That is, the first was an unintended leak (a true lead) and the second was an intended leak (a false lead).
Wonder whether this ‘configuration’ could resolve the consistency/controversy problems so far encountered.
For example, the simulated and Inmarsat data (in)consistency, and the fast dive/slow glide controversy …
@Richard
Yes Z’s prior flights in 9M-MRO suggest he could have known aircraft specifics such as fuel comsumptions differences in the engines. As some have pointed out, we might expect some trying out of rogue flight ideas on prior flights as preparation. Waypoints in the flight computer, etc.
@mash
I am confused which events you are suggesting as accidental, but I agree that is a possibility. Personally I feel the 18:25 logon was intended if this was a deliberate diversion, to send the message to Razak that this aircraft did not crash in the South China Sea. The final logon I am not sure, I can go with intentional or accidental at the moment.
@Richard Godfrey
Richard,
I don’t think that Captain Zaharie did fly 9M-MRO on his MH370/MH371 duty on 20-22 February 2014; I think you’re out by a day.
The Captain commenced duty at 2320 MYT on 20 February for MH370 departing at 0029 MYT on 21 February. 9M-MRO was tasked as MH370 departing at 0035 MYT 22 February (140221 KUL1635 [UTC]). At the time that the Captain was departing KL on MH370 on 21 February, 9M-MRO was returning to KL from Narita as MH093.
I think that you’re also out by a day on the flight simulator data points. I was informed by the ATSB that,
‘a listing of 4 complete .flt files and 2 partial .flt files were provided to the ATSB on 19 April 2014. All files except the final (partial) file contained a [DateTimeSeason] field which contains values for Day, Year, Hours, Minutes, Seconds, Season.
The files with the field had the common values of Day=33, Year=2014 and Season=Winter.
The [DateTimeSeason] field in the files indicated 15:26 to 16:38 (hh:mm).‘
Day=33, Year=2014 is Sunday, 2 February 2014.
A word of caution when interpreting the [DateTimeSeason] data; it is almost certainly the date/time for the simulation flight (ie it is’ sim’ time). There is no guarantee that that is the same as the actual date/time those data files were recorded. While there is generally a correlation between the [DateTimeSeason] value and the actual date/time when a flight file is first created, unless the flight simulation is run continuously in real time, that correlation is quickly lost.
A further issue with the .flt data provided to the ATSB is that we can’t be sure that the first data point (15:26) represents the start of the simulation. As a matter of fact, on the assumption that the six data points provided to the ATSB were AG-References 6, 5, 4, 3, 2 and 1 from the RMP Report (respectively 2N, 45S2, 45S1, 10N, 5N and 3N per Victor and Yves’ paper), given that there appear to be at least two additional data points in RMP report (AG-References 6-1 and 7) that probably precede them, I’d say that there is no certainty that the 15:26 data was actually created at 15:26 on 2 February 2014.
@Mick
The date-times in the RMP folder for the ZS flight schedule and the 9M-MRO schedule are all stated in UTC.
@Mick Gilbert: The ATSB said to you: a listing of 4 complete .flt files and 2 partial .flt files were provided to the ATSB on 19 April 2014
Unfortunately, the flight files provided to the ATSB were not complete. If they were, we’d know more about how the FMC was programmed.
I supplied the ATSB with a complete flight file (not one from the captain’s hard drive) that they could use as a template, and they acknowledged that the flight files they have are missing information.
This has been one of my frustrations. There could be more information that is available the might not have been extracted from the hard drive because of a misunderstanding about what constitutes a complete flight file for the FS9 PSS 777 model. It’s also possible that the information is available but has been withheld.
@Victor
I share the frustration yes it seems like there could be more data, inlcuding things like PSS777 waypoint database and addon programs etc. Let’s ask FBI.
@TBill: If you have a contact within the FBI, go right ahead. I don’t.
@Richard Godfrey
RE: ”The date-times in the RMP folder for the ZS flight schedule and the 9M-MRO schedule are all stated in UTC.”
The date-time stamps in the aircraft schedule are UTC, but those in the pilot schedules appear to be local time. For example, ZS’s schedule on the night of the ill-fated MH370 shows that his rostered duty start time was 2320 on 7 Mar 2014, for a departure at 0027 on 8 Mar. Those times are definitely local. I assume we’re looking at the same documents?
BEBIM is definitely the only waypoint that far south that is also a match. Unless customs waypoints were used, I’d say BEBIM deserves all consideration.
Sorry, I have been absent from this site for some days (including a vacation) and just now saw that my earlier suggestion BEBIM – Antarctica has been discussed here.
@Nederland
Yes I have latched onto the BEBIM idea, I agree with you in hindsight. I could not recall what your final path was, NZPG or WYKS, but I like your NZPG version right now.
@TBill
Assuming the pilot is a scapegoat, then the final logon Imarsat data is a necessity – to ‘prove’ the guilty party. But another necessity is to hide the (contradicting) evidence – the plane. Therefore, showing the other ping rings (except the last one) is counter productive. Because showing the last ping ring is anticipated/unavoidable, it is perhaps necessary to compensate for the weakness by including:
a. A fake fast dive.
b. A real slow glide.
c. A ‘skew’ landing location (not the far south/remotest part of SIO).
[(un)fortunately this plan is not completely successful due to the unintended/accidental leaks (such as the ‘full’ set of Imarsat data).]
Considering CTT paths:
I’ve been looking at CTT paths, and specifically trying to minimize path curvature (between 20:41 and 00:19) by varying the 19:41 latitude (in steps of 0.01 degree) as well as the bias frequency (in steps of 1 Hz). The latter to account for a possible shift after the 18:25 SDU restart and/or oscillator drift. The most straight path I’ve found so far is for a bias frequency of 153 Hz, see the link for a quick impression. This result could suggest a path after 19:41 which initially goes straight south, followed by a turn just north of the equator (near ISBIX?) towards 177 degrees TT. Looking at the speed profile it suggests that a stable cruise condition was reached only around 21:00 UTC (at a latitude around 6 – 7 degrees south).
The 00:19 latitude indicated is 32.8 S.
I’m currently checking the speed profile; does it correspond to a common “schedule” assuming for example level flight, with decreasing weight due to fuel burn?
https://www.dropbox.com/s/f6gvmfacptvxozo/TTpath_FFB153Hz.pdf?dl=0
@TBill
I think both BEBIM – NZPG and BEBIM – YWKS work. The most recent version of my paper had BEBIM – YWKS, but this could change if you take the wind into account:
https://www.docdroid.net/GvlrLaV/mh370-waypoint-30.pdf
Earlier versions had BEBIM – NZPG
The point is that probably neither NZPG nor YWKS were part of the flight computer database, but BEBIM surely was. The end point was therefore probably a customs waypoint perhaps to do with the findings on the simulator.
I think it’s still entirely possible that BEBIM was the penultimate waypoint since the search width was already reduced to 20 nm in that area, but it now seems quite possible that there were pilot inputs at the end of flight, perhaps because the pilot saw on their display that SATCOM reconnected and they decided to not crash immmediately, but glide a bit further instead.
@Mick, @Andrew
Many thanks for your comments.
I stand corrected, ZS only flew with 9M-MRO on 22 Feb 2014 from PEK to KUL. The KUL to PEK flight he made on Friday 21 February 2014 scheduled departure 00:35 (Local Time) was with another aircraft.
I have compared the Malaysia Airlines Timetable valid in February 2014 with scheduled departure and arrival in Local Time, with the 9M-MRO schedule from the RMP report in scheduled departure time in UTC, with the Flight Schedule for ZS in the RMP report in scheduled departure time in Local Time (by deduction but this was not specified in the report), with the Fleet Management Schedule for ZS in the RMP report with scheduled and actual times for duty start, departure time, arrival time and duty end in Local Time.
The route PEK-KUL scheduled on Saturday 22 February 2014 08:55 (Local Time) or Saturday 22 February 2014 00:55 (UTC) was 9M-MRO. ZS’s Duty Start was Saturday 22 February 2014 07:55 (Local Time) and the actual departure was Saturday 22 February 2014 08:54 (Local Time).
@Victor Iannello
Victor, in your correspondence with the ATSB regarding the simulator data did they ever reveal what data elements were represented in the files that they were given? The fact that they had [DateTimeSeason] data for five of the six files means that they had more data than what was presented in the RMP files.
@Mick Gilbert: Relative to the data presented in the RMP report, I believe the ATSB was given the “top” of the file (including [DateTimeSeason]) but not the “bottom” of the file, which would include FMC parameters. I suspect whoever extracted the data from the Shadow Volume did not understand what constitutes a complete FS9 PSS 777 flight file, as there are differences from the FSX PMDG 777 flight file.
@Nederland
You and I are in near complete agreement, as I suspected. My latest write-up is above (post of September 24, 2018 at 6:18 pm).
@Niels,
You said: “This result could suggest a path after 19:41 which initially goes straight south, followed by a turn just north of the equator (near ISBIX?) towards 177 degrees TT. Looking at the speed profile it suggests that a stable cruise condition was reached only around 21:00 UTC (at a latitude around 6 – 7 degrees south).
The 00:19 latitude indicated is 32.8 S.
I’m currently checking the speed profile; does it correspond to a common “schedule” assuming for example level flight, with decreasing weight due to fuel burn?”
What you suggest is similar to the 181.2 degree CMT route I proposed last March (ending at 31.6 S) and available HERE.
Figure 16 in that paper shows the MRC speed profile with time/weight. There is a monotonic decrease in speed when using ECON as the aircraft gets lighter, assuming there are no altitude or large temperature changes.
The predicted true tracks at the handshake times are listed in Table 3, and Table 4 shows the average true tracks for the legs between the handshake times. The track curves eastward because of the varying magnetic declination. At 19:41 the true track is 179.9 degrees at 2.9 N latitude. At 20:41 the true track is 178.8 degrees at 4.0 S latitude. At 21:41 the true track is 176.6 degrees at 12.5 S latitude.
I don’t think a Constant True Track is likely to have been used because it requires setting the NORM/TRUE switch to TRUE, and this is usually only done in the polar regions. There would be no reason to do that near the equator.
@DrB
Did you consider to derive your figure 11 for a range of different FFBs? For example I’m very curious what fig.11a would look like for a bias frequency of 153 Hz (assuming that you used a value close to 150 Hz).
I would like to add that of course I’m not certain CTT was the mode used. However, I am a bit surprised by the straightness of the path I can obtain from such strongly curved input functions. I did not “tweak” the input functions other than shifting the D(t) function up and down.
@Niels: How did you determine the position at 19:41? I think you’ll find that shifting the FFB changes the position at 19:41 that produces a straight path.
@VictorI
Yes, for every FFB I choose the starting location such that the RMS(kappa) or the var(track) over the 20:41 to 00:19 interval is minimal (and of course the location has to fit the r(t) curve). I noted down the “optimal” value of RMS(kappa) for every FFB and if you then plot it as a function of FFB this graph has a minimum around 153 Hz. This is the value for which I posted the path.
[Kappa is calculated for every 10s time step as delta(track) / delta(s), with s the path length]
To implement WGS84 compatibility I designed a control algorithm which for every time step modifies track and GS such that the r(t) and D(t) functions are followed as closely as possible. It requires a P control for the GS (using the D(t) error signal) and a PD control for the track (using the r(t) error signal). The “damping” is needed to avoid numerical oscillations in the resulting track.
@Dennis
Just to breakdown the discussion, natural questions are best resolved through the application of scientific principles and formulas. Human beings have free will and at the local level make unpredictable choices.
We run in to trouble when we try to use scientific approaches to resolve social questions. An apple dropping from the tree will always fall toward the ground (or fall toward each other when the mass of the 2 objects is taken into consideration.)
Social questions are best resolved through consensus. The wisdom of crowds summarizes research that shows that a group is smarter than a single individual.
I think we need to consider 2 different approaches to the scientific and social questions posed by mh 370.
@Niels,
1. No, I have not repeated the calculations in my Figure 11 path fitting optimization for different BFO bias values. I used 150.27 Hz. One set of results doing that optimization requires over 300 hours of computation on the fastest laptop I could buy. I’m not doing it again.
2. I don’t think path straightness is necessarily the best or even a good figure of merit, since CTT is quite unlikely to have been used.
3. However, here is an interesting fact for you. The BFO bias that results in zero mean BFO error for my best-fit 181.2 degree CMT Route is 152.7 Hz. I’m not sure at the moment what this means, but perhaps the close agreement of your 153 Hz and my 152.7 Hz is not a coincidence.
@Niels,
I don’t understand your “Residual Doppler D(t)” values. They don’t seem to match simply subtracting the BFO bias from the measured BFOs. How do you calculate D(t)?
Unfortunately the BFOs depend on location in a complex fashion (plus a small time dependence) because of the partial Doppler compensation applied in the AES. That means that at a given location (and time), the BFO depends linearly on the aircraft speed, but the proportionality constant depends on the assumed location (and assumed time).
I once tried to do something similar to what you are doing, but in the end I concluded that it is not fundamentally different from what I was doing already – assuming a route “shape” (not just a straight path) and minimizing BTO and BFO residuals.
@DrB
1. Yes I understand
2. Difficult to know what is likely or not regarding MH370 flight modes without understanding cause/motive/circumstances
3. Interesting, but could be coincidence. I’m now looking more in detail and the “optimal” CTT path seems to be connected to FFB = 152.9 Hz.
In the bigger picture it seems we are both zooming in on the 31 – 33 S latitude range. Personally I would still want to learn more about the OI scanning “performance” in this area.
One basic question I have is the height above the seafloor the AUVs were operated. In combination with swath width it could give some feel for “shadowing” risks in rough terrain.
@DrB
I’m sorry I didn’t see your last post when hitting the submit buttong
D = BFO – (dFdown + dfsat + dfAFC + dfbias)
Basen on the values tabulated by Ashton et al.
So D represents df_up + f_comp
@Niels
I use the same metric. I like it because residual Doppler represents the Doppler contribution from aircraft motion and location required to satisfy the measured BFO values.
@Niels,
Thanks for that clarification. Your method is mathematically equivalent to fitting a trajectory to latitude/longitude points at each handshake time to match the measured BFOs and BTOs. That simplified fit is an approximation that ignores other information we have.
It ignores the effects of pressure altitude, air temperature, and winds aloft on the air and ground speeds and on the aircraft track. It also is not constrained to follow an air speed profile which is flyable in a B777-200ER. It ignores the constraint on fuel consumption to match the estimated MEFE time. It ignores the constraint to be consistent with the limited number of lateral navigation methods available in a B777-200ER.
Please don’t get me wrong, as this is not intended as a criticism of your method. I am only pointing out that your method is not fundamentally different from what I and others have done, except that is an approximation that ignores numerous additional constraints that can be applied to improve the accuracy of the result.
@DrB
I’m fully aware that my method produces a path and a GS profile (consistent with BTO and BFO data) which you then afterwards have to check against other constraints. This is work in progress as I stated when I posted the CTT path a few days ago. This may be a disadvantage of the method; the advantage is that you quickly have a path consistent with the sat data and that it is not too difficult to optimize towards CTT or CMT paths.
I’m not sure about “Your method is mathematically equivalent to fitting a trajectory to latitude/longitude points at each handshake time to match the measured BFOs and BTOs”: I’m first creating “smooth” D(t) and r(t) function and use these to create the path.
I agree that fundamentally it is not different: it is a different way of “solving” based on the same equations and data. Actually I liked the previous version of my path generator more as it was based on analytical solution for the velocity vector v(t) based on D(t) and dr (t) / dt. However, it was slightly inaccurate as the analytical solution could only be derived for a spherical earth model.
@Niels
I have pretty much the same opinion as you do. The ISAT data is all we really have, and it is pretty much a “turd” analytically. You can polish a turd all you want, and it is still a turd.
@Richard Godfrey. Having looked into drift analysis and now your work I hope you will welcome some comment on both, and that the below, some of it informatory, is of interest.
https://www.dropbox.com/s/xy632c7f1vh30iu/Post%20for%20Richard%20Godfrey.docx?dl=0
@David
Please bear with me, as I am working at full speed to publish an update to my drift analysis.
I have made 2 major corrections:
1. I am very grateful to Victor for proposing using a cluster of 9 particles from each start latitude. This solves a problem resulting from relying on just 1 particle from each start latitude, which is statistically questionable and has proven to be misleading. David Griffin uses 120 particles. My results are now much closer to David Griffin’s. I plan to increase from 9 to 25 particles in future.
2. I am also grateful to David Griffin for pointing out that the drifter data prior to the year 2000 is not accurate. I have removed the old data. That leaves a database of 280 undrogued drifters from the year 2000 to the year 2018, which provides an adequate coverage of the SIO.
I think you will find the results very interesting. I will publish them later this week.
@Victor, @DrB, @David, @Niels, @TBill, @Nederland
I have updated my drift analysis as follows:
Method Update
1. I am grateful to Victor for proposing I should use a cluster of 9 points from each start latitude, it has helped correct earlier errors and statistical anomalies based on using a single particle at each start latitude.
2. I am grateful to David Griffin for pointing out that the drifter data prior to the year 2000 is not reliable and I removed all such data from my database leaving data from 280 undrogued drifters with a good coverage of the SIO.
3. I have updated David Griffin’s chart in order to correct an oversight. He had shown the maximum beaching date in Africa as 26th August 2016 (Vertical Stabiliser found by Jean Viljoen at Linga Linga, Mozambique 23.739314°S 35.400766°E). I have corrected that to show the maximum beaching date in Africa as 27th January 2017 (Right Wing No. 7 Flap Support Fairing found at Mpame Beach, South Africa 32.097481°S 29.062852°E).
