A new article by Hal Hodson on the search for MH370 was today published in the The Economist, and gives more details surrounding Ocean Infinity and its exploration technology. The article discloses that:
- Host vessel Seabed Constructor, owned by Swire and under lease by Ocean Infinity (OI), has been fitted with eight underwater autonomous vehicles (AUVs) for the search.
- The search will be conducted under the basis of “no find, no fee”, which means that OI will bear the economic cost of not finding the wreckage.
- Even though the contract with Malaysia has not yet been signed, Ocean Infinity will proceed with the search in order to take advantage of the favorable weather in the Southern Indian Ocean in January and February.
- The expected scan rate that is achievable using eight AUVs is 1200 sq km per day.
- Some additional testing of the scanning capability of the AUVs will be performed en route between the imminent departure from Durban, South Africa, and the arrival to the search area.
- The scanning will begin in the area designated by the ATSB as most likely (the 25,000 sq km) around 35S. If unsuccessful, the search will proceed towards 30S latitude.
- The advice to proceed north towards 30S latitude came from independent experts. (Readers here might be able to guess the names of the independent experts that have advised OI.)
- Rather than communicating with the autonomous surface vehicles (ASVs), the AUVs will communicate with the host vessel to periodically recalibrate the onboard inertial guidance system.
- If the flight data recorder (FDR) is found, it will be recovered and surrendered to the Australian authorities.
- Recovery of wreckage would require a separate agreement with the Malaysian authorities.
For readers of this blog, there are few new facts presented. Probably the most significant new fact is OI’s decision to start the search without a signed agreement. The article is helpful in that it will provide useful information to a larger, broader audience, and will increase the overall awareness of the new search. There will also be renewed questions as to why Malaysia has delayed signing the agreement with OI.
Update on Jan 3, 2018: Malaysian Transport Minister Liow was asked about recent developments regarding Ocean Infinity and the renewed search for MH370. He replied that the parties were in final negotiations, and there would be an announcement next week. The fact that he offered no stipulations for reaching an agreement, which has been the pattern in the past when Malaysia has wanted to stall the negotiation, is very encouraging.
JUST IN: Malaysia's Transport Minister @liowtionglai says he will be announcing details of negotiations for US company Ocean Infinity to resume the search of #MH370 next week. This after reports of Ocean Infinity sending a search vessel to Australian waters. pic.twitter.com/jBDKBTV0xh
— Sumisha Naidu (@SumishaCNA) January 3, 2018
Update on Jan 5, 2018. Channel News Asia is reporting that Malaysia has accepted Ocean Infinity’s offer to continue the search on a “no cure, no fee” basis. The information was sent to the families of passengers on in an email. (Malaysia in the past has informed the next-of-kin of new developments before releasing details to the public.)
Update on Jan 10, 2018. As widely reported, the agreement between Ocean Infinity has been finalized in a signing ceremony. The tiered payment terms are linked to where the debris field is found, and ranges from $20 million if found in the highest priority, 5,000 sq km area, to $70 million if found beyond the 25,000 sq km area. Here is the complete statement from Minister of Transport Liow:
As good as we could hope for. Certainly a nice, shiny vessel.
And I hope the 30S is a round number — to encompass the entire westerly-flowing anomaly t 7th arc. And some of the projections above with 29.6 or 29.5 crossings.
@lkr: I don’t think 30S was meant to be an exact latitude for ending the search.
I would prefer to state 30°S +/- 1°.
… and you heard it here first … on 12th February 2017 … “The probable End Point of MH370”
“Conclusion: The drift analysis appears to support a probable end point of MH370 around 30°S near the 7th Arc. This fits with a late final major turn south at 19:36 UTC and a flight at the normal cruise speed of 0.84 Mach until fuel exhaustion. There is a good fit to the satellite data and a good fit to a great circle path toward Wilkins Runway (YWKS) as the final waypoint.”
I now use 40S 100E as the ultimate waypoint and not YWKS, as it is not certain that YWKS is in the MH370 navigational database.
https://www.dropbox.com/s/6g82uw9b24r7qf4/The%20probable%20End%20Point%20of%20MH370.pdf?dl=0
@All: Multiple sources tracking Seabed Constructor report that the vessel has left Durban and is headed towards the search area in the SIO.
First heading south-east to the test site, I think.
http://www.dropbox.com/s/gms3c4hpmb2xkf6/2-1-2018.jpg?dl=0
ETA Perth on the 7th 1600 local time, which looks quick.
@Richard Cole: Thank you for the update. Unless we see a change in course soon, I think you are right.
It’s been typical for Constructor to perform some work close to port, subsequent to crew rotation. Now, with two more AUVs, exercising the launch procedures would be extra prudent. The continental shelf drops off to 3000m within 80nm of Durban. Should be evident by morning, SA time.
@Richard Cole. Did not see yours. ETA probably just nominal
Possibly heading for the ammunition dumping ground (bottom right) for tests ?
Seabed Constructor Tracker
@All,
There is a High pressure system moving east currently centered just to the south of Cape Town.
The South African marine synoptic chart for 2018-01-02 1800z is here, and parallel sailing due east at 35°S looks like the best action weather wise, i.e. light winds and the predominate SW’ly swell from abaft the beam.
The Interactive weather map from the Australian Bureau of Meteorology provides a good prognosis for the coming 6 days.
The above Marine Traffic ETA is in fact in February so perhaps the next intended stop, though I would have expected Fremantle.
Fremantle=Perth (8nm apart)
Wonderfull news. The new search is on. I hoped/expected they would go ahead even without a signed contract and they did. Showing great confidence and courage. I wish OI all the luck and a save voyage.
And if the crew of Seabed Constructor also keeps an eye on the ocean surface now and then they could even have the luck of spotting a piece of debris on their way towards the search area.
Anyway, if I was on board that vessel I think I could not resist scanning the ocean surface many times a day..
GR: You do realize that there’s no evidence that any remnant of MH370 has been afloat for a year, and probably much longer!
@Ikr
Yes I realize but still I could not resist looking if I was on board that vessel. There’s no evidence either nothing is still floating around anymore. I (probably wishfull) think some honeycomb-skinned pieces like flap-parts, closing-panels, gear-door pieces, engine cowling pieces etc. could still be floating around. Blaine even still expects debris washing ashore in WA around this times..
And Seabed Constructor is going to cross the south part of the SIO where debris can be expected to (have) travel(led) on their way back towards Australia or caught up in gyres.
Is there any site available with the planning details and live blog?
@ALSM. Fremantle=Perth (8nm apart). Fremantle is the port.
As predicted by Richard C., it appears that Seabed Constructor is performing sea trials at the edge of the South African continental shelf. The depth goes quite quickly from 800m down to 3,000m at this point. This is a good test to being able to handle underwater terrain such as Broken Ridge, where the depth goes from 2,500m down to 4,500m.
Seabed Constructor left Durban last night and proceeded around 30 NM south east and then all but stopped. I presume Ocean Infinity wanted to test out their 2 new AUVs, increasing the number of Hugins onboard to 8.
https://www.dropbox.com/s/6r7tb0fkm2c89uh/SC%20on%20the%20edge%20of%20the%20continental%20shelf.pdf?dl=0
Just noticed something strikingly odd looking at Broken Ridge in Google Earth.
I know you’re probably all going to shake your heads now and think I’ve lost my mind completly at last..
Still I’m going to share. The human mind is a strange thing..
Take a look at the overall shape of Broken Ridge. In my view it strikingly looks like the contours of the backside of a man lying face down, showing from his head down to his toes:
https://www.dropbox.com/s/le2zj9kis5yydea/Broken%20Ridge%20complete.jpg?dl=0
If Shah was the culprit did he provide a cryptical clue?
After all he broke his back in the past in a para-glyding accident.
Have a laugh or just forget it..
@skvery: Welcome to the discussion.
At the risk of making a self-serving comment, I think this site is very helpful in staying abreast of any new developments. Some here have subscriptions to marine data services, and submit comments to keep others informed. Also, some of us have relationships with Ocean Infinity and the officials involved with the search, and so you’ll get a perspective that will be hard to find elsewhere. Many here are very informed on technical aspects, and contribute excellent analyses and opinions.
@Richard : “I now use 40S 100E as the ultimate waypoint and not YWKS, as it is not certain that YWKS is in the MH370 navigational database.”
Richard I get the rationale re YWKS, but what’s the rationale for 40S 100E ?
Update on Jan 3, 2018
Malaysian Transport Minister Liow was asked about recent developments regarding Ocean Infinity and the renewed search for MH370. He replied that the parties were in final negotiations, and there would be an announcement next week. The fact that he offered no stipulations for reaching an agreement, which has been the pattern in the past when Malaysia has wanted to stall the negotiation, is very encouraging.
@skvery
I like to add, don’t regard my latest comment as being typical of this blog at all. Even for me it’s an exeption.
This is generally the blog to go for, for the most objective and scientificaly wise information.
@Ge Rijn: Thank you for the endorsement! Coming from you, that means a lot.
@Victor @Ge Rijn
“Anyway, if I was on board that vessel I think I could not resist scanning the ocean surface many times a day..”
Please arrange helicopter (and Dramamine) for Ge Rijn’s surface search.
Great to hear the search is “on” …I was a little melancholy without MY on board but the new Interview by the Transport Minister is uplifting of the spirit.
On a more serious note, how do you see the search going? From your text, it sounded a little like maybe short search at 35S and then move to 30S. But I think it is probably 32.5-35Sn maybe then move to 30S? or so we move up north through Broken Ridge? What is Broken Ridge strategy? I am not going to complain as long as we have search going on.
Richard- Can you share your spreadsheet calc details for your 40/100 path? I know you shared the YWKS path a while ago, I have studied that with some questions though.
Yes I agree though, I find myself using oceanic waypoints quite a lot instead of defined waypoints, when defining LNAV paths for MH370. NZPG for me is 78S67. The thing is, FS9 has certain waypoints, PSS777 had different waypoints, and of course the actual B777 has company waypoints. I know how to edit the waypoints in those programs with a text editor, but you can cut through it all just saying what lat/long you want. Having said that, I am at the moment favoring CTH or CMH with active pilot.
@TBill: I would be surprised if the entire 25,000 sq km proposed by the ATSB was not searched in the first swing. After that, I think OI will follow a continuous progression northeast along the 7th arc at a width of +/-25 NM rather than skip around to hot spots, but I don’t know that for sure. Perhaps 30S will get special attention early.
Relative to flight plans, custom waypoints that are integer latitude and longitude coordinates are easily created and entered in the FMC. For instance, (-40,100) would be entered as S40E100, without any knowledge of what waypoints are stored in the navigation database. I think that is what you are saying.
@VictorI
That’s quite a compliment coming from you. Thank you.
I try my best in my own way, that’s all I can do like anyone in his/her own way.
All who have been investigating, questioning and pushing to solve this tragedy and mystery to a positive outcome deserve regards.
No matter how critical they have been (or still are) on reasonable grounds.
The time, investment and dedication you and others showed over the years is quite remarkable and has undoubtedly contributed to this new search endavour.
The biggest compliment is to you sustaining this blog with your knowlegde and knowlegded connections.
It hasn’t been in vain. A new search is going to start. Friends and relatives from the ones missing will find some redemption in those efforts.
@all: FWIW, I hit a paywall for the Economist [“exceeded limit” tho I don’t recall even one read]. Would be obliged if anyone posts quotes or summary here, or directs to a non-paywall site.
Thanks
@Victor
Does S40E100 work? That would be entered as 40S00 for me.
Some of you have asked that I publish my MH370 Flight Model for various flight routes. The attached link is to Excel files for great circle tracks between 168°T and 185°T from 19:41 UTC.
If you prefer files in the original Numbers format, please let me know. Please note that Apple Mac does not always export everything from Numbers to Excel correctly, in which case the last value is retained instead of the formula.
You can tell which link belongs to which track by looking at the 3 digits before the .xlsx in each case:
https://www.dropbox.com/s/cf5tw90hdvvnbcg/MH370%20Flight%20Path%20Model%20V17.0%20168.xlsx?dl=0
https://www.dropbox.com/s/k3rt3nuzi1j4004/MH370%20Flight%20Path%20Model%20V17.0%20169.xlsx?dl=0
https://www.dropbox.com/s/inltq6c3amr2uub/MH370%20Flight%20Path%20Model%20V17.0%20170.xlsx?dl=0
https://www.dropbox.com/s/xvljd0gukej351h/MH370%20Flight%20Path%20Model%20V17.0%20171.xlsx?dl=0
https://www.dropbox.com/s/ip4mu6kda3qx4xc/MH370%20Flight%20Path%20Model%20V17.0%20172.xlsx?dl=0
https://www.dropbox.com/s/qi7kqn0l4s6wqug/MH370%20Flight%20Path%20Model%20V17.0%20173.xlsx?dl=0
https://www.dropbox.com/s/9yc6w1gkpo5bpye/MH370%20Flight%20Path%20Model%20V17.0%20174.xlsx?dl=0
https://www.dropbox.com/s/0oyw0jdkrb1lrb8/MH370%20Flight%20Path%20Model%20V17.0%20175.xlsx?dl=0
https://www.dropbox.com/s/rybzvugxmjufn85/MH370%20Flight%20Path%20Model%20V17.0%20176.xlsx?dl=0
https://www.dropbox.com/s/r9kie8cs6p61ki2/MH370%20Flight%20Path%20Model%20V17.0%20177.xlsx?dl=0
https://www.dropbox.com/s/qzvk30z5lk51gat/MH370%20Flight%20Path%20Model%20V17.0%20178.xlsx?dl=0
https://www.dropbox.com/s/0l5yz3zcawm24mj/MH370%20Flight%20Path%20Model%20V17.0%20179.xlsx?dl=0
https://www.dropbox.com/s/oychkr2jbetz2hw/MH370%20Flight%20Path%20Model%20V17.0%20180.xlsx?dl=0
https://www.dropbox.com/s/yfn43l0mhfba7vf/MH370%20Flight%20Path%20Model%20V17.0%20181.xlsx?dl=0
https://www.dropbox.com/s/81wfbe9v1u12bhu/MH370%20Flight%20Path%20Model%20V17.0%20182.xlsx?dl=0
https://www.dropbox.com/s/knrp9byotszu855/MH370%20Flight%20Path%20Model%20V17.0%20183.xlsx?dl=0
https://www.dropbox.com/s/ud3xtgz4tbjm0m8/MH370%20Flight%20Path%20Model%20V17.0%20184.xlsx?dl=0
https://www.dropbox.com/s/ptfn1p5xr2wce1r/MH370%20Flight%20Path%20Model%20V17.0%20185.xlsx?dl=0
@TBill: Yes. The long form is NDDMM.MEDDDMM.M. For integer degrees, it can be shortened to NDDEDDD, where N may be N or S, and E may be E or W.
This is the blurb from the Honeywell guide to the B777 FMS regarding how to enter a custom waypoint using a latitude and longitude:
Latitude/longitude (for example, a lat/long entered as N5000.5 W02000.8 and displayed as N50W020). Leading zeros are required. Trailing zeros are optional when the latitude or longitude is whole degrees (for example, N60W040).
@lkr
Economist article
http://www.dropbox.com/s/9bshp9j4lfoiltn/Economist-2-1-18.pdf?dl=0
@Richard Cole
Thx. Normally I can find a way around a paywall, but no luck with The Economist.
@Richard Cole: Thank you for the article. I didn’t realize that others were having problems with the paywall. I changed the link in my post to refer to a copy of the PDF you supplied.
@TBill
Re: ”Does S40E100 work? That would be entered as 40S00 for me.”
The five character format you mentioned can also be used to enter lat/long waypoints, but only in cases where the lat/long is in full degrees. The five character naming convention for such waypoints is a bit tricky, as the following extract from the Honeywell FMS manual shows:
Positions in the Northern hemisphere use the letters N and E, and positions in the Southern hemisphere use the letters S and W. Latitude always precedes longitude. For longitude, only the last two digits of the three-digit value are used.
The position of the designator in the five-character set indicates whether the first longitude digit is 0 or 1. The letter is the last character if the longitude is less than 100°.
N is used for North latitude, West longitude. E is used for North latitude, East longitude. S is used for South latitude, East longitude. W is used for South latitude, West longitude.
For example:
• N50°W040° becomes 5040N
• N75°W170° becomes 75N70
• N50°E 020° becomes 5020E
• N06°E110° becomes 06E10
• S52°W075° becomes 5275W
• S07°W120° becomes 07W20
• S50°E020° becomes 5020S
• S06°E110° becomes 06S10.
Clear as mud? Pilots prefer to enter lat/long waypoints using the method Victor described, because it covers all cases and avoids any confusion. Here’s another extract from the Honeywell manual that describes the ‘longhand’ method:
Latitude/longitude waypoints are entered with no space or slash between latitude or longitude entries. Leading zeroes must be entered. All digits and decimal points (to 1/10 minute) must be entered unless the latitude or longitude is in full degrees. Waypoints entered as a lat/long are displayed in a seven-character format. For example, N47°W008° is entered as N47W008 and displayed as N47W008. N47°15.4” W008°3.4” is entered as N4715.4W00803.4 and displayed as N47W008.
@Andrew @Victor
I knew that the detailed lat/long waypoint format worked – tried it once – but I got more comfortable with the whole numbers lat/long method. Detailed format even possibly better for explaining MH370 if pilots are used to using that, the long form method could be applied to get to any finely detailed map point in the SIO.
@Damien you asked “Richard I get the rationale re YWKS, but what’s the rationale for 40S 100E ?”
The rationale for YWKS was: starting at 6.507653°N 93.440077°E on the 2nd Arc at 19:41 on a track of 172.6805°T on a great circle with an ultimate waypoint of YWKS at 66.69306°S 111.50389°E, you cross the 7th Arc at 29.7578°S 98.4580°E after 4,068 km at fuel exhaustion and show a relatively good fit to all BTO and BFO data en route.
The rationale for 40°S 100°E is: starting at 6.507653°N 93.440077°E on the 2nd Arc at 19:41 on a track of 173.1161°T on a great circle with an ultimate waypoint at 40°S 100°E, you cross the 7th Arc at 30.0528°S 98.2032°E after 4,097 km at fuel exhaustion and show a relatively good fit to all BTO and BFO data en route.
Both cases are essentially the same rationale: An easy to remember and easy to enter ultimate waypoint, YWKS or 40S00 or S40E100.
The only difference is that YWKS may not have been in the MH370 navigational database, whereas you can always enter 40S00 or S40E100.
Victor has shown that an end point at 29.7°S fits the RNZAF debris sightings 21 days later using the drift data from CSIRO, but these sightings were not proven or discounted to be from MH370, as no debris was recovered.
I have shown that an end point of 30°S +/- 1° fits the transoceanic drift analysis to the locations of the 25 floating debris finds from MH370 after between 508 days and 1056 days, where debris items have been recovered and proven to be from MH370.
Yesterday, Ocean Infinity has added a new video on the web site :
https://oceaninfinity.com/view-seabed-intelligence-action/
After some absence, may I post a quick note that some time ago I have come up here and elsewhere with a similar route to YWKS (or similar waypoint), however via BEBIM. To my mind, this would lead to an impact point at ~31S.
https://www.docdroid.net/GvlrLaV/mh370-waypoint-30.pdf
It seems to be a very good fit BFO/BTO wise.
I have also taken into account available drift studies from all institutes that have published about this or where studies are available (in the appendix)
This is also close to the area where on 29 March an object was spotted (the flaperon?), but not recovered.
https://twitter.com/TimesPictures/status/449541489488441344/photo/1
I’m afraid, though I have promised, I have not so far taken into account wind in my model. This was because of day time job obligations. Once I have a bit more time, I might rework it a bit, but I also think that a further twist in speed would suffice.
Of course, I was not the first to suggest a route to YWKS, but as far as I know, the route via BEBIM (before that ISBIX seems a good fit) has not been suggested before. That would make a slight difference. So perhaps this could be of interest.