4. These simulations are based on the historical undrogued drifter data. There are no adjustments for leeway and drift angle.
Results
1. The best fit to the Flaperon arriving on the North-South line through St. Andre, Reunion and within 200 km from St. Andre, Reunion, just prior to 29th July 2015 (after 508 days from the 7th Arc) is from a start latitude at 31.1°S ± 1.0° (please see link below).
https://www.dropbox.com/s/ow23g0lbw386uia/Reunion%20Arrival%20Time%20-%20Days%20vs%20Start%20Latitude.pdf?dl=0
2. The best fit to the median discovery time of the 13 MH370 Floating Debris items reaching mainland Africa is from a start latitude at 31.1°S ± 2.0° (please see link below).
https://www.dropbox.com/s/d442ejr3cuuvu1g/Africa%20Mainland%20Arrival%20Time%20-%20Days%20vs%20Start%20Latitude.pdf?dl=0
3. Simulated paths from the cluster centred on 32°S show the required bifurcation around Madagascar. The only other start latitude which shows bifurcation around Madagascar is centred on 25°S.
4. Broken Ridge creates a large number of eddies and gyres and start latitudes further south than 32.6°S generally take much longer to get through the Broken Ridge area to leave the 7th Arc for Rodrigues, Mauritius, Reunion, Madagascar and Africa.
5. There is a better fit to David Griffin’s Fig.3 showing beaching dates in Africa (please see link below).
https://www.dropbox.com/s/47c35xgex2jgd3c/CSIRO%20vs%20RG%20Comparison%20Latitude%20vs%20Time%201%C2%B0%20Step.pdf?dl=0
6. There is a good fit to a MH370 end point around 31.0°S as proposed by @DrB, @TBill, @Nederland and others.
Next Steps
1. I plan to run a cluster of 25 particles centred on 31.6°S (as proposed by @DrB) and at ±30 nm as well as at ±60 nm, north, south, east and west of the centre point including diagonals as shown in the following link:
https://www.dropbox.com/s/i5l41nen6q3fxdc/Array%2031.6S.png?dl=0
2. Why has MH370 been missed at around 31°S? If there was some pilot input to recover from the steep descent at around 15,000 fpm to a glide descent at around 2,500 fpm (glide ratio 17:1), then I estimate the maximum range from the 7th Arc is 60 nm eastwards of the 7th Arc.
@Richard G: Thank you for your continued work on the drift analysis. It seems your work is beginning to better match the analyses from David G. and Chari P.
If the aircraft glided some distance from the arc, the assumption of drift from an impact point along the arc becomes less tenable. Have you considered any impact points 60 – 100 NM from the arc?
@Victor
The short answer to your question is no.
The long answer is, I have looked again at the Boeing end of flight simulations.
My question was, if there was pilot input during the end of flight, what is the maximum distance from the 7th Arc that MH370 could have reached, assuming that the BFO data showing a descent at around 15,000 fpm is met?
Case 4 from Boeing, starts at 40,000 feet and 450 knots. After 357 secs, the aircraft is 22.46 NM from the start, the ROD is 15,420 fpm and the altitude is 15,855 feet. After 371 secs (only 14 secs later), the aircraft is 21.07 NM from the start, the ROD is 2,460 fpm and the altitude is 14,020 feet. If the pilot then maintained this descent at around 2,500 fpm (glide ratio 17:1), the aircraft would fly for another 39.26 NM. The maximum distance from the start is around 60 NM.
I agree that it is possible from 40,000 feet with an optimum glide angle to reach 110 NM from the 7th Arc, but this ignores the BFO data at the end of flight.
I have therefore decided to run the next simulations with 25 particles in a cluster up to 60 NM either side of the 7th Arc.
@Richard G: Unfortunately, I believe the potential glide is longer that 60 NM. I think it is possible to recover from an increasingly steep descent consistent with the final BFO values and only lose several thousand feet due to the short duration of the descent and recovery. If the steep descent begins at FL400, I think after the altitude recovery, glides longer than 100 NM are possible.
@Richard
Nice work. Thanks.
I remain concerned that the difference between the arrival time of the debris and the discovery time of the debris is not known.
@Victor
I accept what you say and I will change my 25 cluster points to ± 50 NM and ± 100 NM for the first trial run.
@Dennis
I agree completely that the arrival time can be a long way before the discovery time. Arrival time to beaching time can be long in itself, not just the beaching time to discovery time. The case of Roy shows that beaching and first discovery and re-beaching and second discovery can be 89 days apart. In locations like Rodrigues, the floating debris was trapped behind a reef, so the arrival time to beaching time will be shorter.
This problem is less likely in locations such as Reunion, where there are regular beach patrols, but even in St. Andre there are people who think they saw the flaperon coming close to shore and going away again. This problem is more likely in remote locations which are infrequently visited.
The median discovery time for mainland Africa was 835 days, but this based on only 13 items of floating debris.
@RG: Always keep in mind — for 2/3 of the recovered items, the discovery time is always “whenever it made land” until “whenever Blaine Gibson happened to beachcomb there”. No more, no less.
@lkr
I have focused on the one item found in Reunion (earliest), the one item found in Rodrigues (nearest) and the 13 items on mainland Africa (including the furthest and latest). Only one out of these 15 items was found by Blaine Gibson, at Paluma Sandbank, Mozambique.
Johny Begue, Schalk Lückhoff, Neels Kruger, Liam Lotter, Jean Dominique, Suzy Vitry, Barry McQade, Jean Viljoen and a few others also deserve a mention for being vigilant beach combers.
Most of Blaine Gibson’s items were found at Riake Beach and Antsiraka Beach, Nosy Boraha Island, Madagascar as predicted by Charitha Pattiaratchi.
@Richard
Thank you for the continued analysis. My concern would be in the Alt Config the distance from Arc7 could exceed 100-nm. Perhaps the drift analysis can help us decide just how far from Arc7 makes sense. I currently envision MH370 was possibly heading towards Broken Ridge (125-nm from 30S) possibly as far away as Dordrecht Hole, which is 255 to 265 nm from Arc7 (Lat: -33.5; Long: 101.3333333).
@Victor Iannello,
@Richard Godfrey,
A flight to 31.6S with fuel exhaustion circa 00:17 is inconsistent with FL400. MRC at FL335 fits all observational data. At that altitude, a glide of 100 NM after the BFOs at 00:19 is not possible according to my calculations.
@DrB, @Victor,
I have combined a flight path to 31.566°S 96.744°E at 00:19:37 UTC and altitude of 21,066 feet on track 162.48°T with the Boeing end of flight simulations case 3, 4 and 6:
Case 3 ends up 51 NM from the nominal end point on a bearing of 293°T, 36 NM inside the 7th Arc.
Case 4 ends up 66 NM from the nominal end point on a bearing of 285°T, 45 NM inside the 7th Arc.
Case 6 ends up 54 NM from the nominal end point on a bearing of 118°T, 43 NM outside the 7th Arc.
A lot depends on the altitude at the start as @DrB points out, but around 50 NM is easily achieved and if the start altitude is higher than MRC dictates, then 60 NM or 70 NM is possible.
I have decided to try the drift analysis from the following array of 25 MH370 end points at 50 NM and 100 NM, inside and outside the 7th Arc on all points of the compass.
https://www.dropbox.com/s/phg6tinsh68s0a1/Array%2031.566S%2096.774E%20100%20NM.png?dl=0
Depending on the results, I will make each of the 25 points the centre of a 9 particle cluster and compare the results in terms of median time to Africa, time to nearest pass of Reunion and bifurcation around Madagascar.
DrB: MRC at FL335 fits all observational data.
That is not the only path that fits the data. That is one of many possibilities, especially if we allow pilot inputs after 19:41.
@Richard G: When you estimate the distances from the 7th arc using the Boeing simulations, what do you use as the “nominal end point”? If we accept the validity of the final BFO values, the impact is always less than 10 NM from the 7th arc for no pilot inputs. If we allow pilot inputs, than a controlled glide from a high altitude could result in an impact more than 100 NM from the 7th arc.
@Victor
In this case, I was referring to the nominal end point of @DrB’S flight path of 31.566S 96.774E.
I apologise that I mistyped 96.744E rather than 96.774E in my previous post.
When you say “high altitude”, what do you mean? 45,000 feet? Then the impact point could be 120 NM from the 7th Arc.
I was basing my calculations on MRC mode and @DrB’s FL 335.
@Richard G asked: When you say “high altitude”, what do you mean? 45,000 feet? Then the impact point could be 120 NM from the 7th Arc.
By high altitude, I was referring to a descent from FL400 or possibly higher, such as the example I gave in a previous comment:
I think it is possible to recover from an increasingly steep descent consistent with the final BFO values and only lose several thousand feet due to the short duration of the descent and recovery. If the steep descent begins at FL400, I think after the altitude recovery, glides longer than 100 NM are possible.
@Victor
For FL400, I have already accepted what you say and I have already changed my 25 cluster points to ± 50 NM and ± 100 NM for the first trial run.
Please see my comment:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-19099
@Victor
Looking at Richard’s data for the flaperon, I would characterize any latitude from 23S to 31S as equally probable. Talk of a wider search zone because 31S came up empty is foolish IMO.
@Dennis
My conclusion of 31°S was not based only, on just one criteria of the flaperon reaching Reunion. However, it is important that any candidate MH370 end point fits this criteria.
My conclusion of 31°S was also based on the median time of arrival on mainland Africa, which does not fit 23°S to 29°S as you propose is equally probable.
My conclusion of 31°S was also based on the required bifurcation around Madagascar, which only happens from a start latitude around 32°S ± 1° or around 25°S ± 1°, which also does not fit your proposal.
@Richard
One parameter I am struggling with is starting location.
If we say there was a long glide, say 120-nmm, represents up to 2 degrees in latitude. So we sort of have to say Arc7 starting point=31 South but presumably that could be “Arc8” starting point= 33 South. In other words, the Arc7 location loses some meaning.
@TBill
In this case, when @DrB says the MH370 end point is at 31.566°S 96.774°E on the 7th Arc, the aircraft in his simulation is still at an altitude of 21,066 feet. Therefore the aircraft will finally have flown maybe 120 NM further. You are right, it is possible that the aircraft lands on the 7th Arc 2 ° further south or north. It might also have flown east or west or any other compass bearing. So I take 31.566°S 96.774°E only as a nominal mid point.
The array of possibilities I am considering is
https://www.dropbox.com/s/phg6tinsh68s0a1/Array%2031.566S%2096.774E%20100%20NM.png?dl=0
@Richard
Yes, I understood the scope of your logic for homing in on 31S. I just do not have the same confidence in the arrival times. The flaperon arrival time is probably the most reliable given the location.
Very interesting analysis and thank you all for sharing.
Of the many things that puzzle me, I would like to raise one question, concerning the fact that the possible point of impact may be farther from the 7th arc.
Why would there be any pilot input at the end of the flight, after flame-out?
At that point, it would not matter any more, the fate of the aircraft is sealed anyway.
The only reason I could think of, also based on other’s opinions on this blog, would be to try to keep the aircraft as intact as possible upon impact. And the only reason for that might be to ensure that it is much harder to find.
But why? Why take all these measures so that it isn’t found? This is mind-boggling (at least to me).
@ArthurC
I am completely baffled by the “return to the glide” conjecture. It makes no sense and is contradicted by the data.
One more crazy thought…
What if someone managed to actually put it down “Sully-style”, with minimal damage to the fuselage?
Then it could have floated away to basically anywhere…
If I’m not mistaken (and I hope I am), the pieces that were identified came from areas that could have sustained damage while still maintaining the fuselage’s ability to float: Roy seems to be from an engine, the flaperon attaches to the wing…
@Dennis
You stated “I am completely baffled by the “return to the glide” conjecture. It makes no sense and is contradicted by the data.”
You are right!
A good friend pointed out to me that my steep descent and pull out, then glide scenario is complete nonsense, because it does not fit my previous statement about the IFE transmission.
I stated above:
Case 3 ends up 51 NM from the nominal end point on a bearing of 293°T, 36 NM inside the 7th Arc.
Case 4 ends up 66 NM from the nominal end point on a bearing of 285°T, 45 NM inside the 7th Arc.
Case 6 ends up 54 NM from the nominal end point on a bearing of 118°T, 43 NM outside the 7th Arc.
What I forgot to say was:
Case 3 takes 1020 secs from 00:17:00 UTC.
Case 4 takes 770 secs from 00:17:00 UTC.
Case 6 takes 699 secs from 00:17:00 UTC.
AND I ALSO FORGOT TO SAY … the IFE message was expected from the aircraft at 00:21:06 UTC, which is only 246 secs from 00:17:00 UTC.
My apologies for my nonsense hypothesis about a long glide.
@Richard G.
The whole tone of this post is almost sarcastic, so I had to read it a few times to convince myself. 🙂
Neophyte question: is it possible to drop rapidly and then recover last minute into a glide?
Kind of “Hollywood”, I know, but is it?
@ArthurC
My apologies that I came across as sarcastic, it was not intentional.
It is quite a task with a B777, weighing 174 tons after fuel exhaustion at 00:17:00 UTC, to then achieve a steep descent of 15,000 fpm, just as the Satellite Data Unit reboots at 00:19:37 UTC and the final BFO is measured, then make sure you are over 25 NM from the 7th Arc, but then crash before 246 secs are up at 00:21:06 UTC.
Pilot input would be required to develop a steep descent of 15,000 fpm within 157 seconds from 00:17:00 UTC to 00:19:37 UTC. The fastest time without Pilot Input from the Boeing simulations was 296 secs.
Pilot input would be required to maintain this deep descent at around 15,000 fpm for around 80 secs to impact before 00:21:06 UTC, when the IFE message was expected (after 89 secs from 00:19:37 UTC).
In 246 secs at 500 knots you can travel 34.2 NM, which would take you just outside the 25 NM search width.
@Richard G: The recovery from a steep descent followed by a long glide is predicated on the assumption that the IFE log-on did not occur because the fuel line to the APU ran dry.
At this moment I am reminded of my previous post: http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-19049
containing the question:
“What would happen if the MH370 started programmable VNAV idle power descent before the left engine stopped?”
The above question is imprecise, it should read:
What would happen if the MH370 started programmable VNAV idle power descent before the autopilot was lost?
In my opinion it would cause a completely different trimming configuration in flight without engines and autopilot.
A straight phugoid (perhaps deep) would be possible instead of spiral descent to the water and this does not mean long glide path.
I’m sorry but I can not check it on the simulator.
@Victor
A long glide ending requires much more to be predicated than APU fuel starvation.
1> Why would anyone fly the plane in such a bizarre manner?
2> Why would anyone want to execute a soft ditch after running out of fuel? A soft ditch is much more easily done with fuel.
3> Assertions that a pilot wanted to fool investigators by “inventing” the steep descent and then gliding, assumes the pilot understood the Inmarsat functionality – very unlikely.
4> Assertions that the pilot wanted to hide the plane are similar to suggesting the pilot selected Broken Ridge or Dordrecht Hole. Fantasy.
Lastly, suggesting a wider search at 31S is making the same mistake made over and over again, that is allowing the BFO and automated flight modes to over-constrain flight paths. We have all seen this movie a few times, and so far it has not had a happy ending. Suggesting a wider search at 31S is simply not an actionable alternative. You may as well tell the searchers to give up.
@DennisW,
If it were up to you, no search would ever have been done (as you have recommended in the past). Basically, you were telling them to give up before they even started. It a high risk venture, but we do know a lot more now, especially about where it isn’t and maybe about where it is. Your pessimism never changes, and it does nothing to further the search. As the old saying goes “Nothing ventured, nothing gained.”
@Greg: I’m not understanding the difference in trim that you propose. The elevator trim (really the stabilizer position) depends on calibrated airspeed, and to first order, is independent of vertical speed, which is determined by thrust. Meanwhile, the tendency to roll away from level wings is determined by lateral aerodynamic and thrust asymmetry.
@DennisW: I agree that the prospects for success, based on what we know now, is not high.
@DrB
Your assessment of my search recommendations is correct. I have never felt that the science justified the decision to spend the money associated with underwater searching. I also have opined that even if the wreckage is recovered, it will simply tell us the plane was flown to where it was found.
Sorry, not uplifting I know.
@DennisW,
@Richard Godfrey,
As Victor has pointed out, the potential of a glide after 00:19 is not contradicted by the data. It simply requires the normal electrical configuration so that the SDU reboot is caused by left engine fuel exhaustion circa 00:17:30. The APU then auto-starts by 00:18:30, the SDU logs on at 00:19:29, and the APU runs out of fuel in the line before the IFE transmissions begin circa 00:21.
The unresolved question is why the Boeing (unpiloted) simulations do not show a high and increasing descent rate at 00:19:29/37. Is there a problem with Boeing’s simulations, or does it mean that the plane was already being piloted at 00:19:29 and thereafter?
@DennisW,
You have considered the positive reward of finding the aircraft, but you have not considered the “negative reward” of not looking.
There is a large, known risk to Malaysia/MAS if no search was ever done. The negative publicity would be significant with concerns like “they don’t care” and “they are hiding something”. To avoid this, they did a search. Then they decided they would not continue unless they learned where the aircraft actually was. Then, unexpectedly, Ocean Infinity came along with an unusually attractive offer, which they accepted since it didn’t have any downside.