I hope the area will also be covered, though I would of course be more than happy to have the result double-checked by someone more knowledgeable in that area than me.
@Richard, when you refer to great circle path with track xxx degrees, is that signifying initial bearing? And isnt anything more than one decimal bearing overkill if the tolerance of autopilot is something like +/-0.1.
@Paul Smithson: In LNAV mode, the deviation from the great circle path is continuously monitored and corrected. The high resolution of the initial track angle in our models is therefore representative of the precision of that great circle path, even if the track angle actually flown at any time deviates from the exact great circle path.
@Richard
On your 40S100E suggestion as a possible end-waypoint.
If this easy to remember waypoint was set it would mean it was set deliberately also. But with what objective? There is nothing specific at 40S100E.
If you assume a deliberate waypoint was being set just before or after FMT I can fully support that assumption but not without a specific destination.
There is no logic in flying towards 40S100E (or YWKS or McMurdo or SP) after a flight that shows many signs of well planned, controlled and executed manouvres till ~18:40.
End-waypoints set just before or after that time I fully endorse but than specific waypoints to aspecific destination.
32.5S/97E is easy to remember also but this coördinates end up in a specific ~5200m hole in Broken Ridge.
In the context of the whole flight and what we know about it now, a specific waypoint like this is more logical imo.
If the dissapearance was well planned from the start (which it has all the signs of) the culprit would not leave the end-of-flight and its location to coincidence imo.
@Nederland
Your path through BEBIM to YWKS probably has a little better BFO match, but we still lack a definitive path prediction. We are hoping the 30-31S will be searched fairly soon after the 32.5-35S area is searched, so your path end point should hopefully be covered.
@TBill
Quite a lot seems to depend on the question of whether or not YWKS was indeed part of the database. I hope Ocean Infinity will be made aware of this. My hunch would be it is, otherwise the ATSB would more likely have said it’s not, rather than not commenting on this at all (which is what they did in my understanding).
If YWKS was indeed in the database, then BEBIM would be the only penultimate waypoint that works. Of course, it can not positively be proven that MH370 was navigating along fixed waypoints all the way, only in the early stages of the flight.
In the version above, I came to just south off 31S. (31.1S)
A Malaysian government official told me last month that none of Antarctica WPs were in the 9M-MRO database.
@Ge Rijn
Where do you see a 5200-m (17160 ft) deep hole at 32.5S/97E? Google Earth or other map. The other advantage for a waypoint closer to Arc7 is that the pilot would have fuel estimate to that point. I agree with you targeting Broken Ridge is a possible objective, but it is hard to see a specific target on Google Earth (given Dordretch Hole was apparently out of reach).
Right now I favor the objective was to set something like S31E93 as the target which is right on Broken Ridge. The pilot would have known he could get well past that point, sort of a goal post to the deeper water beyond BR. Possibly after passing BR the pilot could turn Easterly to get into the rougher terrain. After fine tuning, I dropped the 31S93 waypoint in favor a similar path that goes 180S CTH from ISBIX, but basically crosses BR in the same place.
@airlandseaman
Thanks for sharing.
@Ge Rijn, @TBill, @Nederland, @airlandseaman, et al.
Selecting a particular final waypoint at this time is not required to guide the search. A range of latitudes north of the 25,000 sq km will likely be progressively searched, as it will be difficult to justify a particular waypoint-based path while excluding others. We also know that custom waypoints described by coordinates are easily entered, so what’s in the navigational database might not matter.
Let’s assume that OI will first search the 25,000 sq km, and then proceed further north along the 7th arc, as defined at a height of 20,000 ft, at a width of +/-25 NM. Is this the proper search plan? I know that I’ve asked this question several times in different ways.
@Paul Smithson you stated “@Richard, when you refer to great circle path with track xxx degrees, is that signifying initial bearing? And isnt anything more than one decimal bearing overkill if the tolerance of autopilot is something like +/-0.1.”
Please note:
The track was given as a whole number in steps of 1°.
The results are given as 30°S +/- 1°.
If you had bothered to open any of the Excel spreadsheets, you will see this is the initial bearing.
If I choose to run internal calculations in the Excel spreadsheet to more decimal places, that is my choice.
What is your point? Give me a break! Haven’t you got any important to contribute?
I guess you know by now my imagination has no limit..
And my background of working 25 years in psychiatry did not help to scale that down, on the contrary I’m affraid. I know the more how strange the mind can work things out totally believing it’s own logic and perception. Including my own.
I’m still contemplating on the strange shape of Broken Ridge in the context of Shah and his broken spine in the para-gliding accident.
I wonder which vertebra he broke then. Was it a lumbal one? Which one?
The area between ~95E and ~98E just resambles the lumbal spine in the whole context of Broken Ridge viewed as a man lying on his belly facing towards the west with his feet just under the tip of WA.
I know we have to approach this all rational and scientifically in the first place but we can not leave out the irrational as it comes to human behaviour and motivations. Quite some aircraft accidents have proven this.
@TBill
Just scroll to the Google Earth coördinates ~32.5S/97E. If you find something different please tell me.
@Ge Rijn: you stated “@Richard On your 40S100E suggestion as a possible end-waypoint. If this easy to remember waypoint was set it would mean it was set deliberately also. But with what objective? There is nothing specific at 40S100E.”
In my view, the objective was to lose the aircraft in the middle of nowhere. If there is nothing specific at 40S100E, or for miles around, then it is definitely in the middle of nowhere. A pilot who previously tested fuel range and endurance to 45S104E on his home simulator, would know that that 40S100E was beyond the fuel range and endurance on 7th/8th March 2014. Such a pilot might even have calculated that the end point would be around 30S98E.
I am not the first person to observe, that ZS has succeeded so far, in ensuring that MH370 has not been found.
@VictorI
I have no problems with the overall search plan of OI as you describe it.
I only have big problems with the width of +/-25NM.
Not with an initial search along the 7th arc till ~29S but if they take this as a limit in width overall.
IMO they should take at least +/-40NM in certain aereas if they don’t succeed within +/-25NM.
I’m sure they’ll find the wreckage if they search all the way from 35 to 30 (and potentially a little bit further north if it still hasn’t turned up). So I’m too looking forward for confirmation.
@Victor
RE: waypoints
As you know I have come to favor 180S vs. waypoints, but I am very willing to go back to waypoint LNAV if they do not find the aircraft in the search area.
As far as MAS not having polar waypoints, that is understandable, but does not matter too much. FS9 and other flight planners certainly do have polar way points, and nobody is really suggesting a possible rouge flight was based on just hopping into the cockpit without pre-planning and restricting oneself to MAS company waypoints. I routinely use 78S67 for NZPG.
@Ge Rijn, @Nederland, @TBill: It sounds like you agree with the likely search plan. If the plane is not found before reaching 29S, there will be a robust discussion about what to do next. Options include searching wider, searching further northeast along the arc, or re-scanning areas, assuming there is interest to continue searching. There will also be more attention to alternative theories with end point outside of the SIO. Frankly, it will be a mess.
@Richard Thank you for responding to my question.
@Victor
“If the plane is not found before reaching 29S, there will be a robust discussion about what to do next…
Frankly, it will be a mess.”
Nihil desperadum.
“…as it will be difficult to justify a particular waypoint-based path while excluding others.”
Surely the final simulator point should have precedence given that it is the only extant direct communication from the perpetrator. (I regard dismissal of this evidence as absurd}.
Interesting to note that, according to my crude skyvector analysis (I’m not an aviator), 6.507653°N 93.440077°E to 45s104 passes over YPCC and Dordrecht Hole (33s101).
Dordrecht Hole is apparently beyond fuel range, and the path crosses the 7th arc a bit north from the planned search {28.3s104}.
Nevertheless as a back up plan “If the plane is not found before reaching 29S”.
@Ge Rijn
I guess if the whole reasoning behind the final BFOs is incorrect (which I don’t think it is – because it is two independent data points and no strong reason to say the debris doesn’t match), then you have to have a very clear trajectory, including BTOs and BFOs, to make any search economically viable.
@Victor
My feeling is to give priority to Arc7 +- 25nm up until about 20S. I take the Inmarsat data as a gift/contribution to science and I want to stick with it, unless of course we uncover some new facts/secrets.
@all
I still struggle with the notion of a “no cure, no fee” contract. As a payer, I would want to base the fee on level of effort to avoid a “windfall” effect. As a payee I would not want the payer to have an option to cancel after a certan level of effort has taken place, and I receive nothing. My sense is that it is literally impossible to construct such a contract that is satisfactory to both parties. There are also the issues of validating a “find”, and the escrowing of funds. I would think a mutually agreeable third party is needed for both. Good luck with that.
I regard the whole notion as a dumb idea that seems simple and atrractive on the surface, but has a number of difficult issues under the surface (no pun intended).
@DennisW: Be careful about trusting all the statements made in the media reports. First, the fee might depend on where the debris field is found. For instance, if Seabed Constructor rushes to the spot where CSIRO’s David Griffin expects the debris field to be, and is successful there, perhaps OI doesn’t get paid as much as if it is found at 27S, for instance. Second, I’m not sure there really is a time limit of 90 days to collect the fee. I’ve never seen a direct quote that references a time limit. Rather, it suddenly appeared in one story, and was repeated in others. If anybody is aware of a direct quote, I’d like to see it.
@DennisW
Nothing ventured, nothing gained?
@DennisW
In the British legal system, no win no fee seems to be quite common as everyone is advertising this. This normally includes a combination of hourly pay and success fee. Sure, there is more money at stakes, but a law case involving several millions of pounds is not very unusual either. I could imagine there is some upper cap on the eventual total fee.
@Victor
My comments have nothing to do with anything I have read especially the bogus article with the time limit.
@Nederland
A capped fee with a level of effort modifier is probably how things will end up if they end up at all.
@DennisW
I think Malaysia is just haggling. They know OI can not back out any longer at this stage and they are trying to reduce whatever part of the fee a little bit.
There will be elections in August 2018, so it might be good a) to put an end to the saga b) come up with a reasonable spending sheet for this.
@Nederland
Another factor complicating the negotiations, and I have raised this issue before, is assigning a value to finding the wreckage. My own opinion, humanitarian considerations aside, is that it is quite low.
I also believe that if an agreement is reached, that the terms will not be disclosed.
@DennisW
The interest of Malaysia, as a government, or as a country, would to my mind be to make up for reputational damage arising from the incident, including foreign relations, and its less-than-ideal management (some might speculate that was the whole point of the disappearance).
For the government it might secure an election win.
I presume any investigation into the wreckage won’t happen before August 2018.
Now is the time.
@Nederland
The reputation of the Malay government is beyond trashed. The divsersion of public funds by Najib and his pal Jho Low are well known. The fate of MH370 is insignificant in the grand scheme of things.
@DennisW
That might depend on possible coalitions.
Final chance?
Can someone explain the fascination with an ending latitude of S30 as being deserving of a higher priority than any other latitude in the general search area? The CSIRO study first pointed to it as being a location where debris would avoid Australia, but the flaperon would travel too fast. Richard put forward a drift model based on GDP “drogued” buoy data [if I have that right] that picked out 30S as best matching the timing of debris in East Africa (and barnacle data, such as they exist), but drogued buoys travel slower than undrogued buoys, which might solve the timing problem, but the latter should better represent the drift speeds of debris.
I am unable to find any compelling reason that the satellite data would favor 30S, notwithstanding blathering about YWKS and such.
What am I missing?
Esky: Don’t think anyone on this site has exactly placed a eureka! pin at 30S — but combination of drift, specifically lack in WA [with required westerly current according to 3/14 satellite date, found also around 35%], and holes in surface search coverage [see earlier VI post] make this an evident warm spot for areas not covered by Fugro.
So it will be searched. As VI says, it get much harder further north and pretty hopeless off 7th arc.
lkr
You say, “Don’t think anyone on this site has exactly placed a eureka! pin at 30S …”
There are 16 mentions of “30S” (and variants) in the parent article and in the comments section of this thread alone. Whether one calls it a “fascination” as I did or a “eureka! pin” as you did – it gets a lot of attention.
@sk999
I think your perception is correct relative to 30S.
I can only speak for myself in that regard. The drivers for me fall into five categories.
1> I strongly believe in a late FMT with a 19:41 latitude of 6N or so. The Inmarsat study had a 19:41 latitdue of 0N and a 34.7S terminus. YAP’s work had a 19:41 latitude of 0.4S and a 37.5S terminus. The 6N 19:41 latitude will drag the terminus North to near 30S.
2> I like Richard’s drift study.
3> The Z simulator data supports a terminus near 30S.
4> I have weak integrated Doppler work that supports 30S.
5> I have geometric algebra support for 30S.
I never published anything on 4>, but we (you and I) did chat about it briefly. I did publish a lot on 5> but took it down because I was not able to answer questions raised by previous work colleagues. No one here asked any questions about it, so I did not feel bad about stepping away from it.
I am not prepared to defend 4> and 5> above. I feel pretty good about 1>, 2>, and 3>.
Esky: I think Dennis is coming at this in the same way I imagine — that 30S is “warm” simply because it has few strikes against it, and that’s enough [given the Fugro search] to raise the vicinity to a decent maybe [several I think were guessing around 30% likelihood for #0S =/- 1 degree. That’s merely warm in my estimation.
So, very much “underconstrained” as one of our resident sages [?Dennis again] put it.
One thing to remember is that the drift models are little more than permissive. But I do take the WA absence as real, given a serious search effort there over the first two years. Elsewhere, debris finds follow in time and space the itinerary of Blaine Gibson, albeit informed by Dr. Chattiaratchee.
I think the ~30S latitude preferrence on this blog leans rather heavily on @Richard’s drift-study. Which contains a serious basic mistake imo.
His study is based on drogued buoys/drifters while the debris can be regarded as undrogued buoys/drifters.
The difference in drifting behaviour between both is huge though.
The slip of an undrogued buoy/drifter is more then 10 times that of a drogued one. From the GDP site:
http://www.aoml.noaa.gov/phod/dac/gdp_drifter.php
The next text:
“The most important design characteristics that minimize slip are low tension between the surface buoy and drogue, which avoids aliasing wave motion, and a large drag area ratio (Niiler et al., 1987). As long as the drogue remains attached to the drifter, the downwind slip is estimated at 0.7 cm/s per 10 m/s of wind speed (Niiler and Paduan, 1995). If an SVP drifter loses its drogue, it will slip downwind at a speed of 8.6 cm/s per 10 m/s of wind (Pazan and Niiler, 2001).”
@Ge Rijn
I already answered your unfounded criticism previously in this forum. You obviously have a short memory.
My database contains both drogued and undrogued data from the GDP buoys and my simulator contains a switch so that I can use either. The GDP data give you the date and time that a buoy loses its drogue, if that happens. I have compared both types of data from drogued and undrogued buoys.
May I remind you once again, the major point why I used data from drogued buoys in the recent study is that I wanted to remove the effect of historical windage to establish a baseline. Many of the debris items recovered were flat panels with little or no windage. A drogued buoy has a wind slip of 0.014 knots in 19.4 knots of wind. An undrogued buoy has a wind slip of 0.167 knots in 19.4 knots of wind. We are talking about a difference between 0.07% and 0.86% wind slip. This is not huge as you claim. My software allows you to then add windage back in, on top of the baseline, if it is relevant. The larger items of debris found did have significant windage as CSIRO have shown for the Flaperon.
@Richard
My memory is still fine thank you.
What I understand is drogued buoys are designed to follow the deep surface current at ~15m and to level out the effects of wind driven surface currents, wave driven effects and windage on the buoy as much as possible.
Those buoys by their nature don’t represent actual debris floating on the surface which are subjected to those surface forces substantially.
You might say a downwind slip at 0.7 cm/s per 10 m/s of wind speed against a downwind slip at a speed of 8.6 cm/s per 10 m/s is not huge.
I would say it is huge. Over the thousands of miles involved it substantially affects the arrival times and trajectories.
I understand you corrected your data on drogued buoys afterwards. But I still don’t understand why you used the drogued buoys at all if only to remove the effect of historical windage.
Isn’t the historical windage an important indicator too?
Essentially in my thinking drogued buoys data don’t represent actual floating debris on an ocean surface subjected to wind and waves.
If you introduce them in a drift-study they rather contaminate realistic outcomes imo.
For this reason I think it’s better to work with undrogued buoys only from the start like others have done.
Your approach systematically and logically leads to your results. But I keep critical of your approach.
I would like to see your study repeated only based on undrogued buoys and with historical windage included.
Update on Jan 5, 2018. Channel News Asia is reporting that Malaysia has accepted Ocean Infinity’s offer to continue the search on a “no cure, no fee” basis. The information was sent to the families of passengers on in an email. (Malaysia in the past has informed the next-of-kin of new developments before releasing details to the public.)
Malaysia accepts US firm’s ‘no cure, no fee’ offer to resume MH370 search
Read more at https://www.channelnewsasia.com/news/asiapacific/malaysia-accepts-us-firm-s-no-cure-no-fee-offer-to-resume-mh370-9835834
@Richard
To make a simple calculation; 8.6cm/sec makes ~310m/hour, ~7.5km a day. In 500 days this makes 3750km. This counts for the udrogued buoy.
0.7cm/sec makes ~2.5m/hour, ~60m a day. In 500 days this makes ~30km.
This counts for the drogued buoy.
If this is not a huge difference I don’t know what would be.
@VictorI
Wonderfull news! Now it’s official. The search is on again.
All efforts have not been wasted. And Malaysia stepped in at last. Wonderfull. Whatever the outcome this must be comforting to the NoK.
@Ge Rijn
I would love to see your drift analysis, since you are so convinced that I have got everything wrong!
@sk999: I’ll speak for myself, and not others. I view 30S as a warm spot with a probability a bit higher than other terminations. I advocate starting with the 25,000 sq km, and then searching as far north along the arc as resources allow. If there was no drift in the BFO bias from its initial value of 150 Hz, then the plane will likely be found in the 25,000 sq km. If there was some drift in the BFO bias from 150 Hz (and while others here are certain it was essentially zero, I think it possibly occurred based on the DSTG’s study of 20 previous flights of 9M-MRO, and also because we cannot be sure that the bias remained unchanged before and after the in-flight re-boot), then the BFO becomes a poor discriminator of flight paths. For the LNAV path ending at 30S, the RMS BFO error is 5.7 Hz, but the mean BFO error is -5.4 Hz. The debris photographed on March 29, if from MH370, suggests a terminus near 30S, as do certain LNAV final waypoints, such as YWKS and S40E100. That’s not enough evidence to justify searching only at 30S. However, I think it is enough evidence to justify not excluding a search as far north as 30S if the debris field is not found in the first 25,000 sq km.
Regarding drift models, the flaperon is predicted to arrive earlier than when it was discovered for a 30S terminus. However, “Roy” is predicted to arrive much later than it was found for a 35S terminus. In many ways, a later than expected discovery is easier to explain. Before discovery, the piece could have beached and soon after been dragged back out to sea, or caught in a gyre close to shore. On the other hand, an earlier than expected discovery would require being advantageously positioned in a fast-moving jet of water. In light of the timing of the discoveries of both the flaperon and “Roy”, neither 35S nor 30S match particularly well.
@Victor
I agree with you, that is why I only give 30S a probability of 10.3%.
@Richard
I’m just being critical. I’ve read quite a lot of other drift-analyzis over the years. Your way is not the only way.
I regard your work highly but I think I noticed a basic flaw that influences your outcomes to conclude on a latitude just too far north imo.
If my arguments make no sence in your opinion please reply with contra-arguments based on what I stated.
@VictorI
Your remark:
“In light of the timing of the discoveries of both the flaperon and “Roy”, neither 35S nor 30S match particularly well.”
That’s essentially why I opt for the ‘middle of the road’ between 32S and 33S among other reasons. ~32.5S would suite all assumptions and data best imo.
Don’t worry Ge Rijn, the drift analyses are totally off track.