My point is that you have to consider the “negative reward” as well as the positive reward when making a search decision. Unfortunately the value of the negative reward of not looking (again) now is close to zero. Malaysia will correctly assess that they have already done what has been expected.
I would be surprised if recovered debris and data recorders shed much light on motivation/cause, but they may be helpful in eliminating accidental causes.
@ArthurC
We do not know what type of landing was planned, but my current thinking is a flaps up, higher speed ditch (not survivable). As far as why? Reduced debris, quick sinking, but overall the facts seem to me possibly consistent with a deniable hijacking. So I am expecting cockpit appearance management is important, with the pilot probably wanting to be out of the cockpit at impact. If the cockpit data recorder was operating for the flight (questionable) I expect the co-pilot seat will show up as where the pilot in charge sat. I am expecting an attempt to hide landing spot as well as attempt to avoid leaving incriminating evidence on flight computers etc.
@DrB
Yes, there is a public image component to the search. No doubt that is worth more than actually finding the wreckage.
As you say, the parties involved are pretty much free of any rebukes based on seach expense and level of effort. I don’t think there is any appetite left for funding additional search efforts.
My “negative” attitude was always based on the pure science advisor model, and not on public perception concerns, which I am totally unqualified to assess. On the basis of the information we have and have had, an underwater search has never passed my threshold for recommendation.
@DrB
As an addition to the above, I don’t think it is at all appropriate for a government to ask the science team “should we search?”. The appropriate question is “where should we search?”. As I have opined previously, I think the IG has done a very responsible job of answering the “where” question based on the information available.
@DennisW,
Unfortunately, only the Malaysian government has done the asking about where to search, and as far as I know, only ATSB has been asked since the very early days. Nothing the IG or anyone else outside the official team has done has affected the search plans to date. I don’t see any reason why Malaysia would continue to search at all, much less ask someone else where to look.
@Victor, @DrB,
Many thanks for pointing out that there were good reasons why the IFE message was not received other than impact: (1) The APU ran dry of fuel before 00:21:06 UTC … or even … (2) The IFE could have been selected off in the cockpit.
Richard Godfrey says: …/…(1) The APU ran dry of fuel before 00:21:06 UTC…/…
If the plane was pull off the dive, could the APU have run dry from the G forces applied to the fuel in the duct? And didn’t start again because the fuel flew back and there was air in the duct on the APU side?
@DrB said: Nothing the IG or anyone else outside the official team has done has affected the search plans to date.
That’s not correct. The search conducted by Ocean Infinity was heavily influenced by some in the IG.
@Richard G said: or even … (2) The IFE could have been selected off in the cockpit.
The IFE switch in the cockpit does not prevent an IFE log-on. Rather, it inhibits IFE functionality at the passenger seats.
Thank you all for replying and even bothering to consider my post.
@Richard
I realized that you weren’t being sarcastic after re-reading your comment. I guess I was confused by the apologies, as for someone with your knowledge and expertise, I would have expected a more “smug” reply. 😉
And no, I am not polishing the proverbial apple, I merely stated my personal observations.
@Victor
Many thanks again for putting me right about the IFE switch in the cockpit. I was only repeating what I had read in the ATSB report dated 3rd December 2015 without checking: “The fact that the expected IFE system transmission was not received could be due to: the IFE system being selected off from the cockpit overhead panel at some point after the 18:25 logon”.
@Richard G: Yes, I believe we determined that statement in the ATSB’s report to be incorrect.
@DennisW wrote:
“Why would anyone want to execute a soft ditch after running out of fuel? A soft ditch is much more easily done with fuel.”
One possibility would be that the pilot recognised that SATCOM reconnected because that is what they saw on the display, and then decided to glide further before crashing. To me, any scenario makes sense only if you presume two or three changes to the original plan, at least. That makes it too complicated to speculate about the original plan, or possible changes to it.
———-
It should also be noted that the ATSB feels reasonably confident to exclude a crash within 45 nm aroun c. 31 S. Does this mean MH370 crashed further south than those 45 nm? The surface search clearly concentrated on the northern end of that search width, so I think that can reasonably be excluded.
https://www.atsb.gov.au/media/5772655/mh370_ocean_driftii_final.pdf
pdf, p. 22
I don’t think you could exclude the area further south than 45 nm from the arc at 31S.
@Nederland
I am sorry. I simply cannot seem to understand what you are trying to say in your last post. Could you please read it carefully and see if it makes sense to you? Of course, the problem could also be me.
@DennisW
Two points:
a) The pilot saw the final reboot on his screen, then recovered from the uncontrolled descent and glided an unknown distance (“they” is meant to be gender neutral rather than plural … well at least in British English, it may not work in American English). This could explain a “soft ditching” after fuel exhaustion.
b) (general remark, not specifically addressed to you) The surface search around 31 S (the endpoint which I think is most likely) concentrated on the area north of the arc. This means it is still possible that MH370 glided from 31 S to an area south of the arc.
@DrB
I understand that your MH370 flight path analysis concludes MRC mode.
Does your MH370 flight path analysis specifically exclude step climbs to say 40,000 feet during the flight?
@Nederland
You stated “The surface search around 31 S (the endpoint which I think is most likely) concentrated on the area north of the arc. This means it is still possible that MH370 glided from 31 S to an area south of the arc.”
The Ocean Infinity search width around 31°S was reduced to ± 22 NM. At 32°S the search width was still ± 25 NM. There was a balanced search both to the north west and south east of the 7th Arc in this area. The search width was sufficient based on the assumption there was no pilot input, but insufficient if there was pilot input.
@Richard G:
Yes, that was a reasonable assumption at the time. The assumption of no pilot input was, however, challenged, by the French flaperon analysis, published only with the recent safety report.
The ATSB report indicates that it is unlikely that MH370 crashed in an area within 45 nm of the arc around 31S because this area has been covered in the surface search. The surface search, however, concentrated on the parts north of the arc (since that part was searched only once the results of the private simulator became known and the search team had to take into account drift patterns).
If you look at the image on p. 22 (pdf – p. 16 of the paper), left, then it looks very unlikely that MH370 crashed anywhere north of the arc in the region of 31 S, but it is entirely possible that MH370 crashed 45 nm or more south of the arc in that region.
@Nederland
What is not evident from the ATSB graphic is, where the relevant blue area at 31S becomes a geeen area on the day of the surface search.
The black polygons describing the daily surface searches cover a huge planned area. In my analysis of the surface search, there was only an average actual coverage of 11%.
@Nederland
You make a good point. If we envision that Arc7 is not the end of flight arc, then we need to re-interpret everything (surface search, drift studies etc) for hypothetical final arcs presumably further away from Arc6. It sort of makes logical sense too (consistent with not finding MH370 to date).
@TBill
You are right, the 7th Arc is only a central guide.
Even the 6th Arc is relatively meaningless.
We need to define the 8th Arc and a 5.5th Arc.
Even then, we need to be careful, whether MH370 flew 120 NM straight on, or did one or more loops left or right, sideways or even back on itself.
@Nederland
The aerial search of the SIO began on March 18 with a single P3.Any conclusions derived from the aerial search are a joke much less a probability map.
@Richard
I agree with you that the aircraft did not necessary fly further, it could have changed up heading as a misdirection tactic like George Bibel speculated.
However unless we get a new lead or clue, I currently see two main search options:
(1) More northerly between 20-25 South within +/- 25 nm
(2) What I view as the ZS simulator path, keep going beyond Arc7 at about 30 South crossing point on Arc7
Option-1 could reasonably be considered as the unfinished part of the Arc7 search. Option-2 may be the only wider path search currently justified since we have a clue from the pilots sim cases, that that might be the area to look.
@DennisW
The search in the area around 31S started on 28 March and ended on 28 April.
@Nederland
Thx
@Victor,
Are you sure about the IFE? I know from an AMM extract that the IFE power switch in the passenger cabin(flight attendant switch) only partially removes power from the IFE, the AMM I have does not feature the IFE/PASS SEATS switch so I don’t know if it is the same.
From the FCOM below however:
IFE And Passenger Seats Power Control
Electrical power to the in-flight entertainment (IFE) system and passenger seats is controlled by the IFE/PASS SEATS power switch on the electrical panel. With the switch ON, the IFE system and all passenger seats and related systems are powered normally.
Pushing the IFE/PASS SEATS power switch OFF removes power from the
following:
– IFE (all components)
– Passenger seats (including seat motor power, personal computer power
outlets, and telephones).
Is the FCOM incorrect?
@Pilatus: If I recall, there was a change to the functionality of the IFE switch. Are you sure the FCOM extract you provided applies to 9M-MRO? I’d have to dig up documentation specific to Malaysia Air’s B777-200ERs. I believe others have already verified this, but there is no harm in checking. If I recall, @Don Thompson and/or @Andrew have researched this.
@pilatus
The intent of the IFE/PASS switch, originally provided only at the ‘purser station/Video Control Center’ location is that it removed power from the IFE equipment in the cabin: the seat display units, the passenger control units, the seat electronics boxes, the area distribution boxes, and so on.
A Boeing SB introduced the IFE/PASS switch to the flight deck that extended the scope for removal of power to include the components of the video control center itself.
The IFE LRUs located in the MEC, the CMEUs, PIIC, EPESC, RAD, provide integration with other aircraft systems and do not present a risk to the cabin. These other aircraft systems include the Passenger Service System, the Cabin Services System, and the Passenger Address System, and certain AIMS functions.
The SB describes the reason for its implementation:
The change described in this service bulletin will provide new switches on the flight deck than can remove power from the cabin systems if smoke or flames occur. Smoke and flames in the passenger cabin can cause injury to the passenger and damage the safety critical cabin equipments.
We have received numerous reports of smoke or flames in the passenger cabin related to the wiring for the passenger cabin in-flight entertainment system, cabin lighting and passenger seats. It is not easy to remove the electrical power to these systems to stop the smoke or flames.
The implementation procedure for the SB does set out a check for the powered state of the MEC located LRUs after re-work, and with the IFE/PASS switch set to ON. The subsequent procedure for the switch set to OFF does not involve a power-off check for the MEC located LRUs as it does for all the cabin located units. Unfortunately, the SB doesn’t detail the relavent ELMS-EEU data load which should provide an emphatic answer.
@Andrew & I reviewed this previously with the conclusion that the IFE’s MEC LRUs are not powered off by setting of the IFE/PASS switch to OFF.
@Don. Thank you very much for the detailed an authoritative response as usual – though some of the acronyms are new to me…
My key interest is in how the cabin-related components, or components sharing the same circuit, affect the process of error-checks and inputs required for successful IFE logon. Did the implementation of this SB change the way in which the whole IFE system is wired so that the cockpit switch now affects more components than just the pass. seat aspects? By extension, did it affect whether load-shedding of the cabin-side components of the IFE would inhibit successful IFE log-on?
I gather that the opinion to date was that it wouldn’t if this is passenger seat aspects of the IFE only. But do we *know* which components of the IFE are load shed, considering the re-configuration of power under the SB referred above?
@Paul Smithson: We do know that the IFE head was powered on the ground when the only source of power was the APU.
@Don Thomspon
“@Andrew & I reviewed this previously …”
Just for further background, when was this review conducted?
I am thinking it was earlier here in 2018, such that prior to that review, many had presumed the IFE cockpit switch in the OFF position could account for the lack of IFE logon after 00:19:37 (unless the aircraft crashed at that time).
So now, we are mostly thinking either (1) the aircraft crashed before IFE re-logon, or (2) IFE power was cut via MEC Bay circuit breakers or by cutting power to the SDU (e.g. via Left Bus isolation).
@TBill: I think the most likely reasons for no IFE log-on are 1) Crash or 2) APU ran dry of fuel.
@Victor. I understand that you believe these to be the “most likely” reasons. But can load-shedding be definitively ruled out or not?
@Paul Smithson: You’d have to devise a reason why the IFE head was not load shed on the ground under APU power but was in the air. Also, @Andrew has documentation that suggests the IFE head is not load shed.
@all
Coming back to the major turn around 18:40 UTC, there was traffic in both directions around #mh370 just prior to that time. Not only Emirates343 but an Airbus A320 of Indigo coming from Chennai.
Here is a short video illustrating the traffic reconstructed and simulated based on data coming from the 3 major live traffic trackers https://youtu.be/goD03V1wxKE.
It provides additional info to believe that a contengency procedure took place at about 18:40.
@Jean-Luc Marchand: With TCAS inoperative with the transponder in standby, are you suggesting that MH370 was aware of other traffic?
@Victor. Are you sure that on the ground the aircraft is not plugged in to “shore power”?
@Paul Smithson: Good question. I’ll put that out to the group to answer.
@victor
Yes, definitely. I would have been the hijackers I would have checked the type of traffic to expect on my way.
EK343 was coming and converging from the left: so visually fully visible, IGO53 was oncoming so fully visible. Weather conditions: very good. At night, in those conditions the lights are visible several tens of Nm away. Having traffic behind at a close FL and an oncoming traffic at an unknown FL, the safest way is to offset first. Then the crossing of N571 becomes delicate because the “lost ground” of 2×15 Nm due to the offset makes the crossing coincinding with EK343 which was ~27Nm behind. Thus a descent is logical to pass underneath the route.
Without TCAS the People in Command must have been carefull to monitor the surrounding traffic visually as it was their only way to stay safe.
For the record, IGO53 was late by 30min at take-off compared to its regular schedule.
(also referring to K. Tee, she could see the traffic from the ground : the MH370 and TWO other aircraft)
@Jean-Luc Marchand: When I first proposed the lateral offset to satisfy the BTO data, I looked at whether it could have been caused by the proximity of EK343. I concluded that at 27 NM, the minimum distance was too far for the pilot of MH370 to be concerned by EK343. You came to the exact opposite conclusion.
I’d be interested to hear what others think.
@all
Returning to a broader issue – the search strategy.
I do not believe the ISAT data was spoofed. That assumption generates two definite and uncontestable conclusions:
1> The aircraft flew South after the FMT.
2> The flight path ended with a very rapid descent.
The debris findings (also indisputable) support 1> above.
The ISAT cannot be used to refine the terminal location of the aircraft. That has been demonstrated several times. All that can be concluded from the ISAT is 1> and 2> above. Drift analytics might be able to provide some refinement.
The BTO data is very reliable. The speed of light has not changed significantly in at least 400 years ( since the first successful measurement by Danish astonomer Roemer). The location of the last arc is very well known, and the aircraft was descending rapidly at that time. It makes perfect sense to continue searching close to the last arc unless there is evidence to support a descent recovery and glide after the last ISAT exchange. There is no such evidence. In fact, there is evidene suggesting no recovery was made (lack of IFE log on).
A prudent (and intelligent) person would continue to search along the arc to the North of the area previously searched. Searching wider at any latitude has no rational basis.
@DennisW: What happens if the best drift models suggest that the impact was south of 25S latitude?
@Jean Luc
That is a very good picture of the N571/Malacca Straits air traffic. I have tried to save as many FR24 views that I could find (years later), but that video is very nice.
How are you getting EK343 position? I am thinking it is an estimate based on the assumption it was on N571 and adjusting the FlightAware data accordingly.
@TBill
In the video the yellow aircraft icons come from the flight trackers directly. The green icons are extrapolations based on data coming from these trackers, from the knowledge of the MH370 trajectory and radar data.
EK343 was traced by radar data up to it exited the Malaysian radar coverage. Then estimating its speed and knowing it was on N571 we extrapolated its position in time.
For Indigo IGO53 coming from Chennai, we did basically the same but we have extrapolated its position backward in time.
@Victor
The drift models are hampered by not precisely known times of arrival. I regard the flaperon as one of the most accurate arrival times since it was found in a populated area with regular beach patrols. I think we can say a time of arrival of 508 days (stated recovery time) to as much as 60 days earlier. Using Richard’s flaperon graphic there is a very flat area in the 450 to 508 day area spanning 31S to as far North as 23S.
@Richard Godfrey,
You said: “Does your MH370 flight path analysis specifically exclude step climbs to say 40,000 feet during the flight?”
That is a good question. Purely from the standpoint of endurance, the short answer is that step climbs are probably allowed.
Comparing MRC at FL340 to FL400 at W = 190 tones, the air speed increases quite a bit from M0.75 to M0.82, while the fuel flow increases a bit over 2%.
The best fit for 31.6S at MRC is FL336 with an average (predicted) PDA of 2.7%, assuming the normal electrical configuration with L engine fuel exhaustion at 00:17:30. That leaves about 1.2% extra fuel since the PDA is actually 1.5%. That fuel margin would allow step climbs following the optimum altitude as weight decreases.
I’m not sure about the impact of the higher airspeed at higher altitude on fitting the last several BTOs and BFOs. Probably it gets worse, but maybe it is still acceptable. I don’t have access to a suitable computer at the moment to test this.
@Jean-Luc Marchand,
Your air traffic simulation is a nice way to visualize the relative aircraft positions.
One point I would make is that in your simulation 9M-MRO appears to slow down after the lateral offset maneuver. If that did not happen and the speed were maintained at M0.84 through the southwest turn, the separation between 9M-MRO and EK343 at that turn would have remained significant (despite the lateral offset) because of the higher airspeed of 9M-MRO along N571.
@DennisW,
In my opinion assuming 60 days of discovery delay is excessive for a beach on Reunion which reportedly had weekly cleanups.
@DrB
http://www.therakyatpost.com/world/2015/08/02/mh370-was-flaperon-actually-found-in-may-on-reunion-island/
@DrB, @DennisW: Said another way, the weekly cleanup schedule gives a maximum time of seven days for a part to wash ashore, not be recovered, and get carried back out to sea in a high tide.