Here are the results of my latest work. In a nutshell: throw away the offset manoeuver at 1824, the 1822 timestamp is wrong. No manoeuvers required, just a speed reduction from slowing down from M0.84 to M0.81
Pilot had worked out previously that if he waited until he is 25NM out from IGOUGU before his FMT toward waypoint S41.00 E88.00, he will have the flight path set up just as he wants it. When the fuel ran out, he glided a further 97NM possibly more before impacting the ocean. He probably played the aggressive test pilot role in the process, putting it into a steep dive to gain energy before pulling out.
The flight path that was taken after the aircraft had flown beyond radar range can now be reconstructed with a much greater degree of confidence. The latest findings are that firstly, the pilot reduced speed from M0.84 to M0.81 shortly before flying out of radar range. Secondly, that he remained on flight route N571 at FL350 until he was 25NM out from waypoint IGOGU, at which point he commanded the FMC to fly toward a pre-determined manually inserted waypoint, identified as S41.00, E88.00. Thirdly, that as soon as the aircraft began the turn south, the pilot began to reduce speed from M0.81 to an estimated M0.72, as the initial step in a range finding exercise to find the best speed/altitude combination to maintain until fuel exhaustion. The purpose of the ranging exercise was to find the speed/altitude combination that would give the greatest range, on the available fuel. The exercise entailed reducing speed to about M0.72 and then gradually increasing both the speed and altitude while monitoring the FMC’s range projections, until the optimum cruise conditions were identified.
The main points in more detail:
1) Aircraft remained on air route N571 until it was exactly 25NM away from waypoint IGOGU, at which point it turned south towards a waypoint set up by the pilot, a manoeuver strikingly similar to a navigational procedure the pilot had practiced flying on his personal flight simulator, on 2nd February 2014. The pilot-inserted waypoint is identified as S41.00, E88.00. This particular set of coordinates was evidently chosen with the purpose of synchronizing fuel exhaustion with sunrise.
2) A previously proposed lateral offset manoeuver at UTC1824 is now not required. The final radar position was originally allocated a corresponding time stamp of 1822.20. The original time stamp appears to be incorrect because if the time is amended from 1822.20 to 1823.60, it only requires a speed reduction from Mach 0.84 to Mach 0.81 to get a satisfactory match between the flight path and the 1825 and 1828 BTO ping rings. And if this is the case, the proposed lateral offset manoeuver can be dispensed with. The aircraft then maintained a speed of Mach 0.81 up until the FMT manoeuver at 1836.50.
3) Purpose of the previously unexplained slowdown at around 1840 is now understood. The temporary slow-down initiated when as aircraft began to turn south now has an explanation and what is particularly significant, the explanation furnishes a useful insight into the pilot’s thinking, which was to maximize the range attainable from the remaining fuel, an insight that is in turn useful when it comes to identifying the most likely end-of-flight scenario.
4) Total distance covered between 1829 and flameout at 0017.50 is just 10NM short of the estimated maximum range attainable for Mach 0.82 cruise at FL390, as identified by Dr Bobby Ulich’s fuel model.
Revised sequence of events:
A few minutes prior to flying beyond radar range, the pilot began a gradual slowdown from M0.84 to Mach 0.81.
At 1836.50, when he got to within 25NM of IGOGU, the pilot initiated an FMC controlled turn toward waypoint S41.00, E88.00, maintaining FL350. At the same time, began a slow-down from Mach 0.81 to approximately M0.72*. The 1840 BFO (telephone) data suggests that the aircraft’s speed had slowed to between M0.67 and M0.75 by 1840.
*From the BFO data alone, it isn’t possible to arrive at an accurate value for the speed. However, when I calculated the various speed-up times corresponding to a range of minimum speeds between M0.67 and M0.75, I found that M0.72 produced a 17 minute speed-up to M0.82, which appeared to result in the most acceptable overall speed/time profile between the FMT and the 2nd arc. For example, assuming a basically linear speed-up rate (admittedly an oversimplification, as in practice the speed and altitude rate would be steeper at the beginning, and level off toward the end) then M0.67 produced a speed-up time of slightly under 12 minutes, while M0.75 produced 25 minutes. M0.72 at FL350 comfortably fits the 1840 BFO data, and would be a suitable speed from which to start range testing. As soon as M0.82 at FL390 was identified as acceptable, it was duly maintained until fuel exhaustion.
The FMT was completed at 1838.35 at N07.27, E94.61, bearing 186.68985 deg. The FMT was a 108deg, 7.5Nm radius turn.
1839.50 Slow-down to approximately M0.72 was complete. Almost immediately, the pilot initiated a gradual speed-up from M0.72 to M0.82 and at the same time, began to climb from FL350 to FL390. This combined speed-up and climb took an estimated 17 minutes, and was completed at about 1857. The pilot had apparently decided in advance on constant Mach cruise as the most appropriate cruise option, and it’s therefore reasonable to assume that the slow-down, speed-up manoeuver was a range-testing exercise performed in conjunction with the FMC, the purpose of which was to identify the Mach speed/altitude combination that produced the greatest range from the remaining fuel. The FMC was navigating toward a manual waypoint located some 200Nm beyond the aircraft’s fuel range, consequently the most straightforward way to find the optimum speed and altitude was to gradually increase both speed and altitude and monitor the varying FMC range-to-go readout.
Because the range seeking exercise was performed just as soon as the aircraft had completed its final turn, it can be inferred that the pilot’s priority was to get the best possible range out of the remaining fuel. Following on from this, one might then reasonably assume that the pilot was also intending to glide as far as possible towards S41.00, E88.00 once the fuel had been exhausted. Possibly his thinking was simply that the further he flew, the less likely any debris would make landfall? Although the autopilot will disengage after the main engines flame out, the FMC and inertial navigation system (ADIRU) will continue to operate under battery power for at least 30 minutes, possibly longer. The FMC will continue to navigate a path toward the manual waypoint and will continue to drive the pilot’s fly-to needles even though the autopilot is no longer functioning. The pilot would be relying on hydraulic power from the RAT after the main engines stop, but the navigation equipment would continue to operate on battery power.
* If you want to check, go to the Movable-Type Scripts online Vincenty calculator, scroll down to the Direct Solution box, enter N07 14 38.76, E94 36 40.27 as the start point, enter bearing 186 41 23.46, distance 5384329.265 meters and see where it finishes. Just to point out, 0.0039 of an arc sec is 3.6 inches at S41.00. The Vincenty formula works out distances at ground level, as opposed to distances at cruising altitude; to convert from one to the other, multiply or divide by 1.0017. Waypoint S41.00 E88.00 is about 206NM downrange of fuel exhaustion.
MEFE is assumed to occur at 1730, which is 47.5NM beyond 6th arc, at S37.61, E88.66. A 16:1 piloted glide from 38,000ft would terminate at about S39.20, E88.36 allowing for headwinds, and assuming aircraft maintained the same track.
The search area required is no more than 2,500sq km in size. A 20NM wide corridor centred on the flight path would have to be searched. The corridor would start from where the original 120,000sq km southern boundary finished and would extend 15km downrange of the nominal 16:1 glide impact point.
Total distance covered between 1829 and MEFE works out at 2778NM. Doctor Bobby’s fuel model predicts that 2788NM are possible if the aircraft flew at a constant M0.82 at FL390, from 1829 onwards. Fuel exhaustion is predicted at 0018. The aircraft therefore managed 10NM less than predicted for M0.82 at FL390. The difference is due to the less-than-optimal speed/altitude profile flown in the 28 minutes between 1829 and 1857. The average speed between 1829 and 1857 works out at M0.78.
1941 Aircraft makes grazing contact with 2nd arc at S1.0333, E93.65. Distance covered between 1829 and 1941 is 572.22NM , being 60NM from 1829 to start of FMT, plus 14.14NM for the FMT, plus 498.06NM from FMT until 2nd arc crossing. Assuming for simplicity, a time average tailwind of 1.5Kts between 1829 and 1941. The average airspeeds are M0.78 between 1829 and 1857 and M0.82 between 1857 and 1941.
2nd to 3rd arc: 485.73NM, 485.73Kts avg GS, 1.6Kts avg TW, TAS484.13Kts = M0.82 (Mach1: 590Kts, ) 3rd arc crossing S09.105, E92.71
3rd to 4th arc: 492.13NM, 489.11Kts avg GS, 4.6Kts avg TW, TAS484.51Kts = M0.821 (Mach1: 590Kts) 4th arc crossing S17.28, E91.71.
4th arc to 5th arc: 483.43NM, 484.08Kts avg GS, Zero avg wind, TAS484.77Kts= M0.82 (Mach1: 590Kts) 5th arc crossing S25.29, E90.64.
5th arc to 6th arc: 697.18NM, 466.70Kts avg GS, 13.5Kts avg HW, TAS480.20Kts= M0.82. (Mach1: 586Kts) 6th arc crossing S36.82, E88.803.
6th arc to 7th arc: 61.5Nm, 434Kts avg GS, 33Kts avg HW, TAS467Kts = M0.81. (Mach1: 577).
@Ge Rijn: Understood. That’s also close to where oceanographer Chari Pattiaratchi thinks the probability is the highest.
@Ge Rijn
If you have no results to present, then please shut up! You are a waste of time and nobody apart from Rob takes you seriously.
I would prefer that Victor banned you, but that is his decision.
I am no longer willing to take your shut.
I hope I have made myself clear!
@Rob
Happy New Year!
The drift analyzis are based on factual found (and not found) debris. They are not ‘totally’ off track at all.
Respected drift-studies like CSIRO, Godfrey and Pattiaratchi managed to constrain the crash area between ~30S and ~35S.
This is quite an achievement and it’s also within the constraints of the Inmarsat and fuel data.
You keep refusing to let any of these insights in and stick to your ~38S and beyond assumption like you did two years ago. Things have changed.
Since you told us you’re on the spectrum I understand. But please try to take another perspective once in a while.
@Richard
Calm down.
@Rob
Why no debris in WA?
Regarding drift studies, do consider potential differences in seed locations for debris. Assuming all MH370 debris began their journey at sea at the site of resting location could introduce error, perhaps insignificant, perhaps important.
The photographic evidence of Flaperon damage points toward fatigue failure of primary connections. If so, we don’t know what time, location, or altitude the Flaperon began it’s journey at sea.
The IFE interior frame debris points toward failure stemming from high speed impact. We could safely assume the IFE debris started its drift journey at X marks the spot.
@Victor
If there was a step-change in BFO bias, are you thinking it could be about 10 units and we should subtract 10 from the reported BFO’s to approximate this behavior (if it happened)? That seems to be the approx. bias in the 30S path.
Would you expect Channel C bias (the phone call channel) to drift in the same amount?
I am guilty of assuming no bias change but now that I understand what you are saying a step change (vs. just scatter) I can think about it. Worst case the bias changes for no reason mid-flight. My basis for favoring BFO as-reported is that, allowing for logical pilot input, almost exact match of BTO/BFO can be achieved. But possibly the same thing is true for other case.
@Kenyon
We don’t know how the flaperon’s failures and damage occured. There’s still no official report on this issue. All damage though especially combined with the outboard flap section (and other found pieces) indicate low impact speed damage and seperation on impact with the water surface. All examined (and published) debris show mostly signs of tension damage and hardly any signs of compression damage. Which excludes imo a high speed nose down impact.
The IFE piece also was largely intact and the other two found interior pieces don’t show compression damage either.
Those pieces could have easily escaped the hull if the tail broke away on impact or a door seperated.
A mid-air brake up would have left a tremendous debris field and many bodies floating around. This has not been the case for nothing had been recovered or regarded as worthly enough during the aerial search.
@Kenyon: Good to see you here, Tom.
If the flaperon separated before impact, it would have started its drift journey across the Indian Ocean further north than debris from the impact. I think this would make it even harder to reconcile the timing of the discoveries of the flaperon and “Roy”. That is unless you believe that “Roy” could have separated before the flaperon. That’s something I had not considered, as I have viewed it as unlikely.
@TBill: Again, for my 30S, LNAV path (which is positioned at 6.8N latitude at 19:41), the RMS BFO error for measurements at 19:41, 20:41, 21:41, 22:41, 23:14, and 00:11 is 5.7 Hz, which can be separated into a mean error of -5.4 Hz and an RMS of 2.0 Hz error about the mean, assuming a BFO bias of 150 Hz (for all channels). So, if the BFO bias changed to 154.4 Hz, the RMS error is only 2.0 Hz.
Also, we don’t know the environment of the SATCOM for the entire flight, which may have contributed to a change in bias, especially if the plane was depressurized.
And yes, I would imagine the BFO values for all the channels would experience the same drift.
If there was any in flight separations as many of us suspect, they occurred in the final seconds, not minutes or hours earlier. Therefore, the debris field will be small (few miles at most), and all the debris that has been found originated from points within a few miles.
As Victor noted, any drift in the BFO bias would come from drift in the SDU OCXO, which affects all observations the same.
@All: A new independent study for the location of MH370 has just been released. The prediction is a crash near Christmas Island. I don’t think this is likely for a number of reasons, but some of you might be interested in the work.
@Victor, I too have considered “Roy’s” early separation unlikely, but we just haven’t identified data to support its timing. Perhaps biological data can lead to solving timing issues of debris, that’s not an area I can provide fruitful insight…
@All (except Ge Rijn) I stand by content of previous 10:45am post.
Victor: As we have discussed, this new independent study appears on first glance to be well researched and in depth, but after a more careful read, it is obviously full of factual errors and . Misguid
@VictorI
I’m at least confirmation-biased regarding the debris finds and latest forward drift-studies.
I had a glance but won’t take the time to read it all. I regard it as bogus and I wonder why something like this shows up the day the new search is officially announced.
Typing on a phone is challenging! Should read ..factual errors and misguided analysis.
@ALSM
That’s the same thing you said about my Xmas Island theory. I was in therapy for almost 6 months. I will say the new stuff is beautifully formatted. Do you have any idea what tool they used. Anyone?
https://docs.google.com/document/d/1SG79753eTNK1UsUgq4oaepKRlnR5FK8V4tnbohUWjU4/edit?usp=sharing
@Victor
I think it helps me on BFO to understand a proposed path is a close match to an adjusted BFO data base, rather than a not-so-good match to the actual raw BFO data. I understand we do not know if the bias adjustment needed or not.
If I ever do an adjusted BFO path, I will probably show BFO-raw, BFO-adj, BFO-predicted. What I am thinking as usual is Arc3 to Arc5 is possibly without maneuvers, and seems to be no bias change. With those assumptions, one could get solutions for different assumed biases to help determine what bias fits best. I think 150 works quite well for that period, but maybe 155 works too.
@Ge Rijn: I don’t think there is anything sinister regarding the timing. They have been working on it for some time, and they seem sincere. They recently asked the IG for comment, so we were aware it was in the works. For me, there are assumptions on top of assumptions that don’t seem very probable. I also believe there are some factual errors. If their goal is to re-direct the search to 12S latitude, that is not likely to occur.
I don’t know the details of the OI Test Plan, but I know SC is still headed towards deep water. If they continue on their current path (heading 150) at the current speed (~14 kts), they will reach 19,000 deep water in ~10 hours.
https://goo.gl/djH9DA
@Victor
I am surprised the authors of the new Xmas Island study did not site the work of Mikhail P. who provided an independent input regarding the Xmas Island terminus. I did. Maybe they were afraid of pissing off Oleksandr.
https://sites.google.com/site/mh370tibet/updates/05june2014
Ge Rijn wrote (again, without any credible analysis to substantiate his opinion):
“All damage though especially combined with the outboard flap section (and other found pieces) indicate low impact speed damage and seperation on impact with the water surface. All examined (and published) debris show mostly signs of tension damage and hardly any signs of compression damage.”
This is errant bullshit. Repeating it, ad nauseum, does not make this nonsense any more credible now than when it was first aired. The catalogue of debris items published by the Malaysian investigation team, apart from the flaperon and a one-third span length of the adjacent outboard flap, are all shatttered fragments of composite structures. Shattered by some significant forces.. In addition, I am aware of a number of small debris items found on western Indian Ocean shores, the material and manufacturing characteristics of these items are totally consistent with material used to construct aircraft interior ‘monuments’. The significant indications of damage to the flaperon and flap include spanwise fractures of the trailing edge wedge of each control surface assembly and fracture of its attachment structures. Your reference to compression/tension failure modes indicates a misunderstanding of the analysis that was undertaken of these parts.
There is no evidence, none whatsoever, to support a scenario in which 9M-MRO’s trailing edge flaps were deployed (together with the associated ‘droop’ of the flaperons), that a water landing was attempted, or that the flap & flaperon trailing edge damage is a consequence of dragging or upward pressure from water impact.
The known behaviour of the B777, after loss of turbine driven generator power, is that a left wing down roll (and consequent descent) is initiated: that such a descent became established is substantiated by the BFO values recorded during the 00:19 Log On. The last recorded BFO indicates an excessive rate of descent. This is not anyone’s mere opinion, it’s the result of research into the characteristics of the SDU’s behaviour by DSTG, supported by senior design staff at the SDU manufacturer, and the staff of the SATCOM network operator. The outcome of that research is entriely consistent with the knowledge and experience of a number of well informed contributors to this forum. Apart from the inevitable ocean impact resulting from a high rate of descent, it’s entirely possible that aerodynamic forces precipitated departure of flight control surfaces, or more catastrophic effects, during that descent. The only individual who might be regarded as carrying any credibility, and yet advocates for the contradictory dilemma that accepts catastrophic damage alongside relatively minor damage to debris, will not agree to present any basis for his conclusion.
As for the surface search. Some forthright facts: 20 days passed before any aircraft were credibly tasked to a search. There should be no expectation that anything smaller than a ship could have been discriminated on 17-18th March. Until 28th March, the air taskings were completely in the wrong area. The probability of detection in the period 28th Mar thru 3rd April was variable, but generally very low, tasked over an area of 100s of 1000s of km/sq. That must be compared with the search for AF477: in an area of 75,000sq/km, the first debris find was made in an area that had been overflown 5 times. That first debris find was the galley unit, something the size of a typical englishman’s garden shed. No items were found after 20 days had passed. The MH370 surface search was attempted over too large an area, after too significant a delay, to have held a reasonable probability of detection.
Concerning the surface search, again you have offered an opinion that the EO system equippage on the Orion aircraft tasked to the search must be regarded as certain to produce ‘better’ results than the image records released by the RNZAF & RAAF. The Orion’s EO systems are not superior to ‘Mk1 Eyeball’ and handheld DSLR cameras for wide area search & opportunistic observation. The Wescam and Safire EO systems are designed for persistent observation of prior identified targets, especially in low light conditions. The RNZAF’s own SAR briefing material, previously shared by me, categorically states that the EO systems are not suitable for SAR.
Please, if you wish to engage, give us something more to work with than, simply, your ’opinion’.
@DennisW: They are probably not aware. Perhaps you can give them a heads-up about the previous work. @TimR should also be told about the new work.
@Victor
I assume they are (or should be) following your blog as is Tim R. My own conclusion is that the Xmas Island terminus has a number of very serious problems, and I discarded it a long time ago. I don’t want to risk appearing to evangelize it. I actually resisted commenting on it at all. 🙂
Don Thompson: All well said. Those were corrections that needed to be emphasized. There is zero evidence to support a deliberate water landing.
https://goo.gl/ZLsy6i
@Don T. / ALSM
As for the surface search. Some forthright facts: 20 days passed before any aircraft were credibly tasked to a search. There should be no expectation that anything smaller than a ship could have been discriminated on 17-18th March.
This is a very important point, and I tried to make it earlier. My own discussions with some of the Moffett Field people elicitated the same response, and looking for things off the coast of California is a whole lot easier than the 7th arc relative to Perth.
From The Times, by Bernard Lagan, Jan 5, 2018
Underwater drones start ‘last’ search for MH370
A couple of weeks before Christmas, a diverse group gathered in a London hotel to consider an audacious gamble to solve the world’s greatest aviation mystery — knowing that a thousand-day hunt that had cost more than £80 million had yielded nothing.
Those in the room included representatives of Inmarsat, the London-based global satellite network, Boeing, the aircraft maker, Australian air accident investigators and a little-known American exploration company that has leased the world’s most advanced civilian ocean-survey ship.
They are betting that within 90 days the 8,000-tonne Seabed Constructor can find Flight MH370, the Malaysian Boeing 777 that vanished without trace in March 2014 with 239 passengers and crew on board.
The survey ship, which has a large helicopter landing pad, an array of spherical antennae, a recovery crane that can lift 250 tonnes rising from the stern and, most significantly, a fleet of underwater drones, slipped out of Durban, South Africa, four days ago.
She is now powering east across the Indian Ocean toward latitude 35°S, the region where most of the experts gathered in London say they think MH370 must be lying, north of the previous search area, at a depth of perhaps five or six kilometres.
In about ten days she will begin looking for the airliner, which disappeared after mysteriously reversing course early into its overnight flight from Kuala Lumpur to Beijing.
Seabed Constructor’s operators, the Houston-based sea-floor experts Ocean Infinity, have struck a deal with the Malaysian government that will earn them about $70 million, but only if they find the aircraft. A formal deal is expected to be signed in a few days.
The 90 days they have in which to find the aircraft roughly runs from January to early April, when the southern Indian Ocean’s notoriously foul weather eases enough for ships to stay far out at sea for longer.
Ocean Infinity says that its bet that it can find MH370 in 12 weeks, when the sea-bed search co-ordinated by the Australian government found nothing in more than two years, is not the long shot it might appear.
Based on further studies commissioned by the Australian Transport Safety Bureau, which relied on more precise analysis of MH370’s final hours obtained from sparse satellite data and Boeing flight simulators, there is expert consensus that the lost jet is within the new search zone.
That likelihood was bolstered by the searchers’ first big break — the discovery, 508 days after the aircraft vanished, of a large wing component, a barnacle covered flaperon, on a remote beach on the island of Réunion in the Indian Ocean.
The Australian investigators obtained another 777 flaperon from Boeing and set it adrift to determine the speed and direction it moved in ocean currents. Working backwards from where the piece was discovered, they found that it had almost certainly drifted west across the ocean to the island, from where the new search will be concentrated.
In addition to starting with much better intelligence, Ocean Infinity says it has a far superior armoury than the previous search. That relied primarily on cumbersome deep-tow sonars tethered to ships with cables up to 10km in length. While reliable and able to transmit sea floor images in real time through the cables to experts on board the surface ships, the sonars worked laboriously, scanning only about 500 sq km of sea bed a day.
Seabed Constructor will search the seabed with six unmanned, untethered submarine-like search vehicles that can cover a combined 1,200 sq km of sea floor a day and travel to a depth of 6km. Each vessel is six metres long and has a titanium sphere to protect its array of electronics from the pressure of the deep ocean.
The lithium-polymer battery packs should allow the fleet to spend about 60 hours scanning the sea floor before being hoisted back on board the Seabed Constructor. They have sonars that emit sound waves to detect ocean floor objects and metal in particular, magnetometers to detect magnetic fields and high-definition colour cameras. The data they collect will be downloaded each time the search vehicles are lifted back on board.
If MH370 is where the team believes, the world should soon find out. If their search fails, it may well be the last.
From The Australian by Ean Higgins, Jan 6, 2018
MH370: Ocean Infinity attempts to solve the mystery
Sometime soon, maybe next week, an extraordinary, futuristic scene will play out in a remote swath of the southern Indian Ocean in a new quest to find the holy grail of aviation and unlock its mysteries.
An impressive but somewhat bizarre-looking ship with a massive helipad cantilevered over its bridge and a gigantic crane on the rear deck will stop in its tracks.
One after another, the crew will launch eight yellow unmanned boats and eight orange torpedo-like unmanned mini-submarines. The robot boats will stay in contact with the robot mini-submarines as they dart around on their own a few kilometres deep, hunting for the remains of an airliner that disappeared four years ago with 239 souls on board.
And with that, a new, audacious capitalist bid to find Malaysia Airlines flight MH370, after a previous search run by Australian bureaucrats failed, will be on.
The passage of time has done nothing to dampen the world’s fascination with the enigma of how a modern airliner on a regular scheduled flight could just disappear.
The Boeing 777 took off from Kuala Lumpur airport after midnight on March 8, 2014, headed for Beijing. About 40 minutes into the flight something very strange happened: radio contact ceased, the plane’s secondary radar transponder was turned off, and it turned around, flew a course near the Malaysia-Thailand airspace border, turned north up the Strait of Malacca, then headed south on a long track to the southern Indian Ocean.
At the request of Malaysia, the Australian Transport Safety Bureau led a two-year underwater search of a 120,000sq km target zone that ended a year ago without finding a trace of the aircraft. That failed search cost $200 million, paid for by the taxpayers of Australia, Malaysia and China.
Now private enterprise, with American can do-ism coupled with British and European technical skill, is stepping up to the plate.
The people who run Ocean Infinity — a crack international team of engineers, information technology experts, hydrographic surveyors, underwater robot submarine experts and others — and its shareholders have taken a big gamble on their own ability.
A few months ago Ocean Infinity, based in Houston, Texas, and London, put a juicy offer to the Malaysian government, which under international law is ultim¬ately responsible for the investigation into the loss of the aircraft.
Ocean Infinity proposed to launch a new hunt for the aircraft on a “no find, no fee” basis — the Malaysian government would agree to pay a sizeable fee if the company found it. But if no wreckage were found, Ocean Infinity would have gone to all that effort for not a penny’s compensation.
Malaysia’s talkative Deputy Transport Minister Abdul Aziz Kaprawi gave some indication of the size of the deal. The government’s cabinet, he said in October, had agreed “to prepare a special allocation to the Ministry of Transport amounting to between $US20m ($25.46m) and $US70m if MH370 aircraft wreckage is successfully found within 90 days”.
So keen are the people who run Ocean Infinity to get started that this week they ordered the captain of the vessel they have leased, the Seabed Constructor owned by the Swire group, to set sail from Durban in South Africa for the new search zone, even though a final contract with the Malaysian government has yet to be signed.
It looks like a goer, though.
On Wednesday Malaysia’s Transport Minister, Liow Tiong Lai, told journalists: “We are making preparations and we will announce it next week, after we finalise the contract.”
Putting them side-by-side, what Ocean Infinity is about to do makes the previous hunt organised by Australian bureaucrats look like kids’ stuff. The ships in the ATSB-led search used a single tethered “towfish” at a time, or a single mini-submarine, known as an autonomous underwater vehicle. One ship managed to crash its towfish into an underwater mud volcano, breaking the towline, and it was lost for a while until recovered.
Beijing’s contribution to the ATSB-led hunt was a Chinese government survey vessel that was supposed to deploy a single AUV. But that vessel, the Dong Hai Jiu 101, hardly ever did any actual searching; as revealed by The Australian, it spent most of its nearly year-long deployment docked in Fremantle or just offshore, probably spying on Australian submarine and other military movements, according to security experts.
By contrast, Ocean Infinity sounds like a pretty serious mob. Its website describes the company as “ocean explorers mapping the unknown”.
“We go to unmapped locations to survey the seabed using the most advanced fleet of autonomous vehicles in the world.”
David Griffin, a CSIRO drift-modelling scientist who led a project commissioned by the ATSB to work out where MH370 came down based on where a few bits of the aircraft were washed up on the other side of the Indian Ocean, met the Ocean Infinity people in London last month to brief them on the first search.
“It’s a very impressive organisation,” Griffin told Inquirer. “They have got terrific equipment.”
Ocean Infinity plans to use eight AUVs at a time on independent search missions, enabling it to scan the seabed for MH370 much faster than in the first search.
They are mind-blowing machines. In their 6.2m bodies they carry side scan sonar, a multi-beam echo-sounder, a sub-bottom profiler, a high-definition camera, a conductivity/temperature/depth sensor, a self-compensating magnetometer, a turbidity sensor, and a methane and laser sensor.
The most important pieces of equipment for this search are the side-scan sonar, which can create images of things on the sea floor and distinguish metal; the multi-beam echo-sounder, which can produce three-dimensional im¬ages; and the magnetometer, which can confirm if objects are metallic. The AUVs can message the surface robot ships known as unmanned surface vehicles about what they are doing, using acoustic positioning telemetry. They have on-board cameras and machine-vision software that can enable them to keep an eye on what’s ahead of them and dodge underwater cliffs or other obstacles.
The batteries can keep them going underwater for up to 60 hours — 2½ days.
The issue, however, is this: no matter how sophisticated the technology, it can’t find what it’s looking for unless the searchers have the right strategy of where to look.
Most expert observers think that had MH370 been in the ATSB’s 120,000sq km target zone, it would have been found; while the process was slower than what Ocean Infinity is planning, it still used high-quality technology that picked up things such as two shipwrecks with surprisingly high-definition.
The question being asked in the air crash investigation and airline pilot community is whether Ocean Infinity has the right assumptions as it heads into its new search. The plan is to look in a new 25,000sq km zone immediately to the north of the previous one.
It was, in fact, the ATSB and a panel of Australian and international experts including Griffin who had identified this zone as promising even before the old search had been brought to an end.
There are three main clues to where MH370 may be.
One is automatic hourly electronic “handshakes” between transmitters in the engines of the Boeing 777 and an Inmarsat satellite, relayed to ground stations. While not really designed to track aircraft in this way, experts think the seventh handshake shows the arc MH370 was travelling on the long final leg south. That “seventh arc” shows the final track but not the point on it where the aircraft came down.
But that is where the second main clue comes in. Years after it went down, bits of MH370, some quite large and intact, started washing up on the other side of the Indian Ocean on the coast of Africa and the islands off it, including a flaperon and a flap, which are movable parts of the wings.
Griffin and his team did a painstaking analysis of the drift modelling, even building models of the flaperon with adjustments made to reflect damage on the real one, and put them in the ocean and tracked them to see how they reacted to wind, waves and current.
That study determined that the aircraft came down just north of the completed search area.
There’s one other clue that turned up in August last year.
Four satellite images of what could be debris, taken two weeks after MH370 went missing, have been re-examined, prompting Geo¬science Australia and the CSIRO to home in on two narrow strips, no larger than 10km to 30km each. Their report places the likeliest location of the aircraft “with unprecedented precision and certainty” at 35.6 degrees south, 92.8 degrees east.
The satellite images contain up to 70 objects, of which up to 12 were “possibly man-made”, ¬according to Geoscience Australia. Their dimensions match some of the plane debris that washed up on African beaches last year.
Griffin says while these photos have been available all along, the concentration was on the southern search area, and when the decision was made to look to more northerly scenarios, the ATSB asked to re-examine them in high resolution — a slow process.
Griffin has seen the images and says you can clearly see “large, white objects” consistent with debris from “an aircraft hitting the ocean and breaking up into parts”.
“It’s hard to see what sort of other marine objects these might be,” Griffin says.
The part of the seventh arc where Ocean Infinity is going to start searching is, in fact, consistent with this new clue: The Economist got the first interview with the company about the search this week and reported the plan is to begin at about 35 degrees south and work north from there. Because Ocean Infinity thinks it can survey 1200sq km a day using the eight AUVs, the maths work out that the 25,000sq km search zone could be covered in three weeks.
So does Griffin think this hunt will find MH370? “We think there is a very good chance,” he says. “If you are prepared to take the risk and search, this is the most likely place. How likely, I can’t estimate.”
There is, however, one nagging doubt in the professional aviation and air crash investigation community: the ATSB’s theory of what happened on MH370 and how it came down, which it used to determine the old search area and the new one. The ATSB worked on the basis that at the end of the flight the pilots were incapacitated and the aircraft crashed down fast after fuel exhaustion.
Most professional pilots believe the evidence points to Captain Zaharie Ahmad Shah hijacking his own aircraft, depressurising the cabin to kill the passengers and crew while he was on the pilot’s more extended oxygen supply, flying the aircraft to the end, and ditching it in a deep spot of his own choosing to sink it in as close to one piece as possible.
They have now been joined by two of the world’s best known and most experienced air crash investigators: American John Cox and Canadian Larry Vance, who have examined photos of the flap and flaperon, which were mostly intact but had damage to the trailing edge. Vance, who is writing a book on MH370, believes this evidence clearly points to a ditching with flaps down.
He says had the plane crashed down rapidly as the ATSB says it did, they would have been pulverised, as was found in one high-profile investigation he helped lead, Swissair Flight 111, which in 1998 crashed off the coast of Nova Scotia after an on-board fire.
“In the Swissair crash the hydrodynamic forces of hitting the water basically exploded the aircraft from the inside out,” Vance told Inquirer.
When Inquirer sought comment from the ATSB on Vance and Cox’s analysis and whether it still believed its “ghost flight” and “death dive” theory, its spokesman Paul Sadler referred this and other inquiries to the federal government’s Joint Agency Coordination Centre, whose spokesman said such questions should go to the Malaysian government.
Cox does not think that Ocean Infinity should not be conducting the search; even if the ditching theory is right, Zaharie would have flown it somewhere, and the drift modelling and satellite photos show that may well be where the AUVs are about to look.
“Ocean Infinity has ideas about the location of the aeroplane,” Cox told Inquirer. “There is one certainty: we won’t find the aeroplane if we don’t look for it. Therefore, I’m glad they are looking.”
@ALSM, “…they occurred in the final seconds, not minutes or hours earlier. Therefore, the debris field will be small (few miles at most), and all the debris that has been found originated from points within a few miles.”
Mike, while I agree with the likelihood of debris timing and proximity, just as you expressed above, I’m just offering up the concept that starting points for all debris does not necessarily mean they journeyed from the same spot. It then becomes a matter of whether the differences in debris starting points are material to the drift studies. You indicate that a few miles does not matter, I would agree. As I stated in my original post above perhaps insignificant, perhaps important.
We do have EOF scenarios that generate conditions that could produce fatigue stresses but those EOF scenarios do not exclude other potential (undiscussed/unconsidered/unexplored)scenarios from having occurred. I’m unaware of available data that fixes the time, location, or altitude of when the Flaperon separated from MH370 whether it was indeed the final seconds or some time prior.
@Victor
I don’t care what anyone says, I like Higgins.
I do believe that Higgins is messing up the payment terms by referring to a questionable statement supposedly attributed to Kaprawi back in October of last year. Likewise, Lagan’s reference to $70m lacks credibility.
No matter. Don’t mean to get hung up on terms. It is just interesting to me, and probably to no one else.
Kenyon: I think we are in agreement. I was just trying to emphasize that parts of the airplane were not dropped along some extended path of hundreds of miles leading up to impact. Some people have suggested the flaperon could have separated hours before impact. That is very unlikely. I believe the evidence suggests the flaperon and flap segment may have separated in flight during the final 1-2 minutes.
@Don Thompson
Well said. Thank you.
Higgins! No longer any mention of his former pet pundit Byron Bailey? Now Larry Vance is his ‘swami’.
“In the Swissair crash the hydrodynamic forces of hitting the water basically exploded the aircraft from the inside out,” Vance told Inquirer.
That’s not an unreasonable statement. However, the dilemma (at the risk of repeating myself) is that most recovered parts of MH370 do evidence catastrophic damage resulting from “hitting the water”, whereas two parts do not demonstrate the extent of catastrophic damage expected of forces involved in an impact with the ocean.
I’ll remind readers of a previous post where I made reference to two images with which Mr Vance should be familiar. Those images depict debris that is consistent with destruction resulting from a single impact.
The MH370 options are:
a) single impact destruction (not consistent with debris);
b) progressive stages of destruction (consistent with debris);
c) limited energy impact (not consistent with debris);
d) something else.
@Don
d) something else.
Planted debris. Just trying to be helpful.
@all
Well, it (the search strategy) is out of our hands. I am really looking forward to the next couple of months relative to so many different things. Grab a sack of popcorn, and let the show begin.
@DennisW
RE: “DennisW says: January 5, 2018 at 4:57 pm
@Don
d) something else.
Planted debris. Just trying to be helpful.”
Given your penchant for cutting edge witty sarchasm, is your remark purely “Tongue-in-Cheek”, or a hint ?
Are you refering (obliquely) to this ?
https://www.nbcnews.com/storyline/missing-jet/missing-malaysia-airlines-jet-had-ground-collision-2012-n47706
https://media3.s-nbcnews.com/j/newscms/2014_10/238051/140308-malaysian-jet-collision-0912_8785483b83a985db9ae72716530f46ff.nbcnews-ux-2880-1000.jpg
https://pbs.twimg.com/media/BiSZHgEIIAIUyRn.jpg
https://twitter.com/MarySchiavo/status/442647017668759552
@Ventus
Purely “tongue-in-cheek” is this case. I put planted debris in the same category as alien abduction.
@ventus45: At this point, if you seriously think the debris was planted, you are in the wrong forum.
@Victor. Your interview with Marnie O’Neill:
http://www.news.com.au/travel/travel-updates/incidents/the-mindblowing-technical-capabilities-of-new-mh370-search-vessel-seabed-constructor/news-story/1b57123c6ad8408ed6667cdf809e1fcb
@Victor
Thank you for thinking of me. I have been following your blog as Dennis mentions and have seen their “comprehensive” study.
I was made aware of the flight within days of it going missing and that Shah’s intention was to land in Indonesia, release the passengers and place himself at the mercy of the Indonesian authorities.
This flight path does not really fit.
Re: Xmas Island paper
Give them credit for trying to argue for piloted flight (by unknown “people in command”) and also they document the Xmas Island argument.
It is sort of like a modified TimR’s path: they spend a lot of words getting out to IGOGU but end up adopting Ianello’s path including offset and descent past IGOGU. Then it is more like my thinking where the descent is only to FL270 for flight path rules, and then they circumvent Indonesia radar by going over to MEKAR. They claim Indonesian civil radar ends at MEKAR where MH370 could reactivate XPonder and TCAS, and somehow MH370 was content to appear on Sabang military radar as one of many unidentified blips, using the intersection of three FIR boundaries as a misdirection opportunity.
But OK we knew about this general path proposal for some years. Here it is documented with some decent support logic and shows BTO/BFO calcs. Flight path N571 is quoted as N751 more times than I can count.
CORRECTION:
They claim Indonesian civil radar ends at MEMAK (not MEKAR)
@TBill,
I thought the new paper was very well done. I have not looked at every nook and cranny. Anyway, I was impressed with the effort, the detail, and the format/presentation.
@Don Thompson
Thank you for your posts on the wreckage and the efficacy of the surface search, Don. I’d given up on repeatedly rebutting the same nonsense.
@David
I seem to recall that you were trying to calculate the volume of fuel in the APU fuel line between the left tank inlet and the APU itself. Can I ask what you came up with please?
Richard & Victor,
The reason for asking about drogued v. undrogued is that the limited experiments done by the CSIRO their first paper with a replica “Roy” piece showed that it followed the undrogued drifters more closely. Drogues were not even mentioned in Richard’s Feb 2017 paper.
Also of concern is that drift models are a lot of work to create (I haven’t tried), and of those that have been constructed, has there been any serious attempt to intercompare them to determine if they are reliable? With a model constructed from the GDP data alone, a standard method would to divide the data into two sets, use one set to construct a model, then use the second set as “test particles” to see how well their properties were predicted, and then base one’s error estimates on the second set alone. I doubt there are enough buoys to do that in a meaningful way, but maybe it could be done.
Anyway, even with the above concerns, I would agree that any search attempting to be reasonably complete woud extend to North of -30. How much further North is a bit difficult to answer.
DennisW,
You might be interested to know that I have now had a paper accepted for publication in which I cite another paper and explicitly mention that it used “methods of Geometric Algebra”. (I also say that one hits a wall doing so, and suggest one approach the problem from a different direction, which is what the paper is all about. Not relevant here, though, which is a good thing, because even the referee didn’t understand what I was doing.)