@Victor
What we can say is that the arrival time is shorter than the discovery time (I think considerably shorter in most cases,) and that pushes the terminus to the North.
@Victor, @Dennis, @DrB,
The flaperon was found on 29th July 2015. The high tides at Saint Andre were at 11:51 Local Time with a height of 0.62m and the night before at 22:49 Local Time with a height of 0.65m. This was 2 days before full moon and the highest tide in the current cycle at 0.74m.
The previous highest tide at Saint Andre, Reunion was on 17th July 2015 with a height of 0.69m. The high tide before that was on 4th July 2014 with a height of 0.74m.
There were 2 unverified satellite photos supposedly showing the flaperon beached, taken on 23rd July 2015 and 13th June 2015.
@Richard Godfrey. You have the flaperon at unbuoyed drifter speed as I understand it. David Griffin has it at that plus 10cm/sec. Even then the results of his experiments are scattered. The flaperon could travel upside down or the right way up, depending on which way up it was when the wind last dropped below 20 knots. Then there are gyres such as delayed ‘Rodrigues’.
I think to treat this as other than one of a bunch implies an unwarranted accuracy in its modelling. Like the others, there have been no error limits established as to drift speed, whether the assumption is unbuoyed drifters or 0.12% wind speed or that plus 10 cm/sec.
It is only if drift (leeway) is a small percentage of total speed that inaccuracies in its estimation can be disregarded, but your choice now of unbuoyed drifters suggests that the difference can be substantial.
@David
Your understanding is correct, I am using the historic average between the year 2000 and 2018 of 280 undrogued drifters. I select the data relevant for the current position and for the relevant seasonal time frame to determine the position each day along a particular track. I am not accounting for leeway and drift angle.
I agree the flaperon is only one item of floating debris and hardly a statistical basis for analysis. That is why I also looked at the 13 items that have been discovered on mainland Africa, which is still not a large number for statistical purposes.
I agree that the leeway and drift angle of each floating debris item will be different and may change over a time due to water logging and position changes (leading edge, trailing edge, intrados up, extrados up, etc.). This will also change the drift speed and drift direction.
I agree that the drift analysis is an imprecise science, nevertheless I was encouraged that my latest results come much closer to those of David Griffin and Charitha Pattiaratchi.
@Victor
@Jean-Luc
A couple of implications of Jean-Luc’s air traffic visualization:
(1) ATSB ruling out of descent at 18:40 looks like it was a very bad assumption
(2) Iannello/Godfrey have been correct by first suggesting the offset, and furthermore questioning (in a report) the ATSB assumption above
Not saying 18:40 descent and 18:25 offset are 100% definite…we still don’t know exactly…but ruling it out looks silly if not suspicious in hindsight.
@TBill: I’d say both the lateral offset and the descent are possible, but far from proven. The lateral offset doesn’t have much effect on predicting the end point, but the descent does, as it delays the turn to the south and pushes the end point further north.
@DrB
Yes the 9M-MRO was slowing down during the descent and at FL270. We think that this took place something at around 3 minutes before the turn. It is visible in the video on the last part of the Contengency Procedure segment before the turn.
Another aircraft IGO53 was oncoming towards 9M-MRO and closing up. And taking its chances to turn at the same FL would have been like a Russian Roulette gambling.
As speed is concerned, EK343 speed was monitored by ADS-B at around 502kn by Planefinder. 493kn from FlightRadar appears as a computed speed as no ADS-B could be received from there. I see not reason for EK343 speed to be that much different from 9M-MRO in the contrary :-): same weather conditions. In fact EK343 (A6-ECM) is equipped with GE90-115B engines which are much more powerful than the RR892.
@Jean-Luc Marchand: MH370 was not constrained to fly along an airway. If it was concerned about a collision with EK343 (which is doubtful considering the distance), it could have simply turned left. This assumption that MH370 was following SOP and being safe by offsetting to the right and descending before turning to the left seems very unlikely.
@all
As we approach the two year aniversary of this paper I am reminded how much I like it.
http://www.duncansteel.com/archives/category/mh370
@DennisW: You said:
1> The aircraft flew South after the FMT.
2> The flight path ended with a very rapid descent.
The debris findings (also indisputable) support 1> above.
I’d say the debris findings support both (1) and (2). And I am inclined to agree that (1) and (2) are stronger than what we can glean from drift studies because of the non-deterministic nature of drift paths, and because of the uncertainty in the delay between the beaching and discovery of debris.
@DennisW: The reconstruction of flight paths near airports provides some sense of reasoning for why the flight path was selected. But how do you reconcile this with the simulator data found on the captain’s computer, which does not show either Cocos or Christmas Island as destinations?
@Victor
I have nothing that reconciles the simulator data other than the intent of the simulation was focused on the FMT and not the terminus.
@DennisW: If you take the simulator data at its face value, it suggests a plane that runs out of fuel at 40,000 ft in the SIO followed by a glide.
Another possibility is the simulator data was meant to be found, and it was a misdirection. Since the final points were at 45S,104E, we would dismiss this location. That still leaves a lot of ocean to search.
@Victor. The drift studies help in 2 ways, firstly in providing another broad opinion and second in making apparent their own limitations (ie without all that work out we would not know what their limitations are).
However as you indicate, while contributing they should be attributed less weight than the more deterministic or be a deterrent from seeking that.
Remember that the flaperon find, without any experimentation, was seen to be “consistent with” the then search assumptions.
The CSIRO does not rule out the possibility of a crash further north though it believes that its drift analysis indicates that to be unlikely. I know not where “consistent with” would fit between those two extremes.
Were chance delay of some items in a gyre (as per the @Richard Godfrey paper of two years ago and referred to by @Dennis W above) added to the prospects of further delays in discovery/recovery after reaching the target zone and to unknown errors in estimating leeway, the ‘possibility’ entertained by the CSIRO would strengthen.
@Richard Godfrey. You said, “I agree that the drift analysis is an imprecise science, nevertheless I was encouraged that my latest results come much closer to those of David Griffin and Charitha Pattiaratchi.”
Can you tell us whether your model supports the CSIRO assessment that a crash towards the bottom end of the previous search is ruled out by the absence of debris being discovered (ie being retained) on WA shores?
@David
You asked “Can you tell us whether your model supports the CSIRO assessment that a crash towards the bottom end of the previous search is ruled out by the absence of debris being discovered (ie being retained) on WA shores?”
I will do. Please give me a few days to run the simulations.
The common debris origin area based on 11 found debris +/- 2° is in a square 4-5 S and 80-81E which perfectly matches reverse drift model showing all possible origin points in the following article below
https://image.ibb.co/mYMNbe/SIO_11_debris_4_S_80_E.jpg
NOAA buoys: 7 568, 20 331, 32 938, 34 379, 57 894, 57 898, 70 959, 9 421 177, 9 729 754, 9 808 131, 9 810 599, 9 826 019
http://thehuntformh370.info/content/location-mh370-reverse-drift-study-based-debris-found
@Ryan
What is your reason for completely ignoring the 7th Arc and the satellite data?
@Victor
Another wild guess for the 45S 104E simulator coordinates is that they merely serve as LNAV waypoints to create a path slightly to the West of the COCOS. The Western offset would be logical for a 15/33 runway heading.
@DennisW: Thanks for continuing to think about this, as its one of the (many) inconsistencies that bother me.
All airline pilots would know that a custom waypoint can easily be created using a radial (direction, distance) from a standard waypoint, such as an airport.
@DennisW,
@Richard Godfrey,
There are two significant biases in predicting the flaperon discovery date on Reunion using Richard’s drift model. They are in opposite directions, and both are probably large (months). As DennisW pointed out, the flaperon arrival date may actually have been as early as May 2015. That shifts the predicted 7th Arc latitude northward. The second effect is the difference between the drift speed measured using flaperon replicas and the drift speed of undrogued drifters. The flaperon drifted significantly faster. This effect, which is not included in Richard’s current results, shifts the predicted 7th Arc latitude southward, perhaps by an even larger amount.
In my opinion the ensemble of non-flaperon items should provide a more accurate estimate of start latitude because of the much larger number of items being averaged and because they are much more likely to match undrogued drifter speed than the flaperon.
@Dennis, @Victor,
45S 104E is exactly 3,000 NM from 4N 93E.
@Richard Godfrey. WA debris. Thanks.
@Richard. That’s an interesting observation. I’m not sure how rounded the start and end points were, but for what it’s worth using those “whole degree” coordinates GPSvisualizer gives great circle distance of 2989.5NM initial bearing 169.9, final bearing 165.6.
@Richard
I was just backtracking 22 795 NOAA buoys with 35 702 067 GPS coordinates, it has nothing to do with the 7th Arc and satellite data.
1 Saint-Andre, Reunion Flaperon Johny Begue 29.7.15 -20.916183 55.649150
2 Klein Brak River, Mossel Bay, SA Engine cowling Neels Kruger 21.3.16 -34.093767 22.149905
3 Xai Xai, Mozambique Right Hand Engine Fan Cowling Liam Lotter 25.8.16 -25.060000 33.696000
4 Vilankulo, Paluma Sandbank, Mozambique Tailstabilizer Blaine Gibson 27.2.15 -22.088570 35.519020
5 Var-Brulé Beach, Rodrigues Panel segment from the main cabin, associated with the Door R1 closet, Internal Jean Dominique and Suzy Vitry 30.3.16 -19.738711 63.471632
6 Pointe Bernache, Mauritius Left Outboard Trailing Edge Section (external) Tourist 10.5.16 -20.023383 57.701386
7 Gris-Gris Beach. Mauritius Flap Track Fairing Tail Cone Coast Guard Foot Patrol 24.5.16 -20.522448 57.537621
8 Pandane Resort, Mozambique Flap fairing Liam Lotter 30.12.15 -24.078057 35.499796
9 Kojani, Tanzania Right Outboard Flap Locals 23.6.16 -5.076946 39.851445
10 Antsiraka Beach, Cape Anstsiraka, Madagascar Right hand Nose Gear Forward Door Blaine Gibson 12.6.16 -16.849167 49.807500
11 On the beach in front of Linga Linga Lodge, Linga Linga, Inhambane, Mozambique (near Morrumbene) Jean Viljoen 25.8.16 -23.738641 35.399196
@Paul Smithson
https://www.dropbox.com/s/xokje2it4w1bwnc/45S%20104E%204N%2093E.png?dl=0
5,557 km is 3,000.54 NM according to the online calculator I use.
@Ryan: Are you Ken St Aubin? If not, do you collaborate with him?
@Victor
No but I know Ken St Aubin and he know me too.
@Paul/Richard
For a spherical earth (mean radius 6371km) I get a distance of 5557km (3000.54nm). Same result as Richard.
For a WGS-84 geoid I get a distance of 5517.8km (2979.4nm), using Vincenty’s formula.
@Victor
@DennisW
Sorry but I see no apparent significance to 45S 104E other than a flyover point after a coffee break during the sim studies. I suggest the we need to look along that sim flight path to see what the end-target strategy might have been, and to me Broken Ridge/Dordrecht Hole is implicated, and that is also directionally where DrB@31.6 South, Nederland@30-31 South, possibly Richard’s@30S, and drift studies tend to suggest.
Admittedly drift studies are not available for longer glides away from Arc7 to the Broken Ridge area, which I would ask for that (Richard is working on).
@Victor, @DrB, @Dennis, @David, @TBill, @Nederland,
With no pilot input, there are a number of possible flight paths that fit the satellite data, aircraft performance and meteorological data. With pilot input after 19:41 UTC, this number increases substantially. There is evidence, from the consistency of the satellite data between 19:41 UTC and 00:11 UTC, that MH370 was under autopilot control for this part of the flight. At the end of flight, especially after fuel exhaustion, there may well have been pilot input again, even if there was no active pilot input between 19:41 UTC and 00:11 UTC.
@DrB performed a thorough analysis of all possible autopilot methods, fuel data, weather data and the alignment of the possible flight paths to the satellite data in his paper published 7th March 2018 and updated 1st October 2018. He concluded that the optimum flight path reached the 7th Arc at 31.566°S 96.774°E.
In my drift analysis published on 8th October 2018 at the link below, I concluded that the best fit to the Flaperon arriving on the North-South line through St. Andre, Reunion and within 200 km from St. Andre, Reunion, just prior to 29th July 2015 (after 508 days from the 7th Arc) is from a start latitude at 31.1°S ± 1.0°. I also concluded that the best fit to the median discovery time of the 13 MH370 Floating Debris items reaching mainland Africa is from a start latitude at 31.1°S ± 2.0°.
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-19092
I have now run my drift simulation for 25 possible drift start points in a grid ± 50 NM and ±100 NM around a central point of 31.566°S 96.774°E. I have noted the number of days to pass the north-south line through St. Andre, Reunion as well as the number of days to reach land on Madagascar or mainland Africa. I am using the historic average speed of undrogued drifters. I do not account for leeway, therefore I have compared the results for the arrival time near Reunion to the discovery time at Reunion after 508 days in 3 bands of within ± 10%, within ± 20% or greater than + 20%.
The results are shown in the graphic linked below. There was only one drift start location that matched the ± 10% band and only 5 drift start locations that match the ± 20% band. The one drift start location within ± 10% band is still within the Ocean Infinity search width of ±22 NM. However, three of the drift start locations within ± 20% band are inside the 7th Arc to the north west of the searched area and between 22 NM and 100 NM north west of the 7th Arc.
https://www.dropbox.com/s/nfehjixq182pp1b/Array%20End%20Points.pdf?dl=0
I conclude that it is possible that the MH370 end point is in the area around 31°S near the 7th Arc, but may have been missed, because of pilot input at the end of flight taking the flight outside the search area. There are two possible areas outside the previous search area, that fit the drift analysis: (1) between 29°S and 31°S and between 22NM and 100 NM inside the 7th Arc to the north west, (2) centred on 32°S at 100 NM outside the 7th Arc to the south east.
All the MH370 end points that fit the drift analysis, imply that the aircraft more likely rolled to the left after fuel exhaustion. The drift analysis does not support the case that the aircraft carried straight on for 120 NM. All the drift start points to the north of Broken Ridge experience relatively direct transoceanic tracks. All the drift start points near or south of Broken Ridge experience gyres en route. Only one out of the 25 tracks actually lands on Reunion. 16 out of 25 tracks end in Madagascar and 3 tracks make it to mainland Africa.
An example of a direct track is linked below:
https://www.dropbox.com/s/jgj0v0aszwrvo47/Drift%20Map%2030.7332S%2097.5208E%2018.8483S%2049.0455E%20511d.png?dl=0
An example of a track plagued by gyres is linked below:
https://www.dropbox.com/s/zjfp6w3mga4srr1/Drift%20Map%2032.3988S%2096.8649E%2020.0017S%2051.3181E%201200d%20Gyre.png?dl=0
This exercise can be repeated for other possible MH370 end points.
@TBill said: Sorry but I see no apparent significance to 45S 104E other than a flyover point after a coffee break during the sim studies.
In the simulation, the plane was manually moved near 45S,104E, flown a short distance, and then the altitude was manually changed from 37,651 ft to 4,000 ft. The move to that part of the SIO was deliberate, not by chance.
@Richard
Thank you for the analysis…that is perhaps groundbreaking as the new type of calculation we need. Probably most consistent with my current thinking is your option “(2) centered on 32°S at 100 NM outside the 7th Arc to the south east.”
Do you feel the gyres near Broken Ridge would tend to take some debris to the Australia shores too?
@Victor
If DOTEN to NZPG was the flight path showing on the flight sim MAP, then there would be that flight path showing to be able to drag the aircraft icon. Why he chose that particular 45S end location I don’t know, but I was thinking checking out end of flight ideas, but not necessarily locaton per se, or at least not consistent with Beijing flight fuel load. Of course on flight sim we do not get underwater features on the ocean map so just vast blue.
@Richard G: The problem with recommending a search zone at any latitude south of 25S is the final BFO values, the lack of IFE log-on, and the recovered debris together suggest a high speed descent ending in a high energy impact close to the 7th arc. Perhaps the pilot recovered from the high speed descent and glided some distance. Perhaps the missing IFE log-on was due to some unknown system configuration. Perhaps there was a high energy impact after a controlled glide. But on the face of the evidence, there are some important factors suggesting an impact close to the 7th arc.
@Victor/Richard
As usual there is more than one way to evaluate the debris finds. As in the case of BFO error, some pretty advanced degree people elected to compute a mean and variance, which is certainly defined, but is nonetheless meaningless.
Likewise the interpretation of the debris finds. I don’t know why Richard elected to use a median value of the find elapsed time from the diversion date as the transit time. I would be inclined to say the mean value of the three fastest times characterizes the transit time, and mean and variance of the remaining data relative to that time characterizes the mean (and variance) of the discovery time.
When one looks at the data in this manner you get a transit time of 690 days and a mean discovery time of 97 days. This moves the terminal latitude to the area of 25S to 26S.
@Victor/Richard
The discovery time of the three “fastest” pieces is certainly not zero so some additional movement North is appropriate.
@DennisW
Richard did state that he had made no allowance for leeway due to windage. Given that the prevailing winds are generally from the ESE to SE, most objects would have received some positive assistance during their voyage. The Flaperon considerably more than other items, but exactly how much in an open seaway is currently unknown. A matter that could be addressed by a trial.
Notwithstanding, it would be reasonable to assume that consideration of leeway due to windage would bring the start point(s) further south.
@Barry
I cannot comment on the physics. I am not qualified to do that. I can comment on math.