@sk999
I concluded that I was too old and rooted in Gibbs to be comfortable with geometric algebra. I think it is the wave of the future, but strangely, talking to my daughter’s professor colleagues it seems none of them has even heard of it. In any case my own colleagues suggested that I STFU, and I took their advice.
Don: Thanks for your summary! I’m curious about this: “In addition, I am aware of a number of small debris items found on western Indian Ocean shores, the material and manufacturing characteristics of these items are totally consistent with material used to construct aircraft interior ‘monuments’.”
I assume that you mean debris items beyond the official “found, but not proven” list — are these ‘monuments’ interior bulkheads and panels? Or are you suggesting they are fragments of the floor beam, which likely accounts for the greatest mass of composite used in the 777?
@Don Thompson
No published official report on the found debris ever made a conclusive statement on how damage and seperation of the found pieces occured.
The only positive statement I know of that comes close is the statement the ATSB made on the outboard flap that it was most probably retracted when it seperated. Which is not a conclusive statement either.
So we are left to make our own interpretations/opinions based on limited material.
Based on this limited material you and others have formed the ‘opinion’ that the damage and seperation pattern of the pieces was caused by high speed impact and/or aero dynamic forces. And you have some strong arguments I see.
I and others have formed the ‘opinion’ the damage and seperation pattern of the found pieces is better explained by a relatively low speed, ~wings level, nose up, ditch-like impact. And I think I and those others who think the same have strong arguments too.
I spent a lot of effort over the years interpreting the debris and their damage with all possible information available to me.
It’s not an opinion based on nothing, like the way you ridicule this opinion away.
We’ll see. Hopefully within three months we have our answers.
And I don’t mind who turns out to be right.
The objective of all is finding the plane.
@VictorI @TBill
I’ve probably been too harsh on this new paper.
It’s just that every paper that ignores the facts of the debris finds and it’s locations I cannot take serious.
@Don T
Higgins! No longer any mention of his former pet pundit Byron Bailey? Now Larry Vance is his ‘swami’.
Vance and Cox are air crash “experts”. Where is your respect? I am a California certified life coach. I could help you.
@Mick. Re, “..the volume of fuel in the APU fuel line between the left tank inlet and the APU itself”.
The APU fuel line is depicted below, inside a shroud. Andrew confirmed a line OD of an inch, 25.4 mm, and supposed an ID of 20mm. It is of flexible rubber and kevlar and Andrew’s implied thickness of 2.7 mm looks to be consistent with that.
A commonly supposed length is 100ft, yielding a volume of 9.575 litres, 584 cu in. If a fuel density of 0.8 is assumed the line would hold 7.66 kg, 16.9 lb.
Not all that need necessarily be available. I have questioned (ad nauseam) whether at high altitude and with the nose down the APU would draw fuel through the line without the APU DC pump being functional, that is, there being residual fuel available from the left tank to displace it.
That in turn depends on how much of that would be accessible to the APU pump in likely aircraft attitudes and after taking into account its consumption in left engine relights/attempts post LEFE, which the ATSB has agreed were possible, and before the 7th arc log-on.
Nevertheless I can offer no alternative explanation as to how, unmanned, the 7th arc log-on would have been prompted.
https://www.dropbox.com/s/t0y366qqakaqnwi/APU%20fuel%20line.pdf?dl=0
These 3 pages are applicable to the -200ER.
@Ge Rijn
Actually the Xmas Island terminus is very consistent with the Geomar reverse drift analytics. I know you are not a fan of reverse drift.
ALSM wrote on January 5, 2018 at 1:59 pm
I don’t know the details of the OI Test Plan, but I know SC is still headed towards deep water. If they continue on their current path (heading 150) at the current speed (~14 kts), they will reach 19,000 deep water in ~10 hours.
https://goo.gl/djH9DA
Well, looks like your guess has come true!
From the moment the vessel left Durban there was an intent, the course of ~150T has been maintained throughout, and daylight is arriving as the water depth approaches 18,300+ feet. Was this to be a preplanned “treasure hunt”?
Mick/David: I agree with Dave’s fuel line capacity estimate. It is in the ball park for sure. 16.9 lbs would provide about 7-8 minutes of APU power, according to the APU consumption rate reported by ATSB and Boeing (2.2 lbs/min). That is more than enough time for the IEF logon expected by ~00:21.
@Mick. My 5th para for APU pump please read APU DC pump.
@David
@ALSM
Thank you, gentlemen, for those responses.
@Mick. No problem. I do not know why you asked but if related to the 7th arc log-on, please bear in mind that at LEFE the APU DC pump would start, delivering 3150 lb/hr to the left engine. Whether that engine relit or not, that supply would continue. The APU would not auto-start until the left engine dropped to idle, when AC power failed (assuming the left was the last to fail).
At 35% N3 the (RR) left engine would shut down and with it the APU DC pump. So IMO the APU then would be dependent just on the fuel in the line and any residual fuel, which might be nil, it could draw past the stopped APU DC fuel pump and various valves, when it might be sucking air from elsewhere.
As ALSM has said the line fuel would be adequate for APU start and a sufficient run to IFE SMS/email connection if the APU could draw fuel from its line with the APU DC pump stopped and, quite possibly residual tank fuel exhausted or unavailable.
It can be argued that all this is unimportant given the 7th arc descent rates and their decisiveness on search width.
However that is not all there is to it. Neither APU autostart or engine relights were part of simulations yet it is the simulations’ 15NM descent diameter that is the basis of the 25 NM search width.
David: Re “However that is not all there is to it. Neither APU autostart or engine relights were part of simulations yet it is the simulations’ 15NM descent diameter that is the basis of the 25 NM search width.”
That statement is not true. We observed the post FE APU autostart in ALL of our simulations, and in one, we observed a 2 second relight of the #1 (left) engine. Also note, the search width is 50 nm, not 25 nm.
Seabed Constructor reached deep water a few hours ago. It stopped at 2018-01-06_0900 UTC in water about 17200 feet deep. The location is -37.64646 36.87403.
https://goo.gl/bAjsUZ
@ALSM Yes search width 25 NM either side.
As to my statement about APU start and engine relights not being part of simulations I draw that from the ATSB’s Search and Debris Examination 2nd November, 2016 Report, “It was not possible to simulate all likely scenario conditions due to the limitations of the simulator. Specifically, flight simulators are unable to accurately model the dynamics of the aircraft’s fuel tanks. In the simulator, when the fuel tank is empty, zero fuel is available to all systems fed from the tank. However, in a real aircraft, various aircraft attitudes may result in unusable fuel (usually below engine/APU inlets) becoming available to the fuel inlets for the APU/engines. If this resulted in APU start-up, it would re-energise the AC buses and some hydraulic systems. This could affect the trajectory of the aircraft. Similarly, the left and right engines may also briefly restart, affecting the trajectory.”
@Mick. Whereas I said that the APU DC pump would stop with left engine shut down, on consulting my earlier notes again I see that in fact it will stay selected once the APU is running though whether it will be pumping is another matter; as is whether that matters anyway.
@David: For the reasons you mentioned and other reasons, I take all the simulations with a grain of salt. There are so many unknowns about configurations and modeling inadequacies that the likelihood of using the simulations to accurately model what actually occurred is close to zero. That’s why I took a much different approach. I assumed the rates of descent predicted by the BFOs are correct, and determined what flight dynamic conditions would cause this. The answer is either a banked descent, caused by lateral asymmetry with or without pilot input, or a nose-down pilot input. Either way, the impact occurred soon after the last SATCOM transmission, unless there was a drastic change in configuration just after the last transmission.
The cause for the lateral asymmetry could have been many things, include the unpowered left flaperon under RAT power, an engine restart, rudder out-of-trim, reconfiguration of hydraulics after APU start up, etc. I think Boeing under-played the significance of the 8-second time spacing between the descent rates of 4,600 and 15,000 fpm, which would not resulted from an almost wings-level phugoid. Rather, this is an indication of a trajectory that, barring skilled pilot intervention, soon after hit the water.
Victor:
I think there is a lot of misunderstanding and misinformation floating around about Level D simulators. In particular, the notion that they can’t be trusted outside a narrow range of normal operations is not true. Simulated flight beyond extreme conditions (inverted for example) is of course questionable, but as long as the simulation remains within reasonable bounds, the physics will be well modeled. The physics don’t break just because the plane rolls to 60 degrees. In all our simulations, the airplane started to bank shortly after FE. I have no doubt that is what happens in real 777s. The bank angle increases with time in all the sim’s. There is no doubt that also happens in real 777s. You can argue that the precise details may start to diverge from real life after reaching 60 degrees or so, but the fact that all the sim’s indicate this pattern of increasing bank angles must be accepted as typical behavior.
@ASLM: I am not referring to the simulators’ modeling of flight dynamics for out-of-envelope conditions. Rather, I am referring to the shortcomings of the model at a systems level and unknowns about the configuration of the aircraft. I believe the BFO values indicate that either lateral asymmetry or nose-down input caused the plane to enter an increasingly steep descent. What caused the descent will be very difficult to ascertain without recovering the FDR. In neither the Boeing simulations nor in your simulations do I see indications that the descent increased from 4,500 fpm to 15,000 fpm in 8 seconds. (Please correct me if I am wrong.) However, with sufficient roll rate or nose-down input, this is exactly what would occur, and the impact would occur soon after.
Victor:
Yes, I agree with your clarification. For example, the fuel tank models may not work well for residual fuel after nominal FE. But the basic flight dynamics should be reasonably accurate all the way up to speeds of ~mach 0.9 and bank angles to at least 60 degrees. Theoretically, the 0019 decent rate could be the result of either pilot input or uncontrolled flight post FE. Either way, it is close to the 7th arc.
@ALSM: Yes. And if the plane reached M0.9 and a bank of 60 degrees, it doesn’t matter if the rest is accurate. As you say, the plane crashed close to the arc. I think even closer than the Level D simulations suggest, so I view them as being conservative (assuming no extraordinary pilot inputs after the last transmission).
@VictorI @ALSM
If you are right about a close impact along the 7th arc till ~30S, at least we will know this first during or after the second stage of the OI search, when the CSIRO areas/hotspots have been searched.
I see it’s the only sensible way to go, working inside out of the 7th arc based on your data and assumptions.
I myself expect them to find nothing in both areas and if so I hope they’ll have enough time left to search outside the +/-25Nm limits based on extraordinary pilot inputs after the last transmission.
But I hope they succeed based on CSIRO’s and your assumptions. This would only shorten the search and make it a lot less complicated.
@Ge Rijn: I don’t know if 30S is far enough north. I would consider proceeding at least until 26S before re-evaluating the search strategy.
@Victor
@Ge Rijn: I don’t know if 30S is far enough north. I would consider proceeding at least until 26S before re-evaluating the search strategy.
Completely agree. I would put the probability of finding the plane with that approach at 70%+.
For some reason I still worry about the accuracy of the arc location. Yesterday I went through the math again very carefully (even stupidly considered the Sagnac effect and atmspherics – negligible on both counts). It is too simple to screw up. Right?
@DennisW: The location of the arc has been checked many times by many people, so the possibility for math errors is essentially zero. We know the position of the satellite. We know the geometry of the earth. We know the speed of light. We know the position of the ground station at Perth. We know the location at KLIA where the calibration constant was determined. Unless the calibration changed (and we have no evidence that would happen), then the location and uncertainty we assign to the 7th arc should be correct.
@Victor
Yes, I concur with all that. The physics certainly seems solid as a rock.
I wish at least one of the 20 or so previous flights of 9M-MRO checked by the DSTG had an inflight SDU reset. Also the DSTG book did not explicitly state (I could not find it) that BTO bias was the same for all the flights. My concern being that the BTO bias might be off by +/- an internal clock cycle in the SDU depending on how a counter chain comes to life.
Victor is right. The evidence is very strong for an end point close to the arc, but weak for any specific point along the arc. The original null search area and new drift analysis strongly suggests that the actual navigation mode was not one of those 777 pilots thought was the most likely mode, and thus adopted by ATSB, IG and others. Once that constraint is removed, we get a wide range of points along the 7th arc as Victor has demonstrated, possibly as far north east as S26 (or further?).
In hind sight, it is unfortunate that ATSB chose to widen the search at the southern end. If instead the search width had been limited to +/- 20nm, more of the arc to the NE would have already been searched and MH370 might have been found already. That is not a criticism of ATSB’s choice. Given all the conflicting inputs they had to juggle in 2014, their decision is understandable. Fortunately, we have more confidence in the BFO data now, and Ocean Infinity has been well briefed on the reasons for concentrating the search closer to the arc.
Interesting to see that Blaine Gibson hasn’t lost interest [as long as he’s in areas more law-abiding than Madagascar. He visited Ocean Constructor while it was in Durban. [https://www.iol.co.za/ios/news/mh370-south-africa-holds-key-information-12631001
BG will apparently go on to Australia to lead some beach-combing with Dr. Pattiaratchee, hoping for gifts from the gyre.
urgh! I meant Seabed Constructor of course!
DennisW,
We have BTO data for one power-on reset, at 12:55 UT, through the POR satellite. I am probably the only one who modeled the POR data, and don’t think I ever put out a figure showing the entire time span. However, it would be exceedingly uninteresting – the BTOs show normal jitter with no jumps in the residuals from 00:51 UT (before takeoff of MH371) up to 16:00 UT (passengers boarding MH370). One single bias offset throughout; LGA or HGA – makes no difference.
@Mike
Well, that is a valuable observation IMO. I wish the DSTG would have published the BTO bias for all the flights it looked at. It should not change based on physics. Perhaps I am being a worry-wart on the internal hardware design of the SDU.
@Gysbreght
Thanks for your comment over on JW. I am no longer able to post there.
Sorry, for the interrupt Victor.
@DennisW: Recently, I had the chance to interact with one of the developers of the Bayesian model for path reconstructions. One of the things we talked about was the absence of a “hold” manoeuver in the Bayesian model, which could be described by a single random variable (the time duration of the hold). He acknowledged that a “structured” manoeuver of this type would be a next step in developing the model, and also acknowledged that incorporating the hold manoeuver would yield a rather flat PDF, especially if the BFO constraint at 18:40 was also relaxed.
@Victor
I think that would certainly broaden the PDF to the North. My sense is that the result would be a very large area, and not extremely useful. Of course, you could say the same thing relative to all the work that has transpired lately. I think we have basically conceded that the information we have allows a broad range of terminal locations. That is mitigated by the OI search rate capability. I hope their (OI’s) stuff works as well as we think it will.
@Victor
Sounds like you are having second thoughts on rapid descent at 18:40. But there is really an intermediate case that would make sense, and that is (similar to the Xmas Island paper) changing heading about SAMAK more westerly, and then with less rapid descent. Nederland shows this case in his path. Any possibility that helps?
DennisW: Re: BTO bias drift…there is no known mechanism by which BTO drift is possible. The digital clock jitter is now exceedingly well understood, and that caps the 7th arc uncertainty. That’s the only “meaningful noise” in the BTO measurement, and it is very deterministic with zero long term drift/change. Temperature effects are also zero. There is nothing in the SDU that can cause the internal time delay to “drift” like a an OCXO can. And the max delay under all circumstances can be no longer than 200 usec by design. In summary, the BTO bias is very, very constant.
@ALSM
Yes, that was my assumption as well. What I was concerned about specifically is that non-synchronous counter chains can wake up in more than one state. “Drift”, temperature effects, etc. were never a concern – simply different fixed BTO offsets based on initial internal conditions at a power up.
@ALSM
Also, to be clear, when I referred to atmospherics I was not referring to time delays due to changes in light speed in to the atmosphere (although I did calculate that as well just for completeness). I was referring to refraction which according to simple calculations shows ring errors on the order of 10km (depending on elevation angle, of course). Not alarming, but someone else should check it.
@DennisW & @ALSM
@DennisW
I raised the refraction issues with ALSM a long time ago (July 2015) and he dismissed it out of hand. I still think it is worth further study.
http://nvlpubs.nist.gov/nistpubs/jres/69D/jresv69Dn3p395_A1b.pdf
@Victor Re: “A plausible trajectory for MH370”
I wish to withdraw a comment I made yesterday re the study, specifically “This flight path does not really fit.”
Flying between waypoints is the way to go and the final ditch position is certainly plausible.
However, my eyes had glazed over earlier on reading of all the manoeuvring to avoid FIR boundaries.
In my opinion Shah would not have been overly concerned with avoiding detection after he was out of Malaysian radar range.
I did not fully read it right through before commenting and for that I apologise.
@DennisW & @ventus45:
After decades leading programs and companies like GPS/MET, COSMIC, AMSC and Synergetics (all involving in L band communications and/or navigations) , I can say with confidence that refraction in the L band at elevation angles of 40 degrees has no significant effect on the BTO observations. You are talking abut meters, not miles. You have to be at very low elevation angles (grazing the atmosphere) to have any material refraction in the context of BTO delays. (For the radio occultation GPS/MET payload on MicroLab-1 and 6 COSMIC satellites, we were measuring the bending angle to derive atmospheric profiles of temperature and moisture in the neutral atmosphere, and and TEC in the ionosphere.)
@DennisW & @ventus45 If you are concerned about delay at a 40 degree elevation angle, look at the group delay due to the ionosphere and the water vapor in the neutral atmosphere. But again, you are talking about mm delays, not miles.
@ALSM
You are correct. I made a simple error.
If there are any doubters still out there, consider this: We can get ~10m GPS accuracy using L1 only receivers. That would not be possible if the propagation delays varied by many microseconds.
@DW: bummer man, it takes so much to get JWs ban-hammer.
First, he came with it for the psychics’
Then he came with it for the Russophiles.
Now, he hammers our resident whacko-detector!
@ALSM
If there are any doubters still out there, consider this: We can get ~10m GPS accuracy using L1 only receivers. That would not be possible if the propagation delays varied by many microseconds.
True, but… The GPS system has one pass through the atmosphere at nominally ~20,000km. The ranging in the ISAT case has two passes through the atmosphere at L-band, and two passes through the atmosphere at C-band. Each pass at ~36,000km. So in the GPS case you are looking at ~20,000km. in the ISAT case you are looking at ~144,000km. Still, as I said, I made a simple error.
Using the published results of the Chinese AIS transponder ranging system which is more representative of what ISAT is doing, one finds errors that linearly increase with range with typical means of 8m at 100km. This system operates at near 160MHz. If the linear relationship held with range one would expect errors of some 2800 meters at 36,000km.
While I don’t expect errors on the order of 3km in the ISAT case, the errors will be nowhere near 10m.
DennisW: I have no idea what you are talking about. All the delays that do occur do so in the few hundred km above the earth’s surface. There is no excess delay in free space. It is all caused by water vapor and TEC. The delays are on the order of mm and meters at L or C band.
@ALSM
Correction the 8m number above is the standardard deviation, not the mean. Duh.
Take a look at figure 8 in the link below. The L=1 line is the “single shot” measurement error.
https://www.hindawi.com/journals/js/2016/6928961/
@TBill: I still think it is possible that at 18:40 the plane was traveling northwest and in a descent of around 2500 fpm. By removing the BFO constraint at 18:40, the Bayesian model would allow this and other possibilities, which would result in a larger range of possible latitudes at 19:41, and ultimately at 00:19.
http://www.navipedia.net/index.php/Ionospheric_Delay
http://www.navipedia.net/index.php/Tropospheric_Delay
@ALSM
Last one. Sorry.
The mistake I made (as you just pointed out above) was not limiting the atmospherics to the actual “thickness” of the atmosphere ~500km. So I would expect single shot ranging errors on the order of 40meters 1 sigma.
Nothing to worry about.
@ALSM
Just not my day. I spent most of it doing CE stuff, and my brain is fried.
Above should read 160m 1 sigma (accounting for the four passes at ~500km each). Same conclusion relative to worry.