@Barry Carlson: Richard used the undrogued drifters as the baseline, which already includes 1.2% windage as a proxy to account for Stokes drift. Unless there is reason to believe that the debris has more leeway than the undrogued drifters (which is hard to believe since most of the debris floats quite flat on the water), I don’t think additional leeway is warranted in the drift model for most of the debris.
@Dennis
The known/unknown physics here is “fuzzy”, which tends to make the math “fuzzy” and likewise the outcomes.
@Barry
What is not fuzzy are the BTO values. Likewise the BFO values suggest a rapid descent. People advocating a wider search are simply misguided. Richard’s work, as always, is simply amazing. I just question his interpretations. Not a biggie.
The searching is over, IMO.
@DennisW
So you would discourage OI from returning to continue +/- 22nm from where they left off st 26S ?
@Ventus
Yes. The wreckage will not add anything to our understanding of the diversion.
Though the finding of the wreckage will bring closure to the families that nothing else can bring…
@DennisW
I see it potentially like SilkAir, the wreckage may give us circumstantial evidence (eg; flight data recorder turned off). It should still give us better understanding, if found.
@Victor Iannello,
Or perhaps the missing IFE log-on was due to the APU running out of fuel in the line after the left engine was fuel exhausted at 00:17:30 with the normal electrical configuration.
@DennisW,
You said: “That assumption generates two definite and uncontestable conclusions:
1> The aircraft flew South after the FMT.
2> The flight path ended with a very rapid descent.”
I will contest your second claim.
We don’t know from the Inmarsat data that the flight path “ended” with a very rapid descent. We do think we know a very rapid descent was underway at 00:19, but we don’t know when the flight path ended.
I would say that the Inmarsat data indicate a very rapid descent occurred near the end of the flight.
@Richard Godfrey,
Your latest results are quite interesting, especially the clustering of the Reunion matches on the left side of the assumed track. Your plot is an excellent way to present a lot of information on one page.
Can you do similar colored graphics for the Africa mainland and Madascar matches? Unless I am misinterpreting your plot, those points are not in the same locations as the Reunion matches. I noticed you only discussed the Reunion matches.
I also noticed that the predicted transit times don’t always vary smoothly across the grid. Would doing an intermediate grid of start points be helpful in minimizing the model “prediction noise”?
@Victor
You stated:
1. “Perhaps the pilot recovered from the high speed descent and glided some distance.”
At a ground speed of 450 knots, if you recover descent, for just 3 minutes, you go beyond 22 NM.
2. “Perhaps the missing IFE log-on was due to some unknown system configuration.”
We have already agreed that the APU fuel might have been exhausted preventing an IFE log-on.
3. “Perhaps there was a high energy impact after a controlled glide.”
Any kind of impact without engine power and without a certain amount of flaps is difficult. Ask Captain Robert Pearson of the Gimli glider without flaps or Chesley Sullenberger of the Hudson glider with flaps. However, neither had to deal with an ocean surface with waves and swell. Not impossible, but difficult.
I only concluded it is possible that the MH370 end point is outside the ±22 NM search width.
I am not making a recommendation at this stage. I am still investigating alternative scenarios, as to why the search to date did not find MH370.
As @Dennis points out, the mean time of the first 3 discoveries on mainland Africa is 679 days after 8th March 2014 and the arrival time could easily be around 640 days (the first discovery was “Roy” after 655 days, the second was “676EB” after 659 days, the third was “No Step” after 722 days) given that the second and third discovery were in remote locations. This fits a MH370 end point just north of 25°S.
Accounting for leeway on the Flaperon increasing the speed by 15%, then the arrival time in Reunion also fits a MH370 end point north of 25°S up to around 23°S (440 days +15% = 506 days).
At the moment, I do not think it is possible to exclude either further north or further out.
@DrB
My results graphic was perhaps overloaded.
For each of the 25 points, I give the days to Reunion with black text on a light grey background just above the point.
For each of the 25 points, I give the days to the drift end point just below the point. This can be either:
(1) Madagascar with white text on a mid grey background (16 cases).
(2) Mainland Africa with white text on a dark grey background (3 cases).
(3) Reunion with a (R) with white text on a dark grey background (1 case).
(4) Mid Ocean after 1,200 days with a (O) with white text on a dark grey background (5 cases).
I did not discuss the mainland Africa results as there were only 3 cases.
I did not discuss the Madagascar results, because of the 11 discoveries made by Blaine Gibson in Madagascar, only 5 are likely, highly likely or confirmed from MH370. These 5 were all found between 6th and 12th June 2016, but quite possibly arrived a long time beforehand. For Madagascar discoveries, it is not possible to associate an arrival time with any certainty. The items were found simply when Blaine scheduled a visit.
I am planning an intermediate grid to zoom in on the area of interest from the initial results.
@TBill
You asked “Do you feel the gyres near Broken Ridge would tend to take some debris to the Australia shores too?”
All the 25 simulation runs from around Broken Ridge headed initially in a north-westerly direction away from Australia, then in a westerly direction towards Madagascar.
There were 10 simulation runs, where gyres were strongly evident. This caused loops and deviations in the simulation track, but not in the initial north-westerly and generally westerly direction.
@DrB
We don’t know from the Inmarsat data that the flight path “ended” with a very rapid descent. We do think we know a very rapid descent was underway at 00:19, but we don’t know when the flight path ended.
True. I am guilty of adding my own conclusion to the data.
@DrB
You stated “I also noticed that the predicted transit times don’t always vary smoothly across the grid.”
This is true in general and especially around Broken Ridge.
Here are the results in tabular form in an Excel:
https://www.dropbox.com/s/kxxl7p227oqbulb/Array%2031.566S%2096.774E%20100%20NM.xlsx?dl=0
CORRECTION – my apologies.
I stated above: “Accounting for leeway on the Flaperon increasing the speed by 15%, then the arrival time in Reunion also fits a MH370 end point north of 25°S up to around 23°S (440 days +15% = 506 days).”
I should have stated: “Accounting for leeway on the Flaperon increasing the speed by 15%, then the arrival time in Reunion for a MH370 end point north of 25°S up to around 23°S is around 374 days (440 days – 15% = 374 days). This is 134 days before discovery.”
@DrB, @Richard G: Yes, it’s possible that the APU ran out of fuel, explaining the missing IFE log-on. Our best estimate is that it did not.
@DennisW, @DrB, @Richard G: Independent of simulation studies and the final BFO values, the shattered debris from the interior of the fuselage makes it probable that the aircraft impacted the water with high energy.
@Victor
“the shattered debris from the interior of the fuselage makes it probable that the aircraft impacted the water with high energy”
The question: with how high energy. I think there are not too many debris from the interior (there should be thousands of them). The fuselage could have disintegrated completely but more probably it only broke (maybe in several places – like Ethiopian Airlines Flight 961).
@Greg: I suspect there are a number of small fragments from the interior of the fuselage scattered along the shores of Madagascar. The success of Blaine Gibson and the searches he has organized is a testament to quantity of debris, and this success should not be attributed to luck. We are not aware of more debris because either people aren’t searching, or debris has been found and not reported.
Don’t forget that locals in Madagascar searching for debris have been threatened, and the death of the diplomat who was expected to retrieve debris from the Malagasy authorities and send it to Malaysia is still under investigation. The parts he was to collect (including the base of the vortex generator) have been retained by the Malagasy authorities as part of the investigation, and remain in Madagascar.
@David
You asked “Can you tell us whether your model supports the CSIRO assessment that a crash towards the bottom end of the previous search is ruled out by the absence of debris being discovered (ie being retained) on WA shores?”
In my view, the furthest south that we can expect to find MH370 is around 39.6°S, based on the maximum possible fuel range, earliest possible turn and maximum possible aircraft performance.
For a MH370 end point at 38°S on the 7th Arc, the drift simulation shows a track to central Madagascar in 651 days. The track makes a curve around Broken Ridge, but comes no where near Australia.
https://www.dropbox.com/s/iqngkx8u1t46ohy/Drift%20Analysis%2038.0S%20651%20Days%20Madagascar.png?dl=0
For a MH370 end point at 40°S on the 7th Arc, the drift simulation shows a track ending off south east Madagascar in 1200 days. The track also makes a curve around Broken Ridge but still comes no where near Australia.
https://www.dropbox.com/s/65ssxvyiqefu6bv/Drift%20Analysis%2040.0S%201200%20Days%20Mid%20Ocean.png?dl=0
For a MH370 end point at 42°S on the 7th Arc, the drift simulation shows a track ending off the northern tip of Madagascar in 1081 days. The track passes close to Australia, but does not actually beach. The simulation software detects when a track hits land.
https://www.dropbox.com/s/t3rk2r5odzkhltx/Drift%20Analysis%2042.0S%201081%20Days%20Madagascar.png?dl=0
@DennisW
You stated “…people elected to compute a mean and variance, which is certainly defined, but is nonetheless meaningless”
A more positive approach perhaps is to try to find useful statistics to estimate the BFO error margins. If the random walk is a good model for the oscillator drift then it should be possible to “confine” the expected BFO errors using a linearly increasing variance in time (you and @DrB were discussing something like this some time ago; see for example also
http://physics.gu.se/~frtbm/joomla/media/mydocs/LennartSjogren/kap2.pdf
). I’ve built a simple symmetric unit step random walk simulator and indeed the position is usually very well confined within 3SQRT(N), see for 20 subsequent realizations:
https://www.dropbox.com/s/tl73g355mkmeyod/RandomWalk.pdf?dl=0
So would it be possible to estimate the error margins by 2 or 3sigma where sqr(sigma)=sqr(sigma0) + alpha x time ?
The question then is how to define sigma0 and alpha. I’ve contacted DSTG for more details about the data from the 20 previous flights of 9M-MRO. If they are willing to share we could possible make some progress.
@Niels
One could certainly try to put bounds on the oscillator drift, but I believe they would be so large that BFO would not be very useful as a path qualifier i.e. see figure 4 of Bayesian Methods which is derived from the AES on board 9M-MRO.
A variance that grows with time is appropriate. I also thought the data from previous flights would be useful, however, making sense of it would require an awful lot of work (including different satellite parameters for the flights).
@Niels, @DennisW: In addition to the random noise and the random walk drift, you would have to include retrace error. And that assumes the cabin temperature was relatively constant. If there was a depressurization and change in temperature, that would contribute additional changes in the bias.
While I do not wish to disrupt a very interesting conversation, I would like to ask for clarifications on the “pings”.
I was looking at this Wikipedia (sorry) section on the pings: https://en.wikipedia.org/wiki/Analysis_of_Malaysia_Airlines_Flight_370_satellite_communications#Communications_from_Flight_370
How can the difference in times be explained?
Let me clarify this question: second to fifth handshakes are about an hour apart. Why is the sixth handshake an hour and a half from the previous one?
And by the same token, why is the second handshake not an hour from the first?
@Victor, @Dennis, @Barry,
I artificially increased the speed of the Flaperon’s track to Reunion by 15%.
The picture changes, but the hot spot only moves to 2 adjacent locations.
https://www.dropbox.com/s/19bz2otdz0a0foi/Array%20End%20Points%20%2B15%25.pdf?dl=0
@ArthurC: The call at 23:14 reset the one-hour timer. The next handshake occurred 57 minutes after the call, or about 3 minutes early, which is within the accuracy of the timer.
@Victor
Thank you very much.
In retrospect, that should have been obvious to me, as there is consistency there. Both “disruptions” occur after a ground initiated call at about one hour away.
I agree with Dennis et al that the wreckage will not reveal anything. I happen to be in the ZS pijacking camp (basically I fully subscribe to the Ed Baker Thesis) and as such, think that ZS planned this in such meticulous detail that he actually took Fariq’s hypoxic corpse from the other side of the locked cockpit door and repositioned it back in the FO’s seat. I also think he scoured the cabin for anything that would indicate crew or passengers trying to break down the cockpit door, e.g. beverage cart, and rearranged it to stage the scene. Even if some miracle happens in the year 2067 and we find the wreckage, we will never know what happened from the plane itself.
@Sunken Deal
I am totally with you this could be the ultimate deniable pijacking, but it still could be important to find the wreck to see what was apparently done, even if what was done was wiping the evidence.
@Victor Iannello,
You said: “@DrB, @Richard G: Yes, it’s possible that the APU ran out of fuel, explaining the missing IFE log-on. Our best estimate is that it did not.”
Specifically, who are you including in “our”?
How did you estimate it did not?
@Richard Godfrey,
You said relative to increasing the glaperon drift speed by 15%: “The picture changes, but the hot spot only moves to 2 adjacent locations.”
I see more changes if I am understanding your plots correctly.
First, four out of the original six hot spots (with < 20% error) now fall outside 20% error. Second, instead of all the hot spots being on the left side of the track, they are equally distributed left and right. Third, three new hot spots appear with fairly large track offset (and they are not adjacent locations). Fourth, initially two of six hot spots were within 25 NM of the 7th Arc. With the higher speed, only one out of five hot spots is within 25 NM.
I see large changes which I don’t know how to interpret. The mean position of the hot spots went south as I expected when you increased the drift speed. So far so good. Why did the cluster disperse more spatially with the higher speed? Were gyres more prevalent?
@Richard Godfrey. Thank you for your work on the southern-end crash question.
The CSIRO implies (8 Dec 2016 report) p25 that debris would reach WA were the crash site south of 39˚S (and also between 32-36˚S) so rather a different outcome to yours.
This CSIRO animation covers about 35 – 37 (2014, MP4): http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dp_arc7_3634/index.html
This with ‘NEXT’s illustrate crash sites coming up from further south:
http://www.marine.csiro.au/~griffin/MH370/tracks_nonflap/mapwent_tp3l1p2dp_arc7_4422_04_02.html
As to possible contributors to the apparent difference I note that you use the term ‘leeway’ as being the difference between flaperon velocity and undrogued drifters. However the undrogued have leeway without that, being their velocity through the water when added to that of surface water when compared to deeper, Stoke’s effect. The two are added by CSIRO in estimating the total effect of wind.
What that leads to is that the undrogued drifter velocity averages in your data base are wind dependent for this reason. A fair question then I think is how representative is the leeway they experience when compared to that of the MH370 items? That depends not just on the place and month but on actual winds doesn’t it?
Much the same might be said of gyres? An average current may not capture the chance encounters of individual items.
Related, earlier I have alluded to debris retention once landed. It is curious how in the the CSIRO animation below (2014, MP4, see magenta items) they seem not to aggregate. Winds with an easterly component or maybe the animation?
http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dp_arc7_4422/index.html
@all those interested in end-of-flight possibilities.
For my own use I have prepared a ready reference of configurations and SDU reboots they lead to. In case it is of use to others:
https://www.dropbox.com/s/buovl3dgql6rqvq/Aide%20Memoire%20SDU%20reboot%20scenarios.xlsx?dl=0
@VictorI, DennisW
Once a limit on the expected increase of the variance within a certain defined time interval is known my strategy would be to apply this to for example the 20:41 – 00:11 interval. Then assuming a certain control mode (CMT, CTT, great circle..) one could tune the 20:41 location to fit the variance within the expected bounds. Previous shifts / drift in the oscillator frequency would then show up as a certain mean BFO error. The accuracy of this method in terms of range of end-points depends indeed, as @DennisW expressed in his worries, on the possible oscillator drift expected in 4.5 hours.
@VictorI, as a first step it could be interesting to see what happens if you “split” your RMS BFO error shown in fig. 2 of your Oct 2017 posting (for the LRC paths), into a mean error and a variance. Would you still have the BFO errors on the arc crossings (and @ 23:14) for those paths?
@David
I am still working on capturing the data for the actual winds across the Southern Indian Ocean between 8th March 2014 and 29th July 2015. Once the actual wind database is complete, I will add it to the simulation.
Meanwhile, you are quite correct to point out that the historic undrogued drifter data, that I am using, contains an element of leeway and drift angle based on the historic wind data.
My terminology was imprecise. “There are no adjustments for leeway and drift angle.” “. I am not accounting for leeway and drift angle.”
I should have said: I do not account for leeway and drift angle based on the actual winds encountered on the track across the Southern Indian Ocean at the time, the actual windage profile of each item of MH370 floating debris and any changes in the windage profile of MH370 floating debris due to water logging or positional changes (leading edge changing to trailing edge, top changing to bottom). I only account for leeway and drift angle implicit in the historic undrogued drifter data and the windage profile of an undrogued drifter.
@DrB
I would put myself in the rapid descent at the end of flight camp i.e. the missing IFE was not due to the APU running out of fuel.
It is the simplest explanation that does not require a pilot recovery from what would have been a very questionable end of flight maneuver coupled with a timely fuel starvation event. It is very unlikely that a pilot would be aware of the Inmarsat data. Why a steep dive followed by a recovery and glide?
I still don’t understand: Why should Shah have performed the gliding after the rapid descent? Can we consider this as a prove that he knew about the pings? Why didn’t he glide from FL 330? Watering the plane would be possible in both scenarios.
What produces more debris, a high speed impact or a gliding?
@Dennis
You stated “I would put myself in the rapid descent at the end of flight camp.”
I have been in this camp for the last 4.5 years.
It has not helped to find MH370.
I think it is right to question all our assumptions.
@Gerald
ZS did not know about the pings!
Nobody outside of Inmarsat and the aircraft tracking satellite community knew about the pings.
Inmarsat’s revelation of the pings on and after 10th March 2014 came as a surprise to most everybody.
DrB asked: Specifically, who are you including in “our”?