DennisW: You are still way off in the weeds, Dennis. You are confusing AIS VHF delays (basically horizontal ray paths through the atmosphere) with satellite link delays (high elevation angles). Read the papers I linked above. Water vapor delay is measured in mm, and ionospheric delay is measured in meters.
@ALSM, DennisW,
Since I was a professional radio astronomer for many years, I happen to know quite a bit about some aspects of atmospheric transmission of radio waves. There are actually two separate atmospheric effects that come into play in measuring round-trip propagation time through the atmosphere. One is the speed of the radio waves in the atmospheric medium, which will be slightly slower than in vacuum, depending on the refractive index. This is a very small effect in our problem.
The second effect is the bending of the ray caused by the vertical inhomogeneity of density/refractive index. In other words, light doesn’t travel in a straight line through the atmosphere (except through the zenith). In general, light travels in a curved path, and this curved path will always be longer in length than a straight line between transmitter and receiver, and thus the propagation time will always be longer than the geometrical pathlength. Most of the ray bending occurs low in the atmosphere near the Earth’s surface because there the density (and refractive index) are greatest. Thus, the refractive index being > 1 in the atmosphere causes the propagation time to increase due to both effects – the ray slows down and it must travel a curved path.
If the angle of the satellite is, say, 45 degrees above the horizon, the cumulative atmospheric refraction angle is only about 1 arc minute. At ½ degree elevation angle, the refraction is also about ½ degree. Thus, when you see the Sun or the Moon where the top of the image is ½ degree above the horizon, the object is actually at zero degrees above the horizon. In other words, you see the Sun/Moon fully risen just as geometrical sunrise/moonrise actually begins (without the atmosphere being present), because refraction allows you to see beyond the geometrical horizon.
Back to the problem at hand, at 45 degrees the extra pathlength due to atmospheric bending of the ray is also quite negligible – only about 1 mm. So GPS locators don’t need to account for atmospheric refraction to get meter accuracy.
NASA JPL has been doing radio ionospheric and atmospheric sounding on the other planets in our solar system using the data links and going behind the planet as seen from Earth. The signal strengths and polarization measured at Earth display variations as the spacecraft moves behind the planet because the signal reaching Earth passes through, and is refracted (and absorbed and depolarized) by the planetary atmosphere. Thus, you can measure the profile of refractive index (simultaneously at multiple radio frequencies) with altitude in the planetary atmosphere/ionosphere. The technique works quite well.
DrB: Thanks for explaining it, better than I did. In sum, there is some delay caused by water vapor in the troposphere plus some delay caused by charged particles (electrons) in the ionosphere. These delays are negligible for our purpose
DrB: Thanks for explaining it, better than I did. In sum, there is some delay caused by water vapor in the troposphere plus some delay caused by charged particles (electrons) in the ionosphere. These delays are negligible for our purpose. In addition, the atmosphere acts like a spherical lens, so the ray path is straight only looking straight up (90 degrees). As the elevation angle changes from 90 degrees (vertical) to zero degrees, the bending angle gradually increases. But it is insignificant until the angle gets close to the horizon, which is why the Sun is round except at Sun up and Sun down. Thus, at 40 degress (the angle at 0019), the delay caused by bending was also negligible. Pretty sure I mentioned all this years ago.
@David: Based on your continued study of the fuel system, do you have any reason to believe that a re-start of the left engine could not have been responsible for a bank, leading to the two BFO values at 00:19?
@ALSM
DrB: Thanks for explaining it, better than I did.
I thought you explained it just fine. What you and DrB have not done is come up with a number (including the multiplicative effects of the geometry, PDOP). Nor has anyone else. I call this an analytical gap. It should be closed. Saying it is negligible does not cut it for me. I will keep looking at it, although I strongly suspect it can be safely discarded.
The other issue I raised relative to BTO bias consistency should also be validated. If the SDU is repeatably powered on and off, is the BTO bias always the same? I don’t know, and it has never been explicitly addressed in any document I can find. Your POR data shows that for one power cycle is was the same. That is not much of a sample space.
@DennisW: Section 5.2 of the Bayesian report does provide some information about the variation of the BTO bias. For instance, the histogram of the BTO errors in Fig 5.2 for the previous 20 flights of 9M-MRO is referenced to the bias on March 7, so this would include the five power ups prior to MH370 that occurred on March 5 (1), March 6 (2), and March 7 (1). For the duration of 20 flights over 5+ days, the mean BTO error was 10 μs due to slow variation of the BTO bias.
DennisW: OK. Here’s a number you can bank on. I can do it in my head. The 7th arc error due to L and C band ray path bending, ionosphere delay and troposphere delay is <<100m. The "…multiplicative effects of the geometry, PDOP…" has nothing to do with range measurements (BTO values) via one satellite.
PDOP, to the extent relevant here, has always been incorporated in the projection of slant ranges to the surface position. IOW, the ±5.4 nm (99%) number already included the PDOP.
Tropospheric delay: ~2.5 meters at zenith, sea level (1 pass).
Ionospheric delay: Variable. Perhap 10 meters at Beijing, 2-3 meters for SIO, at zenith, single pass, L band. Much less for C band.
@ALSM
The 7th arc error due to L and C band ray path bending, ionosphere delay and troposphere delay is <<100m.
It is not <<100m. Check out the link below.
https://docs.google.com/document/d/15xVqy1qr1_aKLsI6onbk4K2ldDxldRjwLb5Cp2Kp9gM/edit?usp=sharing
@Victor
I regard "mean BTO" error as ambiguous since it could easily be computed using different values of BTO bias. I am not concerned about BTO error. as I tried to explain to ALSM. I am concerned about possible changes in BTO bias.
For Peete’s sake Dennis! The error is negligible compared to the digital logic noise. Besides, what ever it is…10m or 40 m or what ever…it is mostly soaked up in the BTO bias calibration. Geeeez!
@sk999
I think your numbers are definitely in the “ballpark”. Hard to know for sure since the ionosperic delay is subject to a lot of variability. At this moment I am satisfied that it is not an issue.
@DennisW: The DSTG calculated the error shown in the histogram by referencing a single value of bias. From the report: Figure 5.2 shows a histogram of the residual BTO measurement errors for R1200 messages referenced to the 7 March 2014 Tchannel value… The non-zero mean value of 10 μs is attributed to a slow variation in bias over the 20 flights. That doesn’t seem very ambiguous to me.
@ALSM
For Peete’s sake Dennis! The error is negligible compared to the digital logic noise. Besides, what ever it is…10m or 40 m or what ever…it is mostly soaked up in the BTO bias calibration. Geeeez!
I am not trying to piss you off. Just trying to completely tie up loose ends. I am now agreeing with you, but I needed more. Sorry. I am anal. You know that.
@Victor
That doesn’t seem very ambiguous to me.
Me neither. Sorry for being dense. My apologies to everyone, but at least I am happy now. Seriously, thanks for the feedback.
@all
Obviously I am paranoid about the location of the 7th arc. It is the only error I can think of that could defeat OI’s search for the aircraft.
I am a big fan of the terminus being close to the arc.
I am a big fan of Richard’s drift analytics.
I have other reasons to think 30S +/- a degree or so is likely.
I have not been this “content” since the aircraft dissappeared.
A ‘thank you’ to DrB, DennisW, sk999, and ALSM for clarifying the scope for consideration of anomalous effects for L-band and C-band transmissions.
That’s now cleared off the list of ‘itches’.
Grrr, read ‘anomalous ionosphere and troposphere refractive effects’
@All,
A new paper on my analysis of the 18:25 SDU log-on is HERE . This puts a ribbon on the 1-page summary I posted on 26 December.
An updated list of all my publications and links for downloads is HERE .
@DrB: Thanks, Bobby. You make a strong case for the SLOP near NILAM.
DrB: Great job on the latest 1825 transient paper. I think you have perfected the OCXO warmup transient understanding. Only took a year. Not bad considering the complexity of the issues to understand.
I agree with everything up to Section 3.5. Shifting the relative time to fit all the data to one transient curve is exactly what I expect to be needed for different initial conditions (Tamb and internal temp at PwrOn). The peak overshoot, as you say, should be nearly the same for all transients, although we may not always see all of the transient, depending on initial conditions..
That said, I think the specific SLOP maneuver you chose to explain the 1825 transient is only one of perhaps several (or many?) possible maneuver explanations. I agree there was a maneuver of some kind. It is evident from the Radar and BTO data alone. But for example, if the time for logon #7 is shifted ~12 seconds to the right, the first value then fits the curve, and so do all the values circa 140 seconds. Of course, the values circa 75 seconds will be “off” a little more, but that could be consistent with a different maneuver. So, why this specific SLOP maneuver?
Once again, regardless of what time offset is chosen for the #7 transient, the methods you have perfected to explain the transients are dead on. Great job. Holland should be smiling.
@DrB
Nice. It will take awhile to digest it, but I see no flaws on a first pass.
@Victor
Re: Your question ”@David: Based on your continued study of the fuel system, do you have any reason to believe that a re-start of the left engine could not have been responsible for a bank, leading to the two BFO values at 00:19?”
In the absence of a reply from David, I offer the following:
The APU DC fuel pump should have started and continued to run after the left engine flamed out. In addition, the left engine’s EEC should have activated both igniters after it detected the flame out. The left engine may have relit using the limited quantity of residual fuel that was available via the APU fuel inlet; however, any such relight is likely to have been very short-lived. I believe the run-time of several seconds witnessed by @ALSM in the simulator was probably close to the mark, despite the questionable simulator modelling of such a scenario.
Given the short run-time and the relatively slow acceleration typically experienced during an in-flight start at high altitude, the RPM is not likely to have increased very much before the engine again flamed out. The thrust increase during such a relight is likely to have been very small and of such a short duration that it had little or no effect on the aircraft’s flight path.
@Victor. Remember the objects seen from the air during the surface search of 28th/29th March. HMAS SUCCESS was a vessel searching (with a Chinese vessel) and identifying flotsam located by aircraft.
Subsequently AMSA reported that no objects related to MH370 were located despite some items of interest being spotted from the air, by an RNZAF P3 in particular.
Under FOI I sought details of any objects appraised by HMAS SUCCESS, present during that search, with positions. The attached confirms (paras 18, 20) that nothing of interest was found. Details of the objects were catalogued but to me there is little purpose in obtaining those unless they include positions. Other than that possibility I suspect positions would not be recorded in any document since the objects were of no consequence at the time to the job in hand.
However I have asked again whether position details are held anyway, in the catalogue or elsewhere. A ship’s track with times might help but that may be unavailable and/or insufficiently detailed. Anything useful that emerges I will pass on though I am not very confident there will be anything.
I doubt the blacked out sections to be relevant.
https://www.dropbox.com/s/0xzy1rq9ax6aee0/HMAS%20SUCCESS%20Report%20of%20Proceedings%20March%2C%202014..pdf?dl=0
My explanatory notes and acronym expansions:
‘…SURFOR’ – surface forces
‘Dogs watch’ – first dog watch is from 1600 – 1800, 4pm to 6. The second is 1800-2000 (no, no ‘hours’).
‘SME’,’PME’- starboard and port main engines
‘NFS’ – notice for steam (engine availability)
‘DCEX’ – damage control exercise
‘RHIB’ – rigid hull inflatable boat
Thanks, guys. Let me know if you see anything you don’t agree with.
@ALSM,
You said: “That said, I think the specific SLOP maneuver you chose to explain the 1825 transient is only one of perhaps several (or many?) possible maneuver explanations. I agree there was a maneuver of some kind. It is evident from the Radar and BTO data alone. But for example, if the time for logon #7 is shifted ~12 seconds to the right, the first value then fits the curve, and so do all the values circa 140 seconds. Of course, the values circa 75 seconds will be “off” a little more, but that could be consistent with a different maneuver. So, why this specific SLOP maneuver?”
Good question. The answer lies in simultaneously fitting the BFOs and BTOs while starting with the last radar track conditions of speed and course. As we have discussed several times, you don’t need a SLOP to fit the BTOs if you are willing to change the time of the last radar contact by 2 minutes to 18:24:12. All you need then is a turn. However, ATSB tells me they “have no reason to doubt” the Lido timings. Accepting that, then if you consider combinations of SLOP, FL change, and turns, you have to have a SLOP to fit both BTO and BFO from 18:25-18:28. The double turn is the only one that fits all the sat data (excepting possibly some wildly complex combinations).
Besides fitting the sat data, the procedure is in the book – the Malaysia Airlines emergency procedures – to do a 15 NM right SLOP. I was concerned at first because I couldn’t get as far out as 15 NM, more like 12, when I was using the 25 degree bank angle limit. Then Don dug up the B777 equation for bank angle limit versus (ground) speed, and Andrew confirmed it. That reduces the turn rate down to 0.6 deg/sec, and now you can get 15 NM with that turn rate while matching the sat data.
Here’s why the example you proposed doesn’t work. If you shift the 18:25 log-on to the right by about 15 seconds, the 18:25:34 BFO falls right on the peak of the transient curve. Thus, when you subtract the transient error, there is zero left for a maneuver then. That means the maneuver can’t be ongoing at 18:25:34. That pushes the start of the maneuver at least 93 seconds later in time (18:25:34 – 18:24:01). Now, with the maneuver starting that late, you will miss the 2-sigma error bar on the 18:25:34 BTO altogether, and your BTO fit will be poor. In order to pass through the error bars on both the 18:25:27 and 18:25:34 BTOs (and in between the two BTO values), the first turn must begin near 18:24:00. Look at the BTO plot in my Figure 6 and you’ll see what I mean. You have to start turning away from the sub-satellite point well before 18:25:34. That means the maneuver is ongoing then, and that means that the transient cannot be the only contributor then. There must be a BFO contribution from an ongoing maneuver at 18:25:34, just to match the BTO then.
I’ll have some new 18:40 BFO scenarios ready to look at in a couple of days. I’ve found some solutions there that I had not seen before.
@Victor. You asked, “ … do you have any reason to believe that a re-start of the left engine could not have been responsible for a bank, leading to the two BFO values at 00:19?”
I have been delayed and I see @Andrew has answered succinctly. My long winded version has a similar outcome though it has led me to a further conclusion.(Andrew please feel free to comment):
• The ATSB has indicated that either engine might relight from their tank’s residual fuel as the aircraft attitude changes. In the right’s case that would be through the tank suction by-pass point becoming immersed in residual fuel, which could occur in a pitch change. In most circumstances, with AC being generated from the left engine still and autopilot selected, the A/P would reduce the prospects and duration of that and correct any effect. Also, the chances of a relight are lowered at high altitude.
• As to the left, the APU DC fuel pump would access left tank residual fuel and supply up to 3150 lb/hr to the engine at flame-out. At that point AC supply would be keeping the autopilot engaged so the effect of any relight should be corrected and the residual fuel would be consumed before AC was lost. After consuming that, or if there were no relight, AC would be lost as the engine dropped beneath idle. A 50 lbs (say) fuel residue would take a minute to consume if relight supply were at 3150 lb/hr, shorter if that were supplemented by suction fuel, longer if starting fuel consumption, possibly altitude dependent, were less than the 3150. Such a relight would extend the time from MEFE to APU auto-start.
• With no residual fuel left the engine (being RR) would shut down when N3 dropped to 35%, having led to no upset. Were there no relight the APU DC pump supply would continue from flame out to shut down. Below idle that would be split between the engine and the APU for auto-start (on AC loss), the APU’s consumption reducing the engine supply down by about 130 lb/hr. I think the thrust the engine would develop would be small if it ignited beneath idle, again less likely at altitude. Engine acceleration to appreciable thrust from that point is notoriously slow. In any case maximum thrust realised would be inhibited by the meagre 3020 lb/hr supply unless supplemented by suction fuel, which would be by chance and probably momentarily.
• Windmilling would reduce the engine deceleration rate and there is the question as to how long it would take to drop from idle to 35%, particularly if in a dive. However since the autopilot would be engaged until the engine dropped to idle, one would not expect the fuel remaining after that to last longer than the APU’s subsequent 1 minute to start, relight or not, so it is unlikely to have any effect during the log-on descent LOA at more than a minute after that. If the engine had in fact reached 35% N3 before residual fuel exhaustion and shut down with fuel remaining (and being accessed by the APU) obviously it could not relight in the subsequent dive, fuel or not.
• All the above is pre- (and post) amble to your question about it rolling the aircraft before the log-on, leading to those high and accelerating descent rates. As before, a relight above idle would have no effect, autopilot engaged. If it did not light above idle I do not see why it would have any better chance of doing so beneath it, altitude being substantially the same. However, if by chance it did relight then the thrust at that engine speed is unlikely to have much effect on roll, as above; and please note it would be in the opposite direction to the destabilising flaperon and RAT pitch and yaw moments, so would reduce them if anything.
• In summary then I believe the chance of a left engine relight having been responsible for a bank which led to the final BFOs is negligible.
• A corollary conclusion can be drawn that a relight would not affect search width. Not having thought my way through all that before, thank you for the question.
• BTW, one other effect on aircraft dynamics which has gone unmentioned is engine gyroscopic forces, since they have the same direction of rotation. This would act to translate pitch into yaw and VV but I do not know by how much.
@David
Whilst we agree that any relight is likely to have had a negligible effect on the flight path, you seem to believe the APU DC fuel pump starts some time before the APU auto start sequence commences. As I understand it, the APU DC fuel pump starts when no pressure is sensed in the left engine feed manifold AND the left engine N3 decreases below idle. The APU auto start sequence commences when power is lost to both AC transfer buses. That will also occur when the left engine N3 decreases below idle, assuming the left engine is the last engine to fail. The APU auto start sequence should therefore commence at pretty much the same time as the APU DC fuel pump starts.
@Andrew: To be clear, whether an engine restart caused MH370 to roll, I don’t know. But in some of Mike’s simulations, an unsuccessful restart caused a very steep bank.
@Andrew. APU DC fuel pump start on engine fuel supply failure.
The AMM I refer to says the APU DC fuel pump will start on loss of fuel feed manifold pressure and left engine N2 less than the engine run speed, which is for the GE engine. To me that conveys zero duel feed pressure and detection of engine failure. With certification requirements I would have expected the RR to follow the same principle: no fuel and the engine running down as I understand it.
To wait until the engine dropped beneath idle sounds daft, given recovery times from below idle.
Along the same lines you said,”The APU DC fuel pump should have started and continued to run after the left engine flamed out. In addition, the left engine’s EEC should have activated both igniters after it detected the flame out.” There is little benefit from the igniters being active if there is no fuel for it to ignite.
@Victor. The ALSM simulations were with a United Airlines simulator I believe (presumably GE powered?) with it unclear from whence it got its fuel given the reported Boeing engineering simulator experience of no such relights. Presumably that was with the RR. In their simulations there seems to have been no residual fuel, APU auto-starts not being part of simulations.
Also the engine might not be shut down under like criteria to the RR, though again they both will have met certification requirements.
@David,
The UAL B777 simulators are capable of being RR powered, my former colleagues built them and, more recently, commissioned the TCAS updates in 2014.
@David: Yes, we’ve previously acknowledged that the simulators are different. I have been discussing the United simulations that Mike witnessed. The Boeing simulator does not allow an engine re-start after the dual engine flameout from fuel exhaustion. Obviously, the fuel model is different, and possibly more accurate, for the United simulator. I can see what looks like EPR, N1, and EGT displays in the EICAS, which would mean RR engines were simulated, not GE.
Here is the time sequence in the simulation where both engines flamed out at the same time and there was no rudder trim.
Notice that the engine re-start would not have to last long for a very substantial effect on bank angle, and hence the flight dynamics.
00:50 Left and right engines coast down
1:31 N1 for the left engine beginning to slowly increase, bank at 10° to right. Bank increasing.
1:37 APU power up
1:50 Bank at 18° to right
2:09 Bank at 30° to right
2:16 Bank at 45° to right
2:19 Bank at 50° to right, left engine begins coastdown
2:22 Bank at 55° to right
2:31 Bank at 45° to right
2:38 Unexplained rapid automatic correction to level wings.
Without this correction (which I suspect is an artefact in the simulation), the aircraft would have continued in an accelerating steep descent.
@David
1. The AMM for the RR-powered aircraft stipulates a similar condition for the start of the APU DC fuel pump, ie “left engine N3 is less than the engine run speed”. I’m told that ‘run speed’ in this case = idle, for what it’s worth.
2. The operation of the APU DC fuel pump is controlled separately to the auto-relight function that activates the igniters in the event of a flameout. The fuel pump is controlled by ELMS, while the igniters are controlled by the EEC. If the engine does not recover following a flameout and continues to run down below 35% N3, the EEC shuts off fuel (via the HPSOV) and ignition and disables the auto–relight function. However, the APU DC fuel pump continues to run to supply fuel (if any) to the APU.
3. United’s B777-200s are powered by PW engines. PW uses EPR for thrust setting purposes, similar to RR.
@Andrew, @David: True, the United simulator is for a B777-200. I incorrectly assumed it was for the B777-200ER.
Nonetheless, looking at the three simulations Mike completed, all show a failed engine restart of the left engine. For the case in which the left engine failed first, the failed restart of the left engine occurred while the right engine (and TAC) was still running, so there was no effect on the flight attitude. For the other cases, the left engine failed restart caused an increasing bank to the right. The effect was most pronounced for the case in which both engines flamed out at the same time. Why do we believe that the left engine restart was incorrectly modeled?
The simulator I used at UAL Training Center was a Thomson simulator, not UAL specific simulator. It can be configured to simulate different configurations. Pilots from different airlines all over the world go there to train.
Below is a description from my notes of the UAL simulator we used (one of many) and the configuration. Note that the Simulator is a Thomson Model: 700-200 (not a Boeing model number). It is configurable to various -200 versions and engines to serve many airlines. As far as I can remember now, we simulated the same model and conf as 9M-MRO.
Manufacturer: Thomson
Model: 700-200
Year of manufacture: June 1997
Level: D
Configuration
AutoThrottle: Under control of FMS in ECON Mode.
CI=50
ZFW (Zero Fuel Weight): 395,000 lbs
Fuel at 1707 (ACARS reported): 96,500 lbs
Fuel at end of flight: 200-800 lbs per tank, depending on scenario
CG: 28%
Altitude: 35,000 (for most of the simulations)
TAS @ IGARI: 471 kts
Wind Velocity at 38S: 57 kts
Met Direction at 38S: 240 degrees (direction from which the wind is blowing)
Heading at 38S: 186 degrees
Track at 38S: 179 degrees
@ALSM: Thank you for the details. So it’s safe to assume the simulator was configured as a B777-200ER. My question remains. Does anybody have reason to believe the left engine restart was incorrectly modeled? If correctly modeled, then the failed restart, leading to a banked descent, could explain the value and timing of the two BFO values at 00:19.
@DrB
I had no problem and appreciate the nice shape of the SLOP turn in your paper. However I thought it was controversial to say the SLOP turn and SDU reboot suggests the pilot was dealing with a mechanical issue, and not intentional diversion of some kind. The good thing is as least you said your opinion, and that could set off some discussions, which appear to be starting over on JW.
When you do the next time slot 18:40 work, I want to know how close UAE343 and MH370 come together. Perhaps ATSB can tell us the separation and flight path.
@TBill: I didn’t make this comment because it has been said many times before, and it seems to generate a lot of unresolved debate. However, in my humble opinion, a pilot dealing with an emergency would stay close to a suitable airport, and descend and land. The pilot of MH370 did not descend below cruise altitude as the plane approached Penang, and then the plane flew away from it. That makes no sense. The power up of the left bus around 18:22 could have been for many reasons, but I don’t see how you can relate it to a pilot dealing with a mechanical failure in light of previous actions.
@Victor
I agree. The only scenario, IMO, that fits all the details is an intentional diversion. The simulator data alone makes all mechanical scenarios questionable in addition to your disconnects above.
@DennisW: At this point, anybody that has followed this closely already has an opinion about the cause. I don’t see many people changing their opinion.
Luckily, we might not need to reach consensus on a cause to find the plane.
@ALSM
RE: ”It is configurable to various -200 versions and engines to serve many airlines. As far as I can remember now, we simulated the same model and conf as 9M-MRO..”
Fair enough.
@Victor
@ALSM
RE: ”Why do we believe that the left engine restart was incorrectly modeled?”
The modelling of the aircraft systems’ behaviour might not be entirely accurate once the FQIS shows zero fuel remaining in the tanks. The ATSB estimated that about 30 lb of fuel was available to the APU after left engine fuel exhaustion. It is unclear if that quantity includes the fuel in the APU fuel line in addition to that available in the tank. However, based on the calculated maximum run time of 13 minutes 45 seconds, I assume the 30 lb figure includes both the fuel in the line and that in the tank.
David calculated that the APU fuel line holds about 17 lb, leaving about 13 lb of fuel available from the tank. That fuel would also be available to the engine via the APU fuel inlet and would be expended very quickly during any relight attempt. I would question a modelled relight that lasted more than about 10-15 seconds, given that fuel would start to flow several seconds before ignition occurred. As explained in my earlier post, I don’t believe the engine would accelerate to a high RPM during such a short period and consequently would not produce much thrust.
With reference to your earlier post @Victor. Notwithstanding the provisos about accuracy of the simulator, I am intrigued by the following [last line]:-
“2:22 Bank at 55° to right
2:31 Bank at 45° to right
2:38 Unexplained rapid automatic correction to level wings.”
Does anyone have an informed view on why the simulator might have done this? And whether there is a real-life possibility that this might occur? The reason that I ask is that it provides a possible scenario for recovery from a steep bank and continued flight beyond ~25NM of 7th arc.
@Andrew: In the simulation, before the re-start, the engine is windmilling at essentially idle thrust at N1 around 25%. As EGT and speed begin to increase, the net thrust (thrust minus drag) difference between the engines produces a small yaw moment. It looks like N1 eventually gets to about 60% (rough guess) for about 8 seconds and then coasts down. It’s during the time that the speed is held at 60% that you see the engine banking the plane. Whether or not there is enough fuel for all that, I don’t know, but considering that Thomson modeled this extra fuel, it seems possible that they would get it correct.
@Paul Smithson: That’s a good question. Envelope protection, which is only available in “normal” flight control mode, would reduce the bank angle to 30° after it exceeds 35°, and would not level the wings. After dual engine flameouts, the flight control mode would be “secondary”, which should offer no envelope protection. I have assumed the correction to level wings is is an anomaly in the simulation, but perhaps that is an incorrect assumption.
@Victor
RE: ”…considering that Thomson modeled this extra fuel, it seems possible that they would get it correct.”
Possible, yes; however, it’s also possible they modeled a small amount of residual fuel, sufficient to allow a brief relight, but not necessarily representative of that found in the real aircraft. I would urge caution in accepting that the relight seen in the simulator reflects exactly what would happen in real life. Simulators are notorious for odd behaviours and, in my experience, it is not uncommon for simulator instructors to tell their trainees to ignore something the simulator did or didn’t do. I have had to do that myself on.a number of occasions.
@Andrew: Caution acknowledged regarding the precise behavior modeled in a simulator. That said, what else besides an engine restart could cause a 0.65g downward acceleration 2 minutes after fuel exhaustion with the APU running?
With reference to the recent Christmas Island paper linked by Victor.
The authors claim air traffic control expertise, so should know a thing or two about standard procedures & manoeuvres. Shortly before the first ping ring(s) 1825-1828 they suggest a turn to the right, then a left turn in order to cross airway N571 at close to perpendicular. They also suggest descent to below minimum altitude for aircraft using that airway.
This sequence of manoeuvres has potential to fit the [Dr.B-calibrated] BFO and BTO at 1825-1828. Indeed it could do so even better without a descent. In other words, instead of a SLOP you have a right turn, then transition into a left turn that ends up on a course of something like 200 degrees. Of course you would have to be “lucky” to get somewhere close to constant BFO that just happens to match a heading of ~290 over the 9 seconds between 182806 and 182815 as the aircraft continues turning towards “perpendicular to N571”.
My question:
Does this make sense (from a procedural point of view)? If your intention is to turn left, why not turn left away from the airway? Would you really turn right, descend, then turn left in order to cross at 90 degrees the airway you have just left?
@Victor
Re: ”…what else besides an engine restart could cause a 0.65g downward acceleration 2 minutes after fuel exhaustion with the APU running?”
As you previously commented:
“The answer is either a banked descent, caused by lateral asymmetry with or without pilot input, or a nose-down pilot input.”
“The cause for the lateral asymmetry could have been many things, include the unpowered left flaperon under RAT power, an engine restart, rudder out-of-trim, reconfiguration of hydraulics after APU start up, etc.”
Of those, I think the unpowered left flaperon while the flight controls were powered only by the RAT, with or without a mis-trimmed rudder, and/or pilot input are the most likely culprits.
hi all,
if this OI distributed data processing octopus doesnt find it, the its not there; but I believe already that it simply must be; although still inside a game somewhat, as SC is quite similar to star-trek uss-enterprise… peace;
@Andrew: It’s the timing relative to the last engine out along with the descent rates that is hard to explain. By the time the BFOs registered, the APU was already running for a minute. The asymmetry introduced by the RAT would have already occurred and been corrected by the APU well before that. For the rudder out-of-trim, again it is hard to find a scenario that produces a 4,500 fpm descent after a minute, and then progresses to 15,000 fpm 8 seconds later. Pilot input of course would explain it.
@falken: Welcome. Seabed Constructor can certainly cover a lot of seabed every day. However, it still has to look in the right place, and the debris has to be distinguishable. I am hopeful, but not certain, it will be found.
@Victor
@Andrew
Victor, re “However, in my humble opinion, a pilot dealing with an emergency would stay close to a suitable airport, and descend and land. The pilot of MH370 did not descend below cruise altitude as the plane approached Penang, and then the plane flew away from it.”
Your scenario assumes a conscious, capable and willing pilot at the controls between say 1730 – 1815 UTC. What if that wasn’t the case? Two possible circumstances come to mind;
– that the flight crew were incapacitated sometime around 1730 UTC and revived (or replaced) around 1815 UTC, or
– that deliberate malicious action was disabled around 1815 UTC by say the locked out flight crew member regaining control of the flight deck.
Re “The asymmetry introduced by the RAT would have already occurred and been corrected by the APU well before that.”
I don’t understand how the APU coming on line would “correct” the roll introduced by floating left flaperon. Restoration of power to the Left and Right hydraulics when the APU started would certainly cause the left flaperon to retract to the stowed position but it wouldn’t bring the airplane back to wings level. Depending on how long the airplane had been running on only the RAT enough roll may have been introduced for it to continue to increase even after the left flaperon had been retracted.
@Mick Gilbert
What about the coordinates on Shah’s simulator?
What about lack of communication of any kind?
Scenarios like yours are simply not compatible with all the information we have.
@Victor
@Mick Gilbert
@ALSM previously commented:
”In all our simulations, the airplane started to bank shortly after FE. I have no doubt that is what happens in real 777s. The bank angle increases with time in all the sim’s. There is no doubt that also happens in real 777s. You can argue that the precise details may start to diverge from real life after reaching 60 degrees or so, but the fact that all the sim’s indicate this pattern of increasing bank angles must be accepted as typical behavior.”
In my experience in the simulator, the aircraft has a tendency to roll left with both engines failed and the RAT supplying hydraulic and electrical power. The accepted ‘wisdom’ is that the roll moment is caused by the left flaperon floating up when it is unpowered. That roll moment would be exacerbated if the rudder was mis-trimmed in the same direction. If there are no control inputs following the second engine failure, the aircraft will develop a significant bank angle during the one minute it takes for the APU to start generating electrical power and with it the restoration of pressure in the L & R hydraulic systems. That bank angle will not be ‘corrected’ when hydraulic pressure is restored, but the roll moment caused by the flaperon will cease when it returns to the neutral position. If a significant bank angle had already developed by that time, then I think it’s possible the aircraft would enter a spiral with an increasing rate of descent.
If I remember correctly, @Gysbreght commented somewhere that a vertical acceleration of 0.65g would require a bank angle greater than 60°. That said, I accept that none of the simulator scenarios demonstrated an increasing rate of descent of the order suggested by the BFOs. That being the case, pilot input might be the most likely cause of the increasing rate of descent.
How accurate are the BTO calculations? These determine the location of the 7th arc. I compared my calculations with three other sources: JON Table 9, Godfrey version v16.1, and Ulich Jan 9, 2017 true heading.
Short answer: agreement is good to excellent.
JON and Godfrey construct routes that are designed to land precisely on the arcs at a given time. Thus, the BTO is the value given in the signal logs and the position is as provided. (JON only uses this method post-IGARI, so I will omit the pre-IGARI points as they are not useful.) Ulich supplies the calculated BTO for all positions, but only starts at 18:25. I use the supplied positions, times, and altitudes to calculate BTOs.
JON only provides positions accurate to 0.1 degrees, so some quantization error is present (+/- 19 microsec). From 18:25:27 to 00:19:29, my calculate BTOs are in good agreement with the JON, but there is a systematic slope in the residuals, from -23 microsec at 18:25 to +41 microsec at 00:19. If real, this would represent an uncertainty of about 4 nm at the 7th arc. Note that my bias offset calibration, tied to Kuala Lumpur, otherwise works perfectly, so the error may be somewhat less.
The comparison with Ulich is much better. The rms error between 18:25 and 0:19 is only 13 microsec; if I adjust the bias offset by +7 microsec, the agreement becomes excellent, with no systematic slope over that interval. The rms error in the location of the arc is less than 1 nm.
The comparison with Godfrey is even better. The time interval now runs from gate 16:11 to 00:29 and directly ties the location of the 7th arc to gate C1 of the satellite terminal at KL International. After removing a systematic offset in the bias of 15 microseconds, the peak difference is 6 microsec (for the KL point), and the rms is 1 microsec. An error of 6 microsec corresponds to an error in the location ot the 7th arc relative to gate C1 of 0.4 nm.
All,
Just a thought on the relation between bank angle and radius of curvature in a downward accelerating situation.
The bank angle equation, tan(bank) = velocity * angular rate / gravity
expresses the conversion of lift (needed to balance gravity) into a horizontal centripetal force (perpendicular to the direction of travel of the aircraft), which is balanced by centrifugal force as the plane turns.
In the presence of downward acceleration, the effective gravity is less, and thus one needs a greater bank angle to achieve a particular radius of curvature. This might be part of the explanation as to why the paths shown in Fig. 6 of the Nov 2, 2016 ATSB update (showing flight paths at the time of the final BTO) are not as curved at the 2 minute mark after flameout as one might expect (as pointed out by Gysbreght recently.) I can also imagine that other effects dominate.
Random question, which I suspect no one can answer, but here goes.
The BFO error consists of two components – one is the improperly calculated Doppler compensation (maximum when the satellite is furthest above or below the equator), and the other is the Doppler from the satellite motion itself (maximum with the satellite is at the equator). Suppose an inclination, node, and latitude were downloaded to the Honeywell/Thales AES, will the AES correct for the first effect, the second, or both?
@Mick Gilbert: I don’t understand how the APU coming on line would “correct” the roll introduced by floating left flaperon. Restoration of power to the Left and Right hydraulics when the APU started would certainly cause the left flaperon to retract to the stowed position but it wouldn’t bring the airplane back to wings level. Depending on how long the airplane had been running on only the RAT enough roll may have been introduced for it to continue to increase even after the left flaperon had been retracted.
I never said the APU coming on line would correct the roll. I said the APU coming on line would correct “the asymmetry introduced by the RAT”. The asymmetry introduces a roll RATE. I didn’t think I needed to explain that. Remove the lateral asymmetry of the flaperons, and the roll rate is reduced, not the bank angle.
The roll RATE introduced by the hydraulics on RAT hydraulic pressure would start once the RAT came on line and would end once the APU came on line.
We believe that at the time the BFOs were recorded, the APU was running, and therefore the flaperons were symmetric and the induced roll rate small. Yet the descent rate increased from 4,500 fpm to 15,000 fpm in 8 seconds. At the time the descent rate was 4,500 fpm, the bank angle was still modest. Yet, in eight seconds, both the bank angle and descent rates increased to large values. How was that possible on APU power? Something is missing. I proposed the engine re-start as the mechanism for this to occur. People here think there was not enough fuel for this scenario. A nose-down pilot input is another explanation, perhaps in combination with rotation of the yoke. That would be quite effective.
It almost doesn’t matter. If the plane was accelerating downward at a rate of 0.65g, it impacted the ocean soon after, and should be found close to the 7th arc.
By the way, flaperons don’t “retract”. When the flaps are not extended, the flaperons pivot about a neutral position. When the flaps are extended, the flaperons “droop”.
Your scenario assumes a conscious, capable and willing pilot at the controls between say 1730 – 1815 UTC.
Yes, based on the facts at hand, including the timing of the disabling of the transponder and ACARS, the lack of communications, and the incriminating evidence found on the captain’s computer, I believe the evidence overwhelmingly suggests the captain diverted the plane. You feel differently. You’ve proposed many other scenarios. I have not been persuaded. I have not persuaded you. Let’s not rehash it. It doesn’t change the recommended search zone.
@Andrew said: That bank angle will not be ‘corrected’ when hydraulic pressure is restored, but the roll moment caused by the flaperon will cease when it returns to the neutral position. If a significant bank angle had already developed by that time, then I think it’s possible the aircraft would enter a spiral with an increasing rate of descent.
A descent rate of 4,500 fpm is not consistent with a significant bank. More like around 20°. (I can provide a more exact value.) That wouldn’t cause a spiral.
@sk999
Good question. How do you quantify your ring data? Subsatellite point and great circle radius?
@sk999: If the orbit is circular but inclined, and the value of the inclination, ascending node, and longitude used by the AES were exact (which they can’t be due to the limited resolution), then the AES compensation will be perfect. That was the basis of the BFO spoofing scenario I developed some time ago. If the inclination in the AES is set to twice the actual inclination, a northern path looks like a southern path. If the inclination used by the AES is exactly the true inclination, it is impossible to use the BFO to distinguish a path. Recently, I asked a Thales engineer about this, and he agreed with my assessment.
Victor,
Your spoofing scenario only works if the AES compensates for both of the effects that I mentioned. That is why I asked the question the way I did. If the satellite is exactly over the equator, only the motion of the satellite affects the BFO, which means that only the location of the aircraft relative to the satellite matters, not the track or velocity of the aircraft. If the Thales engineer is correct, then yes, the AES must be computing the Doppler due to the motion of the satellite.
It was just a thought.
DennisW,
“How do you quantify your ring data? Subsatellite point and great circle radius?”
I put out 3 posts in a row (whew!) so I presume you are asking about the first one.
I calculating everything as a timing offset (BTO) since that is the form that the data are supplied. A BTO actually defines a sphere, not just a ring. Important when you want to include the effects of altitude of the aircraft at the 7th arc. [As an aside, the JON paper set the altitude to 0 for post-IGARI points.] Further, the Earth is spheroidal, so those “rings” aren’t really spherical.
As a further aside, I still have flags to turn the spheroidal components of the model off (back to a spherical earth, which is how we all worked early on, including Inmarsat) and the impact on the BTO calculation is huge. Conclusion: everyone is using a spheroidal Earth model.
@sk999. I understand your question. The Thales engineer is quite knowledgeable. Also, it would be odd to model the position of the satellite with an inclined orbit, but not the velocity.
@all,
Ian Holland has revised his paper. New version is at :
https://arxiv.org/abs/1702.02432
@VictorI
On your remarks in the 9.05PM answer to @Mick Gilbert and @Andrew:
” A nose-down pilot input is another explanation, perhaps in combination with rotation of the yoke. That would be quite effective.
It almost doesn’t matter. If the plane was accelerating downward at a rate of 0.65g, it impacted the ocean soon after, and should be found close to the 7th arc.”