There have been many comments here regarding the fuel available for the APU, notably by @David, @ALSM, and @Andrew, using data from the ATSB and other sources. In short, there was about 17 lb in the fuel line to the APU, and at a consumption rate of about 2.2 lb/min, the APU would run for about 7 or 8 minutes. That is considerably longer duration than the time required to log-on to the IFE after the SATCOM was re-powered at around 00:18.
If new insights or data is available to change those numbers, I am not aware of it.
@DennisW asked Why a steep dive followed by a recovery and glide?
And then followed by another steep descent ending in a high speed impact that produced the fragments from the interior of the fuselage.
@Gerald
I would say, a couple things-
Re: SATCOM Signal Awareness of Pilot
Even if the pilot did not know about the pings, he may have realized there must be a logon to the network. The pilot’s SATCOM understanding may have optionally been: (1) Full, (2) Partial, or (3) Total Ignorance. I would suggest at least partial understanding of SATCOM signals is not only possible, it may explain some of what we saw.
Re: End of Flight Descent/Glide
Why descend at 00:19? If it is active pilot, then there are a number of possible reasons:
(1) Climb to e.g. FL400 and descend when right engines fails. With one engine can only sustain around FL250, less if lower thrust.
(2) Get RAT power
(3) Intentionally sending end-of-flight message
(4) Pilot’s choice for various situational reasons (get below clouds, etc)
(5) Preparing flight surfaces for a powerless, extended, and stable glide, which will give time to prepare (for leaving cockpit etc.)
Loss of IFE Signal:
If we allow for active pilot trying hide evidence, then the explanation for loss of IFE at end of flight includes, beside loss of APU:
(1) Pilot turns off Left BUS/SDU again to prevent signals
(2) Pilot had pulled MEC Bay circuit breakers for IFE/Flight Data Recorder etc.
Since we have been mostly trying to say this was a passive flight/unmanned crash, it is a paradigm shift to try to think of all of this as intentional. But I would say that active flight is just as likely as passive flight at this juncture.
Sometimes I think Shah or the man in the cockpit knew about the pings, because this would bring sense into the EoF scenarios. Maybe he knew about the AF flight discussions where The possibility of Flight Data Recovery was discussed. https://www.bea.aero/uploads/tx_elyextendttnews/triggered.transmission.of.flight.data.pdf.
So in order to hide the plane it was necessary to make the EoF a riddle. The previous pings seemed not to be important to him and/ or he didn’t want to shut down the Left Bus for the whole flight. If things went that way, he did very well up till now.
@TBill: Didn’t read your post. This sums it up more elaborated
@TBill (you said)
Since we have been mostly trying to say this was a passive flight/unmanned crash, it is a paradigm shift to try to think of all of this as intentional. But I would say that active flight is just as likely as passive flight at this juncture.
@Richard (you said)
I think it is right to question all our assumptions.
Sure, a glide (active flight) is possible. And yes, questioning assumptions is a good thing.
However, my “solution filter” eliminates solutions which are not actionable. What good can possibly come from an extended glide hypothesis when it leads to an impractically large search area. DrB is in love with his latest solution, and the plane was not found close to the arc at his preferred latitude. So he would advocate seaching wider there. We should all know by now that no solution based on the Inmarsat data is worthy preferential treatment.
I categorically reject an extended glide hypothesis because it has no value relative to locating the aircraft, and it is not supported by any evidence whatever. In fact, there is evidence against it (no IFE logon).
@Dennis
You stated: ” What good can possibly come from an extended glide hypothesis when it leads to an impractically large search area.”
You are looking at the problem from an investigator’s point of view, not from a perpetrator’s point of view.
I do not believe that ZS was at all bothered, that an extended glide hypothesis might lead to an impractically large search area.
On the contrary, ZS may have considered, that an extended glide (perpetrator’s hypothesis) might lead to an impractically large search area, especially if he was interested in burying evidence, was a good thing.
@Dennis
You stated: “We should all know by now that no solution based on the Inmarsat data is worthy preferential treatment.”
I am currently repeating the grid analysis ± 50 NM and ± 100 NM around 24.0000°S 102.4793°E for exactly this reason.
@Richard
I don’t believe ZS had any idea that the Inmarsat AES could be used to track the aircraft. If that is true, the motivation for an extended glide cannot be attributed to a desire to generate an impractically large search area. That is another reason to reject the extended glide hypothesis.
@DennisW, @Richard G: If the captain was aware that Inmarsat date would give clues about the location of the aircraft AND he wished to hide the plane, he would have not allowed the SATCOM to power up and log-on.
@Victor
Yes, one could certainly reference that. I was reluctant to do so because I am still not convinced the SATCOM was powered down. The accuracy of the BFO value at 18:25 relative to what we believe the location, speed, and track of the aircraft was at that time is simply too accurate (maybe just plain luck?). However, I don’t have an alternate explanation for the 18:25 logon.
IMHO, a glide scenario would have been considered by ZH to reduce the debris, not to extend the search area. I believe that, regardless if he knew of the satcom capabilities, he would not have considered that factor.
@ArthurC
But why start the glide with a kamikaze like dive? Why not just glide down from the altitude at fuel exhaustion?
Sorry to appear to be beating this to death, but being in the glide camp is just plain untenable from any perspective, and I just cannot rationalize how people can cling to that scenario.
@Gerald, and any interested others.
The BEA Working Group Report concerning Triggered Transmission of Flight Data does not discuss the exploitation of BTO data, a.k.a. ‘pings’.
That report describes the usefulness of ACARS transmissions by a cohort of aircraft in the accident database.
9M-MRO did not initiate any ACARS messages over SATCOM after 1707Z. That aircraft systems may initiate ACARS messages after the diversion would have been well known, whether routine msgs such as the periodic FMS downlink reports or the CMCS reporting power loss of certain LRUs.
The SATCOM datalink and ISO-8208 SSN signalling was/is much less widely understood.
@ArthurC: If the captain wanted to reduce the debris, he would have performed an entry into the water with flaps down and with engine thrust available. This would minimize both the airspeed and the vertical speed at the time of contact with the water. However, likely flaps were up (based on the ATSB’s investigation of the flap) and one or more engines had flamed out (based on the SATCOM log-on at 00:19).
@Victor,
I understand that, the only explanation I could think of was literally, a steep dive, then a possible recovery into a glide, to match existing data.
Not likely, even insane. But the point I was trying to make was that the captain would probably have not thought of satcom implications or the future search area.
@Dennis Maybe the first log off was an attempt to go silent and that attempt failed, because flying with the isolated left bus for the next coming hours proved to be not a good idea. So the pilot logged on again.
Denying ideas just because they make the finding impossible or better to say more complicated isn’t productive.
Fact is that the unpiloted flying theory didn’t find the plane and thinking about a wider search area is a good idea. But this needs a longer gliding phase, obviously with somebody living.
By the way in 2014 a man like Shah who was an aviation man could have read lots of articles about locating a plane by the Inmarsat satellites. This was not a secret at all. Just look at my link above. It is a BEA paper from 2011.
@DennisW wrote “I am still not convinced the SATCOM was powered down”
I held out on that for a long time. Ultimately, the expanded ‘Stratos Log’ confirmed to me that the SDU had lost power. The SDU Log On Request defines, within a particular data field, either a valid ID for the previous satellite used or ‘no valid data’.
‘No valid data’ is recorded if the SDU had previously completed a Log Off (unnecessary to signal prev sat ID to the ground segment), or if the SDU lost power.
@Gerald: Please read @Don Thompson’s comment.
It is very, very unlikely that the captain would expect that Inmarsat could locate a plane using BTOs and BFOs and without ACARS working.
@Don Thompson: The BFO sequence at 18:25 also suggests the oscillator was thermally stabilizing after a power up following an extended period of no power.
@Don
Thx. How is an SDU Log Off accomplished without depowering?
@Gerald,
Denying well established detail isn’t productive.
You suggest that “Maybe the first log off was an attempt…”. No SATCOM Log Off was commanded from the aircraft subsequent to 9M-MRO’s departure from WMKK at 1635Z.
The GES reverted its record of 9M-MRO’s AES state to ‘logged off’ because the AES did not respond to a ground-to-air transmission initiated at 1803Z.
See my response just above concerning the 1825Z Log On and the AES-SDU state prior to that event.
There were no “lots of articles about locating a plane by the Inmarsat satellites>/em>”.
Refer, again, to my comment above: the Triggered Transmission of Flight Data report concerned ACARS transmission of flight data in response to a non-normal event, or series of such events. No ACARS messages, none, were initiated by 9M-MRO after 1707Z.
@Victor,
BFO & oscillator warm-up: yes, no doubt persists in my mind about that.
@DennisW,
Both the AES and the GES manage Log On state.
An AES may execute an explicit Log Off (as 9M-MRO’s did at 15:59:45Z) or, if it loses sync with the forward P-ch signal for >10sec, it assumes the Log Off state and attempts to renew its Log On with a GES.
A GES maintains a table of AES that are ‘logged on’. If the GES fails to correspond with an AES that is recorded in that table, it marks it as logged off.
The purpose of a long glide, among other things, could be getting to a specific end point target such as Broken Ridge. Another purpose could be preparations for whatever landing strategy the perpetrator might have been planning. Who knows what that was, but seems to me high speed glide, with intent to leave minimal evidence behind, is my current thought experiment.
As far as search implications, we would need a decent consensus, which may require another year or two and first searching 20-25 South to rule that out. Then I wonder if there could be some way to spot check wider areas versus saturation coverage of the sea floor.
@Victor/Don
BFO & oscillator warm-up: yes, no doubt persists in my mind about that.
We had extensive discussions about that when the Holland paper was made public. You might recall Mike and DrB generating elaborate 2nd order temperature control and corresponding oven response profiles.
Consistently ignored in Holland paper (and by Mike and DrB) is the fact that the 18:25 log-on was unique among all the cases presented. It was characterized by an accurate BFO value at the first response time whereas all other log-ons started with a large offset followed by a settling period.
@Don
My question is how does an SDU execute what you are calling an “explicit” log-off.
@DennisW
Just as there is a Log On Request SU (type 10), there is a Log Off Request (type 12).
During the prior MH371 service the AES explicitly commanded Log Off 11 times as it swapped service between POR & IOR.
@Don
Yes, swapping satellites is automatic in flight. I would not charcterize it as explicit (which I am perhaps confusing with deliberate). When the aircraft parks at the gate, I suppose removal of power constitutes the log-off? Likewise when power is applied at the gate?
@Victor
You stated: “@DennisW, @Richard G: If the captain was aware that Inmarsat date would give clues about the location of the aircraft AND he wished to hide the plane, he would have not allowed the SATCOM to power up and log-on.”
I agree!
As I have said before: “ZS did not know about the Inmarsat satellite data.”
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-18979
To be clear, my view is:
ZS was not aware that the Inmarsat data would give clues about the final location of MH370.
ZS was not at all bothered to hide the plane, other than dumping it in the middle of nowhere.
ZS was not thinking about a subsequent search for MH370.
ZS did not consider a subsequent search strategy or how large a search area might become.
ZS was not bothered that investigators would subsequently spend over 4 years trying to solve the mystery.
ZS just planned to dump MH370 in the Southern Indian Ocean in the middle of nowhere.
Why else do you simulate such a flight on a home computer and then try to delete all trace?
@DennisW wrote “I suppose removal of power constitutes the log-off? Likewise when power is applied at the gate?”
Removal of power from the AES doesn’t constitute a Log Off on the part of the GES. The GES will only record that AES as logged off when no response ensues from the Log On Interrogation initiated as the idle timer expires (set for approx 60min at the time of 9M-MROs loss).
@DennisW said: It was characterized by an accurate BFO value at the first response time whereas all other log-ons started with a large offset followed by a settling period.
The first BFO value at 18:25 shows evidence that the measurement was corrupted by a packet collision from a transmission from two AES sources. For this reason, most of us ignore it. The remaining BFO values (which show no evidence of a collision) are consistent with a asymptotic decay to the final value, just like the other power up cases presented by Holland.
@Don
Thx. Got it.
@Victor
Yes, I recall all that. I suppose I should attribute the accurate BFO value to a random result of a number of coincidences. I have a hard time with that, but you are probably correct.
@TBill: You said “The purpose of a long glide, among other things, could be getting to a specific end point target such as Broken Ridge”
If there is a behavioral science expert to evaluate some of the “Why’s” that don’t make sense based on solid physics and Math, that might help as it is highly probable that his attachment to his daughter and intent to settle near her has something to do with the “End of Flight” location.
Geelong at 38S could have been on his mind and mentally getting as close to that as feasible without having her know that he was responsible for the loss of the flight. It is only a theory but it is supported by possible intent considering he made a flight to Melbourne before this final flight.
@Richard Godfrey. Thanks for your advice that your model predicts no debris on WA shores at the 7th arc’s southern extreme and of your intention to add wind to it.
@Victor. You said to Dr B, “….there was about 17 lb in the fuel line to the APU, and at a consumption rate of about 2.2 lb/min, the APU would run for about 7 or 8 minutes.”
That was based on an internal diameter estimated by @Andrew as 20mm. Later he adjusted his estimate to 10mm, reducing the 17 by three quarters. At 2.2 lb/min of APU running that amount would be marginal for 2 mins running.
However as usual, that depends. There is an estimated 30 lbs of residual fuel in the left tank with the aircraft at 1˚ nose up. Andrew has described how the aircraft, slowed at APU fuel pump start (at left engine fuel exhaustion), would be slightly more pitched up than that but with the APU fuel pump suction being at the left tank’s rear that should not cause a reduction. However during the manoeuvres leading to the second transmission at 2 mins 8 secs that fuel may have sloshed about, or forward.
Complicating this, the left engine could relight, tapping into this residual fuel via the APU fuel pump supply. Relight or not it could have consumed it or some of it.
This is all analysed in Andrew’s thorough paper here:
https://www.dropbox.com/s/ynyr3i5dwo38dml/MH370%20ENG_APU_Start.pdf?dl=0
(I add that the yaw from a relight could bring forward the timing of the descent and increase its acceleration, the descent being too late in Boeing relight-free simulations and the acceleration too little in the simulation Andrew describes.)
Also, the engine fuel supply manifold most likely would have air in it and that would need to be replenished.
As if that were not getting complicated enough I have written a paper on APU fuel consumption estimation. It is too detailed to reference here. Because there are no reliable figures for APU fuel consumption at altitude, ie without intake drag being included in the gross, I analysed test bed results and made adjustments for aircraft installation, then converted that to an approximate estimate at altitude.
The ATSB has not described the basis for its estimate, including at what altitude it applied and under what loading. There is no APU fuel flowmeter in the 777. It could even be an average over a 2 mins start and run (which Andrew points out includes intake opening time). My figures, from more than one source, were a deal higher but supposed an electrical loading which could well be different to the ATSB’s. As to altitude, if the aircraft had been lower than 22,000 ft during SDU reboot and second transmission, APU pneumatic air delivery would cut in to replenish cabin pressurisation and ventilation, loading it more. (The normal engine bleed supply would have ceased at fuel exhaustion.)
Finally I drew attention to the risk of vapour lock if the APU became dependent on fuel just in the line, its fuel pump having lost suction; or if the engine demand exceeded the APU fuel pump’s limited delivery rate and reduced its delivery pressure thereby.
A controversial topic as you will remember.
@ST
You are correct that, if this was an intentional flight path, we do not know what the strategy was. And I agree too, what we seem to be missing in the search planning/flight path guesses, is FBI-style criminal intent analysis of what might have been done. I would cite the example of the Egypt Air 990 apparent suicde, where the NTSB wanted to turn over the case to the FBI, but Egypt refused to do that, which was apparently Egypt’s choice via Interational law.
However, long ago I personally adopted the Broken Ridge crash site hypothesis, which I took out of Peter Lee’s book, from a comment by a NOAA retiree. Originally the only reason I adopted the BR hypothesis was to help me comprehend Arc7, but also I felt it was intentional pijacking.
So what I am saying, that the BR hypothesis is still working. The pilot’s simulator flight flies over BR, and the BTO/BFO and drift studies seem like they could be consistent with the fact the pilot simply flew MH370 on the flight path shown in the simulator cases.
If you have another hypothesis, I am all ears, but I will need to feel better about your proposal, than I do mine.
@David: Thank you for reminding me of that discussion. Perhaps we can’t say with any reasonable degree of confidence whether or not the APU running out of fuel was responsible for the missing IFE log-on. That said, if the APU ran out of fuel, so did the left engine, so scenarios in which the left engine continued to produce thrust for some time after the final SATCOM transmission are unlikely.
We still have the two BFO values which suggest an increasingly steep descent. With no pilot inputs, that still suggests an impact close to the 7th arc.
@Victor. My view is that whether the APU would run for 2 mins is also unconfirmed.
There is also the issue as to whether a brief relight would bring forward the BFO related descents in the Boeing simulations to about the right timing, together with the possibility that a piloted initial dive was at perceived onset of depressurisation after engine bleed air was lost.
@Victor. Standby.
https://variety.com/2018/tv/global/spiral-co-director-set-to-write-helm-mini-series-about-missing-malaysia-flight-1202982644/
@David: I suspect it won’t be helpful to solving the mystery.
@Richard Godfrey. David Griffin has found that updating his Fig3 was easier than he had anticipated and has now posted the result under his 18th October entry. Besides updating he now includes observed beachings as a comparison and has removed some bugs in the old one.
http://www.marine.csiro.au/~griffin/MH370/
@Richard
The primary lesson, an old one, repeated, and driven home, very savagely, yet again, from what became known as the battle of Savo Island (Naval Action) in WW2, was:
Never assume what your enemy/opponent –
(a) does not know
(b) can not know,
(c) can not do
(d) will not do.
Your list of what Z did not know, might well be a repeat of those mistakes.
DRAFT POST 18th Oct 2018
SUBJECT: The significance of the final two simulator points near 45S 104E.
I propose that the final two locations in the simulation was a “simple” planning reference tool.
The mission was to begin by going dark, at Igari (approx 7N 104E).
He needed to know:
(a) how much time was there to sunrise from diversion to 45S, and (most importantly)
(b) how much time would he have from just before sunrise at TOD (top of descent), to sunrise at circuit height, BOD (bottom of descent) at 4,000 feet.
Choosing 45S is significant for reaching the circumpolar current in an attempt to hide any debris, and choosing 104E is significant for two reasons:-
(a) it is about the same longitude as Igari, and
(b) it is also in his time zone (since, simplistically, he only need to know required endurance from Igari).
Plug the figures in for day, date, location, and the altitudes (37,651 ft and 4,000 ft) for Malaysian Time = (UTC+8).
Assume the simulation was done for 8th January 2014, what you get is:
For 8th January 2014
Sunrise_37,651(feet)_11,479(metres) = 05:01am = 21:01utc
Sunrise_4,000(feet)_1,220(metres) = 05:19am = 21:19utc
Sunrise at cruise altitude is 18 minutes before circuit height.
Assume the simulation was done for 8th February 2014, what you get is:
For 8th February 2014
Sunrise_37,651(feet)_11,479(metres) = 05:46am = 21:46utc
Sunrise_4,000(feet)_1,220(metres) = 06:02am = 22:02utc
Sunrise at cruise altitude is 16 minutes before circuit height.
Assume the simulation was done for (mission day) 8th March 2014, what you get is:
For 8th March 2014
Sunrise_37,651(feet)_11,479(metres) = 06:28am = 22:28utc
Sunrise_4,000(feet)_1,220(metres) = 06:42am = 22:42utc
Sunrise at cruise altitude is 14 minutes before circuit height.
In other words, the differences between sunrise at cruise altitude and sunrise at circuit height (at 45s) are 18 minutes, 16 minutes and 14 minutes respectively, ie, as the months roll through the southern hemisphere autumn, the time for descent (to remain in the dark, “under the rising sun”, ie, descending in twilight) is slowly decreasing by 2 minutes per month.
(Note: the actual longitude is irrelevant for time for descant).
In any case, it is clear that a descent rate of at least 2,500 feet per minute would be required from TOD (top of descent) to 4,000 ft to arrive at 4,000 ft just before sunrise.
The actual time of sunrise poses an interesting question with regard to longitude. I said above that the simulation was a “simple” planning tool. Taken at face value, 104E is “convenient” (as stated above), simply in as much as it shows how much time is “available” from the “start point”, ie, Igari. That “time available”, for “mission day”, is 5 hours and 7 minutes ( Igari 17:21utc to TOD 104E 22:28utc). However, he is not flying direct, and he has much more than 5 hours fuel from Igari.
The problem now is, after the diversion across Malaysia and onward to west of Sumatra, some of that time and fuel is used up. What he really needs to know are his options when he crosses the equator (near 95E).
Available time to sunrise at 00:00utc (and fuel remaining) will limit how far he can get. It turns out that time becomes the only factor, since there is fuel for a bit “beyond 00:00utc”.
So, if he crosses the equator a little before 20:00utc, he has 4 hours to fly south. He can average 480knGS, so he can fly a little over 1900nm. There is no way he can reach 45S.
On “mission day” he has to now pick a lat/lon that will be reachable, ie, within the 1,900nm arc from 95E, at 00:00utc.
That location (consistent with proximity to the 00:11utc 6th arc), turns out to be directly due south, at 95E 31.8E, right on top of Broken Ridge.
He passes his sun optimal planned TOD at 00:00utc and does NOT commence descent yet. He is “driven” to get as far south as possible, and he is in high cloud, and he still has some fuel for another few minintes.
After first engine flame out (right engine) he would begin to descend to maintain “speed” for “distance over the ground”. That descent rate would be low, and the flight path angle shallow, ie, it would be less than normal idle descent from TOD.
He crosses the 6th arc (at 00:11utc) still in a shallow cruise descent on one (left) engine at 95E (right engine had already flamed out). He keeps going.
Soon after left engine flame out (approx 00:17utc), he deliberately “pushed over” into a very rapid descent, crossed the 7th arc (00:19utc) during this rapid descent (probably between 20,000 and 15,000 feet).
Why did he do that ?
Everyone seems to think it is crazy.
Perhaps not.
Consider this.
Taking the recorded 00:19utc BFO’s at face value, I propose that the very rapid descent that they indicate, was in fact deliberate, and under pilot control, because, he had deliberately made the decision to delay descent (to get further south, remembering that every minute is another 8 nautical miles south) later than he should have (in other words, after sunrise at cruise altitude at 00:00utc) because he felt that he was still safe from detection, because he was still hidden from visual detection from satellites or other observers, by still being in high cloud.
But, he knew that every minute that he delayed descent, meant that he would have to descend faster, if he was to still reach 4,000 feet just as dawn broke.
But, as it turned out, he had a little more fuel than had thought he would have in his planning. He deliberately used it to “keep going” instead of following his plan to meet dawn at 4,000 feet at 00:14utc. That time has now passed. It is 00:17utc and now the left engine flames out.
He has gone as far south as he can, but now it is daylight below, and that is now his worry. He is now 3 minutes “late” on his planned BOD, and he still high. He has to get down “in a hurry”.
He begins descent with a prudently moderate “push over” into rapid descent, and with provision for a prudently safe recovery from the dive in mind, he would aim at accelerating to a target descent rate, (in mid descent), of around 15,000 feet per minute, (which is consistent with the recorded BFO’s). He would commence recovery at about 10,000 feet.
With the high speed passing 10,000 at the beginning of the recovery, the pull out would bottom out at around 5,000 feet, and with a partial zoom to convert speed to height (would take a minute or so), such that he is now at about 8,000 feet and speed has reduced to 200kias, and he pushes over in clean glide configuration, to prepare to ditch.
Thus, I propose that the overly simplistic (and politically convenient) “ghost flight with spiral dive” scenarios be dismissed, (along with the “no IFE log on” proves it mindset). I don’t place much credence in the lack of IFE log on at all. There could be reasons for it already discussed, and there could be others not yet thought of. To elevate it to the level of a slam dunk “got-ya”, to “prove” an unpiloted EOF, is, in my view, very unwise.
So I have to disagree with DennisW and all others who place such faith in the lack of the IFE log-on, and also those who dismiss a steep dive as insane, both thus supposedly “evidence” of an unpiloted EOF. (Try convincing a coroner of that).
I suppose that means that I disagree with just about everyone here on the EOF.
In conclusion, I contend, that the various “piloted dive and recover” scenarios are “sunrise dependent” and credible. They should be studied more closely, since, in my view, they have a sufficiently reasonable basis for serious consideration.
Moreover, and most particularly, anyone with an open mind would have to admit, that the search so far, has dismally failed to find the aircraft, in the (supposedly certain) +/- 22nm “death dive zone”. If the search has been thorough, and if we are 99.9% confident of that, then I will concede, that finishing the “death dive zone” from 25S up to 20S, is justifiable, on the existing logic, so let us hope that OI returns to do that.
But if OI does so, and it is still not found, then it is near enough to certain that it is not anywhere in the “death dive zone”, so the current EOF logic that says it is there, is wrong. Other EOF logic will be required to find it.
@Ventus45
The only other poster I can recall who had a fixation with the sunrise is Rob.
@David (F.): That Fig 3 is really a very nice way to present the data. If the timing of the discovery of debris is indicative of when the debris arrived, a search to the north is not warranted, as David G suggests. In fact, the predicted drift from a latitude of 22S provides just about the worst match to the timing of the discoveries.
However…although it is convenient to represent all the beached debris collectively, the time to reach Reunion Island is different than the time to reach Mossel Bay. Interestingly, some of the debris that was found the earliest (“Roy”, the flap) should have taken the longest to beach due to the southerly latitude. This may be due to the significant delay between a beaching and discovery for most of the debris that landed further north. For instance, the debris findings in Madagascar were really precipitated by the efforts of Blaine Gibson to organize search efforts and inform the locals. The “Blaine effect” is an externality that is not included in the drift models.
So, I agree with David G’s conclusions if we accept the qualifications he makes. However, those are important qualifications, and are debatable.
@Ventus45: Thank you for your comment. However, in the future, comments of that length are better suited as document that you can link to here.
You describe the captain’s simulator as a tool that was used for planning purposes. However, it would actually be quite a poor tool compared to other resources that would be available to him that are easier to use and much more accurate (e.g., Google Earth, skyvector, fuel planning tools). In fact, the way that the user manually moved the plane and modified the fuel level would indicate that there was no attempt to use the simulator for time studies or fuel planning.
As for a long controlled glide after a rapid descent, it is difficult to prove that it did not occur. The coordination with sunrise is also possible, but I don’t think we can mount a search based on this conjecture. In fact, Ed Baker has a scenario that includes planning for a prayer at sunrise, yet his prediction is an impact near 22S latitude, so even if we assume the flight path is related to the sunrise, it still might be difficult to estimate an impact location.
You said: If the search has been thorough, and if we are 99.9% confident of that, then I will concede, that finishing the “death dive zone” from 25S up to 20S, is justifiable, on the existing logic, so let us hope that OI returns to do that.
So, I think you are advocating that we start with a search to the north unless we have reason to believe the debris field was previously scanned and not identified.
@David,
Would you be able to get David G. to explain how he/they have applied the leeway due to wind in respect of the Flaperon. The SE trade winds in the SIO can create short steep seas mounted on the the underlying long period predominate swell. The surface wind speed between 1000 and 1600 LT can be in the range of 15 ~ 20 knots, and sometimes more; least overnight.
The day-time conditions could become problematic when determining the leeway angle and applied amount. Without detailed sea surface conditions and winds for the Flaperon’s assumed position(s), its leeway due to wind will only be a ‘lottery’. The Stokes drift component is likely to be only a small percentage of the total leeway, with the windage vector / distance the big unknown.
The Stormy Bay tests, were a bath-tub exercise when compared to the reality of the SIO.
@David,
Thanks for refreshing Victor’s memory on the fuel available to the APU. You beat me to it and did a very thorough job. My conclusion was, and still is, that the APU could have run anywhere from 2-9 minutes. We simply cannot say with any certainty that it would or would not have run out of fuel by 00:21:06 (when the first IFE transmission would have occurred) in the normal electrical configuration.
You also said: “. . . with the possibility that a piloted initial dive was at perceived onset of depressurisation after engine bleed air was lost.“
That is an interesting and new idea to me. Would the aircraft depressurize that quickly while descending from cruise altitude when the second engine failed? Why not just use the supplemental O2 mask?
Another novel idea is from TBill: the pilot might have shut off the IFE CB in the MEC. In my last paper on the FMT Route I suggested that ZS may have shut off the data recorder CB in the MEC during a HOLD. However, I don’t see why he would also shut off the IFE. What is to be gained?
@Victor
Re: “So, I think you are advocating that we start with a search to the north unless we have reason to believe the debris field was previously scanned and not identified.”
The search so far has been predicated on the unpiloted EOF theory.
The apparent agreed bounds of that theory are +/-22nm between 38 and 20 south.
The combined Fugro and OI searches have only so far covered from 38 to 26 south.
Therefore, I advocate that the unpiloted EOF theory should be “prosecuted to exhaustion”, by:-
(a) completing the remaining search area from 26 to 20 south, so as to have covered the entire area specified by the unpiloted EOF theory, and
(b) if not found between 26 and 20 south, to carefully revisit the Fugro data holiday areas and surrounds, and any similar areas that may exist in the OI search to date.
(c) if still not found, the whole search data set might have to be reexamined, perhaps by new more powerful software algorithms or AI of some kind.
The reason I so advocate, is that the unpiloted EOF theory is either shown to be correct by finding it, or if not found, we must be “sure” that it has not been missed, before it can be deemed incorrect.
So long as any doubt that it may have been missed remains, it will be impossible to shift those so far committed to the unpiloted EOF theory, to any other theory.
What level of confidence equals “sure” ?
That is an open question at this point, particularly when it comes to searching in “difficult” terrain, and needs to be carefully explained and justified by those in that line of work.
Unfortunately, due to the commercial sensitivities and implications for the selling of their services to commercial interests in other sub-surface projects, I don’t think they are likely to oblige us by being very forthcoming on these issues.
I fear it will remain a grey area, and thus some residual level of doubt will remain for those committed to the unpiloted EOF theory, thus stalling any serious consideration of any other theory.
If you have been following the so far frustratingly unsuccessful OI search for the ARA San Juan, you might begin to wonder if the ocean floor will ever give up it’s secrets easily.
@VictorI
The June 2016 peak in David G.’s new fig. 3 (items reported) seems to be part of the “Blaine effect”, so yes only based on this the dotted lines should probably already move to the left if they were to represent percentile of beached items. However, to have 50% of items beached in Oct. 2015..that’s quite a stretch.
Difficult issue though. I’ve looked at the items containing barnacles (“Roy” (Mossel Bay) and Right Outboard Flap (Pemba)) as @RichardG suggested before. Their number is too small. Where should you “place them”, close to the 10, 50, or 90 percentile line?
@Niels: I wish I had better answers. For some time I’ve been wanting to make a recommendation to OI. I have not because I don’t have a recommendation with a reasonable chance of success.
Ventus45 said: If you have been following the so far frustratingly unsuccessful OI search for the ARA San Juan, you might begin to wonder if the ocean floor will ever give up it’s secrets easily.
I have only followed it casually, but it is sobering that even armed with a last known time and position of reasonable accuracy, and even with the speed limitation of a submarine, the submarine has not yet been found.
I don’t know if the “A-team” is working on that one. That may be part of the problem, but I am not speaking from knowledge.
@Niels
There can be no “Blaine Gibson” effect on mainland Africa, because only 1 out of 13 MH370 debris items were found by Blaine.
The vast majority of Blaine’s discoveries were on Madagascar, which obviously does not belong to mainland Africa.
If David Griffin counts Madagascar as mainland Africa, then … I am lost for words.
@DrB
Re: IFE CB
I am working on a paper I call “How to Conduct a Deniable PiJacking”.
One possible element is to manage the Flight Computers such that they do not contain any incriminating info on the memory chips. I am fixated on the IFE flight map which would be presumably fed data from the IFE Computer. So that is why I think pulling various circuit breakers possibly gets involved in this accident.
David has an interesting point on bleed air. The pilot could use O2 but if the aircraft is FL400 that is hard. Also in a deniable pijacking the pilot is trying to manage cockpit appearance by leaving the cockpit after the glide is stable.
@Victor, @Ventus45,
Ventus45 stated: “If you have been following the so far frustratingly unsuccessful OI search for the ARA San Juan, you might begin to wonder if the ocean floor will ever give up it’s secrets easily.”
OI stated that they planned 100 days to find the San Juan.
So far they have spent 37 days in the search area.
https://www.dropbox.com/s/55op830ev02qqat/San%20Juan%20Argentina.png?dl=0
@Ventus45
Returning to my thoughts near the very beginning of this exercise (since the release of the Inmarsat data) it made sense to me to postulate a diversion for a reason. No suicide. No aircraft failure. A diversion with an associated negotiation. Crashing the aircraft into the SIO was not the preferred outcome. It was the execution of the “or else” associated with a negotiation.
For that reason a flight path with an end point of last resort seemed the most logical. That is why I proposed Christmas Island (long before Jean-Luc et. al.). The Cocos could also fulfill that role.
Postulating a route based on sunrise (your latest soft landing scenario) or a route based on difficult underwater terrain (a recurring TBill theme) assumes a motive to make the aircraft difficult to find. Why would a hijacker care about that? I have heard no even remotely plausible reason for wanting to hide the wreckage. Likewise a DrB route based on optimizing the flight mode and Inmarsat data is devoid of any form of causality. It is reminiscent of the early attitudes of Duncan Steel and Richard Cole who refused to entertain any form of underlying motive.
My search stategy would be to focus on the Victor Cocos scenario end point, and the Jean-Luc et. al. end point near Christmas Island.
@DennisW
I look at the Cocos/Xmas proposals as either real, or possibly a subterfuge. But it is possible to approach Cocos via BEBIM to make it look like that was a plan, and still make Broken Ridge as your “or else”.
@TBill
A subterfuge? Please explain. Also please provide your reason for thinking the hijacker had a “hide the wreckage” motive.
@Richard Godfrey
I was looking at (new) figure 3, which includes beaching from E21 to E55 and 29 items, so that seems to include Madagascar. I don’t think it mentions “mainland Africa”, it includes the non-flaperon items.
This is indeed relevant for the interpretation of the percentile lines.
@Niels
The problem with including Madagascar is, that all the items deemed likely or certain from MH370 were all found within one week in June 2016 by Blaine Gibson. The discovery date was determined by Blaine’s schedule. The relation to an arrival date cannot be ascertained.
I have therefore discounted the discoveries in Madagascar from the statistical analysis of the drift studies.
There are not 29 items of floating debris confirmed or likely from MH370, even including Madagascar.
It may be a deniable hijacking so subterfuge to leave word behind that this was not a suicide mission, even though it may have been, as well as getting advance assistance for peaceful diversion mission, and hiding the black box. Many are saying without the black box, we must assume innocence, or if the black box is wiped, same thing. That logic does fly for most Americans due to the Duck rule, but I do not think the Duck rule applies in Malaysia.
@DennisW above to you
@VictorI
It is indeed difficult to confine the length along the 7th arc to less than say 500 km. Adding the problem of understanding the final BFOs in combination with the “non-detection” in previous searches, I’m afraid I agree with you: this is not good enough for a search recommendation. I can at this moment only recommend to carefully check the “coverage” degree of previous searches, perhaps with emphasis on S29 -S34.
@Richard Godfrey
Yes I think we agree there is a “Blaine Gibson” effect on the “find statistics” if you include Madagascar. Good point regarding the “29 items”. I don’t know.. The 30-04-2017 list contains 27 items with several “unidentifiable” items.
@TBill
No “official” effort has been expended on a Cocos or a CI scenario. If any subterfuge was involved it did not work.
The new trend (in the US at least) is guilty until proven innocent. The black box is needed to prove ZS’s innocence not his guilt.
@DennisW
Going back to the beginning, and revisiting our earliest thoughts, as you said, returns us to reconsideration the early “plan B” or “or else” scenarios.
If I recall, back then, many of us had basically similar ideas, but with different intended landing spots. For example, if you recall, I spent a lot of time working on Barrow Island.
However, by late 2015, I had pretty much decided, that we were all “barking up the wrong tree”.
I decided that Z knew for certain that no negotiation, with what he saw as a fundamentally corrupt regime, that had been in power for decades, and would remain so, would ever succeed.
He had decided that “democracy is dead”, and that regime change could only be precipitated by decisive action.
That is why he had planned the mission in the first place.
He had decided that he would be that “decisive action”, and if you look at subsequent (and ongoing events) in Malaysia, it is possible that the historians may eventually verify that he was the catalyst that got the ball rolling.
Thus, I decided, that in his mind, there never was a “Plan B” at all. What we initially thought may have been his “Plan B”, was in fact his “Plan A”, from the outset.
There may have been, and most probably was, some sort of “negotiation scenario” in play initially (as “a ruse”), to buy him the time he needed for the “escape” from Malaysian Airspace. How else do you explain the MY government’s total lack of initial response, that “allowed” him to fly back to Penang without “any” RMAF reaction, and the subsequent actions of the government in the first few days ?
So by late 2015, I had discounted that he had any intent to land on any runway ever again, and I also decided, that he also had to ensure his safety after Penang, just in case the government decided to intercept him, and try to sell it to the world, as an Asian nine eleven scenario.
Thus, my rejection of the Malacca Strait flight path, and the genesis of the escape to the safety (from perhaps a belated attempt at interception by the RMAF) of Indonesian Airspace, by immediately continuing on to the nearest point of Sumatra, and thus the development of the via Medan theory.
All that lead me to decide, that since he never intended landing, he must have meticulously planned the terminus end game deep in the SIo, and, in my mind, that leaves only one sensible scenario, to ditch at dawn, as far away as possible.
Therefore, I still personally hold to the theory I published on AunntyPru, way back.
slight correction, 4th last paragraph: INSERT “NEVER”
So by late 2015, I had discounted that he “NEVER” had any intent to land on any runway ever again, and I also decided, that he also had to ensure his safety after Penang, just in case the government decided to intercept him, and try to sell it to the world, as an Asian nine eleven scenario.
I will try that again:
So by late 2015, I had DECIDED that he NEVER had any intent to land on any runway ever again, and I also decided, that he also had to ensure his safety after Penang, just in case the government decided to intercept him, and try to sell it to the world, as an Asian nine eleven scenario.
@Ventus45
Good points. You have taken it to a level beyond my own thinking.
@DennisW,
You said: “Likewise a DrB route based on optimizing the flight mode and Inmarsat data is devoid of any form of causality.”
If I were to be picky, I would say this is not true. The 180.0 True/181.2 Magnetic Route I have previously identified as a candidate for MH370 has obvious “causality”. It is created by flying south from BEDAX toward BEBIM, which is at 180 degrees true. If a pilot wanted to keep flying south, the simplest way to do this is to set TRACK HOLD, which thereafter maintains a 181-degree magnetic track, since the default reference switch setting is NORM.
Thus, assuming my proposed route is correct, the “causality” is clear and obvious – fly “south” until fuel exhaustion with a crash in an indeterminable location (with ACARS off) in the middle of nowhere, which would make the plane unfindable in ZS’ mind in the relatively near term.
What is less clear is “motivation”, and perhaps that is what you meant to express (instead of “causality”). If the motive was to embarrass the then-current government, that goal has so far been only minimally achieved. In my opinion ZS could have done a number of things to make his point in more spectacular fashion with greater effectiveness. He could have flown into the Petronas Towers. He could have sent a manifesto to the media. Such actions would have made him a “public” (not a “private”) martyr. I am not a student of Eastern thought on martyrdom and suicide, but others have indicated that suicide, even in martyrdom, would bring dishonor on his family. Perhaps ZS thought he could do both: embarrass the government by disappearing an airliner full of passengers and avoid being proven to be the perpetrator (achieving plausible deniability), thereby avoiding the stigma of suicide on his family. He may also have correctly surmised that the government would be reluctant to blame him because of the ensuing increased financial liability and embarrassment; the lack of enthusiasm on the part of the Malaysia MOT for this search is evident. ZS has succeeded in demonstrating some ineptitude on the part of the airline/MOT/government while simultaneously avoiding being blamed by that government. In that sense he also proved he was the smartest guy in the room, and maybe ego also played a role in his decision-making.
A planned, secret “negotiation” would have several drawbacks, including no control of the post-crash narrative in which he might be immediately blamed and characterized simply as a crazed individual, a criminal, and a mass murderer. In addition, he could not escape blame even if the “negotiation” were initially “successful”.
@DrB
Thanks for the feedback. It is useful. We are all on the same page here. Like Victor, I don’t feel strongly about what to do next.
Gentlemen, thank you for the many months of fascinating reading. Following the comments and discussions on this site is incredibly informative and enlightening.
I would like to enquire as to your knowledge on satellite photography. I distinctly remember in the days following the crash, some websites were established to “crowd-source” analysing satellite imagery of the SIO. I remember in fact spending hours and weeks looking myself.
My query is, surely some of the big companies or governments were funding some high resolution imagery of the general area of ocean along the arc in the days and weeks following the crash, and indeed there may have been imagery from around that time regardless. How or why have we got little to no photographic evidence of debris on the ocean?
@Ventus45 said: Thus, my rejection of the Malacca Strait flight path, and the genesis of the escape to the safety (from perhaps a belated attempt at interception by the RMAF) of Indonesian Airspace, by immediately continuing on to the nearest point of Sumatra, and thus the development of the via Medan theory.
The civilian radar data has the plane rounding Penang Island and turning northwest up the Malacca Strait. Even if you don’t trust the military radar data, a track towards Medan is not consistent with the civilian radar data.
@Ventus45
I would be interested in your thoughts relative to Shah’s relationship with Tim Pardi.
@Victor. About your earlier, “I agree with David G’s conclusions if we accept the qualifications he makes. However, those are important qualifications, and are debatable”, and latterly about making a recommendation to OI, here I dwell on the case for a new search up north.
To me a recommendation for that would need to include a good reason why David Griffin’s now-reinforced conclusion should not prevail.
To me the only potentially serious weakness in that conclusion is the supposition that MH370 items will drift at the unbuoyed drifter rate, ie 1.2% of wind speed, and direction, ie downwind.
First a related digression. Noting some comments from @Richard Godfrey and @Neils I do not know what extra items he has included in his 29 but his use of the term “reported” on the left of the arrivals rate panel below in the latest Fig 3 is deliberate.
Earlier I did post the 50% and 90% arrival times of the 18 items that beached in Africa and which have been assessed by Malaysia as at least “likely” from MH370, appended to the David Griffin’s original Fig3. I have taken the liberty of transferring the same to his later version:
https://www.dropbox.com/s/nmock7ltz9zxf8f/CSIRO%20FIG%203%20AMENDED%2C%20ILLUSTRATING%20%27LIKELY%27%2018.jpg?dl=0
i can say that both the 29 and 18 include Madagascar since for his purposes he defined Africa in his first and second reports (the Figures 3 there), as extending west of 50˚ E, which is inclusive of Madagascar.
Now returning to the validity of the CSIRO model’s results, documented offshore tests included a replica of a low windage flat item, ‘Roy’. Unaccountably at one point in a 7 days’ offshore trial, wind 15 knots NW-SW, this suddenly turned and it tracked downwind, cross current, giving the impression (to me anyway) that it might have lifted from the water and cartwheeled. Even though it was out of contact for a while the track does suggest this, though it also could be attributed to a small local current eddy, being offshore. See p8 Fig 2.3.5 of the CSIRO’s 1st report:
https://publications.csiro.au/rpr/download?pid=csiro:EP167888&dsid=DS1
As can be seen a replica of the curved flap track part-fairing recovered in southern Mozambique that trailed initially, apparently in a squall (though it is unclear why that would be), at the end of the trial had rejoined the undrogued buoys, matching their drift.
Earlier in a 2 days’ trial, wind 20 knots NW-SW the ‘Roy’ replica and particularly the flap track fairing replica both trailed the undrogued buoys (see 2.3.4 above the above). Yet that looks to be inconsistent with the actuals’ travel being quicker and farther (South Africa and southern Mozambique) than their like low-windage brethren. All are represented in CSIRO modelling by the undrogued buoys.
Of my above 18, 12 would float flat in the water, the remainder having curvature. Of those floating flat, 6 were thicker and so should have projected out of the water like those curved. Those with curvature might flip and have different stable postures having done so, their downwind speed and even direction varying. Those flat in the water might also exhibit bizarre reactions to strong winds.
So it can be argued that in some circumstances, particularly strong winds, the items classified as low windage by the CSIRO could make leeway at quite a different rate than the supposed 1.2% of wind, that altering leeway-plus-current vector additions and hence both speed and direction of travel.
However at first glance (at least), a change in speed of advance westward, even if true, looks likely to be an increase. Thus it is unlikely to affect David Griffin’s conclusion. There, the error is that the observed arrivals were too late from northern crash sites, not too early. Putting that another way, for the model to be adjusted so that its predictions were shifted to later to match the 50% and 90% arrival dates would require their speed of westward advance to be slowed, not sped up.
Even were the error in the other direction it would need to be very large for his conclusion to be overwhelmed by that alone.
Thus to me there is no basis to question the conclusion he has drawn form his Fig 3.
PS For completeness I add that there is another theoretical source of possible modelling inaccuracy. At 2.5 of the CSIRO 2nd report (13th April, 2017) it says, “Figure 2.3.1 confirms that our replica undrogued GDP drifters also have a total leeway velocity close to 1.2% of the wind velocity. More precisely, the magnitude of the leeway velocity is close to 1.2% of the wind speed. The direction, however, was sometimes significantly non-zero. We think this is because the wind direction changes more rapidly than the wave direction, so making point measurements of wind direction is not an adequate way of monitoring the direction of the waves (and hence the Stokes Drift).”
As it reads this suggests that a modelling assumption that there is no time lag for waves to change direction after the wind does could lead to errors. However this statement’s context was offshore trials outcomes where the wind apparently was squally, in a lee (albeit distant) and I assume that the delays are short. There should be a lesser instances of encounters with such squally conditions in the open ocean; and anyway the effects should tend to “cancel out” over a long period.
@Barry Carlson. While uncomfortable as an intermediary in questions to David Griffin, because I had been in recent contact with him over his new Fig 3 I did pass on your question. His answer was to the effect that he could not add to the explanations in his reports.
My understanding is that Stokes effect can be related to wind and wave size but as to the latter it is counter-intuitive I agree: small waves and wavelets have a large effect, big waves and swell much less so.
I note that at page 3 of the below, the CSIRO’s 1st report, it says, “The Stokes Drift is mostly due to the shortwavelength, locally-generated waves and, importantly, applies equally to all floating objects (of the sizes considered here).”
https://publications.csiro.au/rpr/download?pid=csiro:EP167888&dsid=DS1
At 2.3 of the CSIRO 2nd report below there is the following, “We showed in our earlier report that undrogued GDP drifters in the Indian Ocean had a total leeway velocity that was indistinguishable from wave model estimates of the Stokes Drift, which is very close to 1.2% of the wind velocity. We thus concluded that the true windage of those drifters was small compared to the Stokes Drift.”
I think one can deduce from that that the effect of Indian Ocean seas and swells also have little effect.
https://publications.csiro.au/rpr/download?pid=csiro:EP172633&dsid=DS4
Finally, recently he passed me a copy of a 2018 research paper (I do not know if it is on the web) that says in its abstract, “Thus Stokes drift
is therefore not primarily due to remotely driven swell; rather, it is mainly due to the much shorter waves that the local wind generates. By taking into account when the short waves break, it is shown how Stokes drift
can be approximately estimated directly from the local wind.”
@David
I have written to David Griffin asking him:
1. He showed 29 reported beaching dates, whereas the Malaysian report contains 27 beaching dates. The difference is 2 dates in September 2016 on his chart, which are not in the Malaysian report. What are these 2 items please?
2. The beaching dates in the reported items per month table include Rodrigues, Mauritius, Reunion, Madagascar and mainland Africa. Do the percentiles of modelled dates of debris beaching within 21-55E, 5-35S also include particles arriving in Madagascar and not just mainland Africa? Obviously Rodrigues, Mauritius and Reunion are excluded by the 55E boundary in the graphic showing percentiles.
In addition, I would comment:
If, as you say above, Madagascar is included, and Rodrigues, Mauritius and Reunion are excluded, then this should be true both for modelled dates and reported beaching dates. Otherwise we are comparing apples with oranges.
Out of the 27 items in the final Malaysian Report, 7 were unidentifiable. Out of the 7 that were unidentifiable, 6 were found by Blaine in Madagascar. David Griffin included all unidentifiable items, which distorts the results.
David Griffin also includes all Blaine’s finds on Madagascar, of which the first 8 were all discovered in one week in June 2016. The debris discovery time in Madagascar in this case was dictated by Blaine’s schedule. Debris on Madagascar could have arrived between 0 days and 270 days beforehand. It only takes on average 41 days for floating debris to track from Reunion to Madagascar. Blaine’s initial finds on Madagascar where 313 days after the Flaperon on Reunion.
The item handed to Blaine by Milson Tovontsoa, Rija Ravolatra, and Eodia Andriamahery, in September 2016 was found in February 2016 in Saint Luce, Southern Madagascar, but was also unidentifiable.
The inclusion of debris items found in Madagascar leads to a false peak at around 23°S and 22°S in David Griffin’s Fig. 3 results. Based on this false peak, David Griffin states a MH370 end point north of 25°S is highly unlikely because debris items arrive far too soon on African shores (whereby he apparently includes Madagascar in the term African shores).
An analysis of beachings on mainland Africa (excluding the unknowns surrounding the finds on Madagascar) shows a possible MH370 end point north of 25°S.
@Dr B. “Would the aircraft depressurize that quickly while descending from cruise altitude when the second engine failed? Why not just use the supplemental O2 mask?”
This to me is more a cannot-not-rule-it out type possibility than a likelihood. However in my mind was a non-professional pilot like Helios Airways Flight 522.
In MH370 once cabin altitude had increased to 11,000 ft the outlet valves would have shut, supposing these to be in auto. Thence there would be continuing slow rise. Warning heights depend on landing altitude and can be set to 15,000 ft but for landing altitudes of 9,660 ft or lower there would be an alarm at 10,000 ft cabin altitude. (Cabin oxygen masks would deploy until 13,500 ft cabin altitude.)
Thus in all likelihood there would be a warning before the outflow valves closed, which might prompt him to push the nose down. @TBill’s point about an O2 mask pilot being needed above 40,000 ft would apply to a near-complete decompression I think.
So for my part I do not think the aircraft would depresssurise quickly or that an oxygen mask would be needed by the pilot.
Incidentally just one item of interest in the wreckage, which I have mentioned earlier, would be whether the cabin oxygen cylinders were burnt or not; a normal result of their operation. An earlier deliberate depressurisation would result in that. Indeed for those who have the perfectly planned and undetectable crime in mind, that evidence would be a deterrent from deliberate depressurisation.
@Dr B. ….Cabin oxygen masks would NOT deploy…..
@David said: To me a recommendation for that [search to the north] would need to include a good reason why David Griffin’s now-reinforced conclusion should not prevail.
First, I think you would agree that I have not yet made that recommendation.
I think we are still trying to understand David G’s model, and why that model differs from Richard G’s estimates. I know that Richard is working on some new results. I also question whether we have valid timing data for the beaching of debris on the shores of East Africa, for the reasons mentioned. So perhaps I will eventually conclude that the drift models exclude a northern search. I’m just not there yet.
@ventus45
@DennisW
I do not really think MH370 did much to damage Razak politically, if that was ZS goal. I presume ZS anticipated that Razak would immediately realize that ZS stole the aircraft and ZS presumed Razak would blame ZS. However, as I say Razak handled this so “well” by refusing to acknowledge that the flight was pijacked and when Obama called Razak’s bluff, Razak cunningly only admitted to deliberate intervention by some unknown party.
Slightly humorous to me is recent defense of ZS saying it was not a rogue idea to be angry at Razak as many others were also. Sort of like damning with faint praise.
[Comments here are closed. Please continue the discussion under the new article.]