“A descent rate of 4,500 fpm is not consistent with a significant bank. More like around 20°. (I can provide a more exact value.) That wouldn’t cause a spiral.”
I think this issue (pointed out yesterday by @Gysbreght on JW-blog) could turn out to matter a lot if the plane not gets found close to the 7th arc.
If the high descent rates were induced by pilot-input then there is also the possibility of recovery/pull-out by pilot-input.
China Air 006 recovered/pulled-out using only 1500ft altitude coming out a near vertical high speed dive without running engines (engines restarted after the pull-out).
@Victor
“I never said the APU coming on line would correct the roll. I said the APU coming on line would correct “the asymmetry introduced by the RAT”.”
Ah, there you go. On the basis that the discussion had concerned itself with the attitude of the airplane, I had read “asymmetry” as relating to an asymmetry in orientation, in this case around the longitudinal axis (ie bank) rather than an asymmetry in flight controls.
“The roll RATE introduced by the hydraulics on RAT hydraulic pressure would start once the RAT came on line and would end once the APU came on line.”
I understand what you’re getting at but the circumstance whereby both PCUs on the left flaperon change to bypass mode actually isn’t related to when the RAT comes on line; it’s related to when the pressure in the left and right hydraulic systems falls sufficiently consequent to the second engine flaming out. The deployment and proper operation of the RAT isn’t a factor with regards to the left flaperon floating.
“By the way, flaperons don’t “retract”. When the flaps are not extended, the flaperons pivot about a neutral position. When the flaps are extended, the flaperons “droop”.”
Thank you, I understand how the flaperons work. They “droop”, that is move down and aft, to provide lift augmentation but when providing roll control, which is what we’re talking about here, they deflect up and down out of the faired position. In the circumstance we’re addressing, with both PCUs on the left flaperon in bypass mode, aerodynamic lift causes the flaperon float up. When the PCUs are repowered by the APU coming on line the flaperon is pulled back down to the faired position. I didn’t think that “retract” would be a confusing or incorrect verb to describe that process.
With regards to the rest of it, agreed; scenario bashing isn’t going to refine the search area.
On the matter of search area, to rephrase your questions on likelihoods, if OI search the 25,000 km² zone and then proceed to search 25 nm either side of the 7th arc all the way up to say 25°S, what do you think the likelihood is that they will find the underwater debris field?
@Mick Gilbert said: I understand what you’re getting at but the circumstance whereby both PCUs on the left flaperon change to bypass mode actually isn’t related to when the RAT comes on line; it’s related to when the pressure in the left and right hydraulic systems falls sufficiently consequent to the second engine flaming out. The deployment and proper operation of the RAT isn’t a factor with regards to the left flaperon floating.
The situation we are discussing occurs when there is asymmetry in the position of the two flaperons, i.e., the left flaperon is bypassed and floats up and the right one held in the neutral position. That is induced by the RAT. If the RAT is not supplying pressure to the center hydraulic flight controls, there is no asymmetry, and no induced roll rate.
You said: Depending on how long the airplane had been running on only the RAT enough roll may have been introduced for it to continue to increase even after the left flaperon had been retracted [sic].
Rather than speak in generalities, it would be more instructive if you provided the specifics of your scenario, starting with flameout of the second engine. Please describe, as a function of time, left and right flaperon position, bank angle, descent rate, airspeed, and whatever else you think might be necessary to show that your scenario produces descent rates of 4,500 fpm and 15,000 fpm separated by 8 seconds.
@Victor
Yes, I see what you mean regarding the RAT being required to produce the asymmetry.
With regards to the latter, sure, I’ll get right on that next time I have access to a sim.
@Mick
With regards to the rest of it, agreed; scenario bashing isn’t going to refine the search area.
I disagree with both you and Victor on that point. The scenario can have a significant impact on the search area.
Pilotless vs piloted has a huge influence on terminal locations. Victor’s comment reminds me of the “Duncan days” when talking about scenarios or motives was strcitly forbidden and Occam’s Razor ruled the day. I thought we were past that view.
I think at this very moment virtually nothing will impact OI’s search area. OI has likely architected a search plan, and will stick to it. That is a much different notion than a statement regarding refinement of the search area.
Mick: Re “…that is move down and aft,…”. No, they don’t. That is what Victor was trying to explain. They move up and down, like an aileron, not “aft” like a flap.
@ALSM
Mike, flaperon is a combination of flap and aileron; it works like a flap for lift augmentation and like an aileron for roll control.
The B777 FCOM describes them as follows under Flight Controls:
“The flaperons are located between the inboard and outboard flaps on both wings.
In the normal mode, they are used for roll control with the flaps either retracted or extended. For increased lift, the flaperons move down and aft in proportion to trailing edge flap extension.” My bolding.
Mick:
The flaperons “droop” when the flaps are extended. They are hinged (see below), so they cannot extend aft like the flaps do. I think the manual you referenced is confusing. The hinges are on the bottom of the leading edge of the flaperon. The actuators attach to the upper leading edge and extend fore-aft to move the flaperon up-down, leaving a gap. (see “Functional Description” below) IOW, the actuator linkage moves aft, not the flaperon.
From a Boeing Training Manual
Physical Description
The flaperons are light weight composite materials structures. The skin is carbon fiber reinforced plastic bonded to a core of Nomex honeycomb. The overall dimensions are approximately 62 in by 95 in (1.6 m x 2.4 m). Each flaperon weighs about 110 lb (50 kg). The flaperon attaches to the airplane structure with 2 hinges. An aluminum strip diverts lightning strikes on
the flaperon trailing edge. A cove lip door at the leading edge of the flaperon
supplies a gap when the flaperon is full down to make aerodynamic performance better. The cove lip door attaches to the airplane structure and to the flaperon with two main hinge fittings.
The flaperons move a maximum of 11 degrees up and 34 degrees down. Two PCUs operate each aileron and flaperon.
The flaperons are standard inboard ailerons which also operate as flaps. They are between the inboard and the outboard flaps on the wing trailing edge.
During extension of the flaps, the ailerons and flaperons droop to increase the lift of the wing. The ailerons and flaperons of both wings move down. When drooped, the ailerons and flaperons are fully operational for roll control.
Functional Description
When the flaperon moves down, a mechanical linkage pulls the drive link attachment point of the fitting aft. This rotates the cove lip door up, which supplies a gap along the leading edge of the flaperon.
Drawing of flaperon here: https://goo.gl/YHdsCh
@ALSM
Thank you for that explanation and the diagram, Mike.
I took the phrase “move down and aft” straight out of the FCOM and it was broadly consistent with my (albeit limited) understanding of how the flaperons work. Now, I’m no book-schooled, high-price engineering-type but if you have a look at this video from the 3:40 mark you’ll see that the flaperon definitely moves up and forward (the rub line at the crest of the upper camber is a bit of a give away) as the flaps are retracted. That, by corollary, means that they “move down and aft in proportion to trailing edge flap extension“. It’s not a lot of movement to be sure, especially when compared to the flaps, but it certainly appears to be consistent with the description in the FCOM.
@ALSM, Victor, Mick Gilbert. Please let me join this discussion even if of obscure import. The flaperon moves down and aft. If it did not move aft the cove lip door would not clear it and likewise there would be no horizontal gap between the flaperon and the top of the wing.
Take a vertical line through the hinge, an aircraft datum. Droop the flaperon and that line moves forward on the flaperon ie the flaperon has moved aft.
The outer flap operates in a similar way, rotating down and aft, its pivot link pivoting about trh bottom of its pivot link, though the flap also rotates slightly less than the pivot link, increasing its aft movement slightly.
https://www.dropbox.com/s/uc7ptmahph1y2hv/Outer%20flap%20movement.pdf?dl=0
That video showes that the inboard flaps extend and retract. The flaperon does not.
@ALSM
@David
I’m not invested in this one at all but to my untrained eye, using the gap between the leading edge of the flaperon and the wing and rub line on the flaperon as a guide, it looks like the flaperon moves forward.
I’m happy to chalk it up as an optical illusion and move on.
@ALSM
@Mick Gilbert
Gents,
The flaperons do move slightly aft as they droop, although they obviously don’t extend anywhere near as far as the flaps. The flaperon hinge is located beneath the leading edge of the flaperon, while the PCU is attached to the leading edge. When the PCU extends, the flaperon moves slightly aft as it rotates about the hinge.
Other videos that demonstrate the flaperon movement:
http://www.youtube.com/watch?v=YxOskxXiYMk&t=1m32s
http://www.youtube.com/watch?v=qvQSllUyti0&t=0m36s
@Dr.B re “Ian Holland has revised his paper”. Have you identified the revisions? Anything substantive to be aware of?
@Victor. Another influence on roll before and during the final log-on is the inherent static roll stability. The aircraft has dihedral. If that prevails under APU driven hydraulics, where controls should be neutral, it will tend to right itself and of course the converse. Some evidence for the former:
From your post at 8:49 AM, 8th January, from ALSM’s simulation:
2:19 Bank at 50° to right, left engine begins coastdown
2:22 Bank at 55° to right
2:31 Bank at 45° to right
That suggest a static stability roll reduction, 10 deg in 9 secs or 66 deg/min.
I took a look at your 6th June, 2016 simulation but that looks inapplicable.
However while the timing of the log-on descents is difficult to explain and the above makes that even more difficult, I note the ATSB, Boeing and implicitly the SSWG appear to be in agreement and sanguine about it. Though plenty of attention has been paid to the most recent simulations, including by me, they were just supplements to those earlier, covering some particular configurations not looked into.
We have no details of the earlier but I think the 2nd November 2016 ATSB report page 8 dot points cover the aggregate, not just those illustrated above them.
I do think on reflection that it is unlikely the descent rates observed were inconsistent with those of the BFOs; and timing likewise.
@David: Yes, there is some dihedral effect. I do think that is the reason that the 55° bank reduced to 45°. In the Level D simulations, you see evidence of it for high speed and high bank angles. It is not a constant roll rate. Rather, the restoring moment is a function of bank and speed.
Paul Smithson,
There are numerous changes and reorganizations in the paper. The one thing that stood out is that the new Fig. 8, which overplots all the startup transients shifted to align with one another, is quite different from the previous version. In the text it is pointed out that there are TWO Doppler compensation algorithms in the AES, with the second, called the “closed loop” approach, being a Rockwell-Collins type correction, and it is used when navigation data are not available. In that case any errors in the OCXO frequency are doubled.
Thanks sk999. If I recall correctly, ALSM had insisted that the AES simply would not transmit if it did not have the required navigational data input. Your post above says that Holland’s revised paper reveals that it uses a different/secondary BFO algorithm in such an instance – suggesting that it could transmit. Is that correct? Does this have possible bearing on interpretation of final BFOs?
@David, @Andrew, @Ge Rijn, @ALSM, @Mick Gilbert, et al.
Understanding what occurred after fuel exhaustion is useful because it helps us decide the distance from the 7th arc to search. However, if we believe the validity of the final BFO values and the associated descent rates (and at this point, I am not aware of a reason to doubt them), then the 0.65g descent means the plane impacted close to the 7th arc. We can speculate about the cause for the rapid descent (and I lean towards left engine restart or pilot input), but barring an extraordinary and skilled recovery on the part of the pilot, the +/-25 NM search width should be sufficiently wide to find the plane.
@Paul Smithson
Re: Chritsmas Island Paper
I am very excited about the new paper as far as flight path past IGOGU to MEMAK. I have also been trying to say that MH370 may have descended to FL280 consistent with moving around on the flight paths. Also nobody seems to mention UAE343 was right on MH370’s tail. The paper is a little wordy, to put it mildly, so I did not see the maneuver you are talking about. I thought it was SLOP and descent with left turn to MEMAK. See how few words it took me to say that?
Anyways we finally have a 3rd party, rigorous path proposal that fundamentally agrees with the Ianello/Godfrey descent+loiter scenario, and it makes perfect sense to me. It is similar to Nederland’s path and my 180S CTH path, except me and Nederland went further west to get out of Sabang radar. Also, on FlightRadar24 I have seen the aircraft holding to the lower FL280-300 altitude when crossing over (to DOTEN) and I also sometimes see the aircraft cutting over from N571 to B466 etc.
So if we say descent at 18:40 makes sense as a valid option, which it probably does, then that says the Bayesian analysis missed an important option.
@Andrew said: If I remember correctly, @Gysbreght commented somewhere that a vertical acceleration of 0.65g would require a bank angle greater than 60°.
That seems about right. The significance of the vertical acceleration is underestimated by many. I believe that either there was nose-down pilot input, or the plane was in a steep bank. A phugoid would not develop such a strong vertical acceleration.
@VictorI
I repeat part of your comment from yesterday:
“We believe that at the time the BFOs were recorded, the APU was running, and therefore the flaperons were symmetric and the induced roll rate small. Yet the descent rate increased from 4,500 fpm to 15,000 fpm in 8 seconds. At the time the descent rate was 4,500 fpm, the bank angle was still modest. Yet, in eight seconds, both the bank angle and descent rates increased to large values. How was that possible on APU power? Something is missing. I proposed the engine re-start as the mechanism for this to occur. People here think there was not enough fuel for this scenario. A nose-down pilot input is another explanation, perhaps in combination with rotation of the yoke. That would be quite effective.”
I still think this could be significant. The final 8 sec. BFO’s would be right but then the timing is off considering an uncontrolled high speed descent regarding the ATSB figure 6 simulations. It’s happening too early and the coincidence with the 00:19 log-on sequence happening at the same time is at least very remarkable imo.
To me it seems the start-up of the APU has been the trigger that has initiated the very high 8 sec. descent rate. While it should have been a stabilizing factor bringing the left wing flaperon in neutral again.
It’s asif it was a sign to dive down rapidly once the APU started up.
To me this bears the signs of a pilot-induced high speed steep descent.
Did not see conclusive arguments yet that refute this possibility.
Following closely all comments on this issue.
@Ge Rijn said: If the high descent rates were induced by pilot-input then there is also the possibility of recovery/pull-out by pilot-input.
China Air 006 recovered/pulled-out using only 1500ft altitude coming out a near vertical high speed dive without running engines (engines restarted after the pull-out).
I don’t think anybody doubts it is possible to recover from a steep descent by a skilled pilot. It does require careful operation of the controls so that wing and tail parts are not damaged by high aerodynamic loads. The use of speedbrakes would help.
I think this issue…could turn out to matter a lot if the plane not gets found close to the 7th arc
If the plane is not found before reaching 29S latitude, there will be debate about whether to search further north, wider, or re-scan areas. We won’t resolve that now.
@VictorI
Just read your last comment.
I can agree even if a pilot-induced recovery took place after the steep descent indicated by the final BFO’s the width of the search area would not be very much wider than +/-25Nm of the 7th arc.
It would all depend on which altitude a possible recovery could have taken place.
If this happened at ~20.000ft the widht maybe could exceed +/- 40Nm.
If it happened at ~5000ft it would be well within the +/-25Nm probably.
@GE Rijn said: To me it seems the start-up of the APU has been the trigger that has initiated the very high 8 sec. descent rate. While it should have been a stabilizing factor bringing the left wing flaperon in neutral again. It’s asif it was a sign to dive down rapidly once the APU started up.
Remember, the log-on and the 0.65g acceleration down occurred about a minute after the APU started, not at the same time.
To me this bears the signs of a pilot-induced high speed steep descent…Did not see conclusive arguments yet that refute this possibility.
Whether or not it occurred, it remains a possibility that is difficult to refute.
Victor: I agree with the main point of your statement at January 9, 2018 at 9:32 am…the +/-25 NM search width should be sufficiently wide to find the plane. But I would add to your list of possible reasons for the rapid descent this: The plane simply rolled to one side or the other, all the way to 90 degrees or more (with or without an attempted engine restart, with or without a pilot) and dropped like a rock. That is what the sim’s suggest (without a pilot). That is also what I see in experiments I have run in my own planes. It does not take much to get the turn going (turbulence, rudder trim, CG, etc.), and once it starts, the plane “augers in” as pilots say.
BTW…from a pilot’s perspective…if I wanted to make an airplane go down rapidly, I would not simply push the stick forward. I would roll the plane inverted and pull back on the stick (standard split S maneuver except you don’t complete the half loop).
@ALSM: Yes, the lateral asymmetry can come from many sources. Once a critical bank angle is achieved, the plane will accelerate down.
Your comment about inverting the plane and pulling back on the column brings to mind the SilkAir 185 suicide by pilot incident. In fact, there are many similarities between MH370 and SilkAir 185. Some are:
1) Flight control surfaces separated from the tail at high speeds due to flutter and were recovered in relatively large pieces some kilometers from the main debris field.
2) The pieces from the main debris field were smaller and shattered.
3) Investigation by Singapore and Indonesian officials pointed towards mechanical malfunction. Meanwhile, the US NTSB ruled that the events and data suggested a deliberate action on the part of the pilot.
The Aircraft Accident Report from the Indonesian NTSC is very interesting, including the description of the recovered debris.
I also found Appendix N, which are the comments from the US NTSB, and the Indonesian NTSC’s comments to the NTSB’s comments, very illuminating.
@Paul Smithson
Re: ”Does this make sense (from a procedural point of view)? If your intention is to turn left, why not turn left away from the airway? Would you really turn right, descend, then turn left in order to cross at 90 degrees the airway you have just left?”
There is no procedural requirement for such a manoeuvre. The established contingency procedures allow aircraft to turn left or right away from an airway, subject to the relative position of other aircraft, adjacent airways and terrain. There is no requirement to cross the airway the aircraft has left, at 90 degrees or any other angle.
@Andrew. Thanks for that. I also couldn’t see the sense in it and I wonder where the authors got that from.
@Paul@sk999
If I recall correctly, ALSM had insisted that the AES simply would not transmit if it did not have the required navigational data input. Your post above says that Holland’s revised paper reveals that it uses a different/secondary BFO algorithm in such an instance
I remain confused by this change as well. My previous understanding was that the AES in 9M-MRO did not perform in this manner i.e. this implementation is unique to a Rockwell AES.
Figure 4 of the new paper also had a change which highlights the fact that the BFO error ar 00:11 cannot be “zeroed” without a slight rate of descent at that time. An observation I made some time ago.
@sk999
I feel like we left your previous post on arc location computation “dangling”. Yes, I agree that the “rings” are really the result of a spherical BTO surface “intersecting” with an oblate spheroid (the earth). So what is the method you use to generate this curve? Do you generate it at all or merely compute BTO error at discreet points? Of course, for a search vessel a path needs to be established. How do you suggest doing that? In my case I used an average earth radius and the sub-satellite point to generate the rings. Of course, there will be residual errors with this approach. The errors are insignificant for path modeling, but I would be skeptical of that approach for search path purposes.
@DennisW said: In my case I used an average earth radius and the sub-satellite point to generate the rings. Of course, there will be residual errors with this approach.
I’m not sure why you’d do that. The BTO can be calculated as a function of latitude, longitude, height, and time. To generate points along the curve, choose a time, height, and latitude (or longitude). Then iterate until the longitude (or latitude) matches the measured BTO at that time.
@Victor
Sure. I know that. That is what I meant by computing BTO at discreet points above. That works perfectly correctly. Is that what you do to generate your BTO rings on Google Earth? Probably not.
@DennisW: Yes, that’s what I use. Why not?
@DW. At aqqa.org you will see (Barry Martin’s) method for generating the long for any given lat and assumed aircraft altitude and time (thus satellite position) for a WGS84 earth and BTO. Two alternative satellite orbit/position models. You can create a series of such lat/long points with a trivial iteration then use that set of lat/long points (with whatever separation you choose) as a csv in gpsvisualizer to create an arc displayed in google earth. At least that’s how I do it.
@Victor
OK. Just seems like overkill to me. I do that for specific arc crossing times and coordinates. I do not do it to generate a graphic.
@DennisW: I use Barry Martin’s ping arcs for graphics, but he iterated to find the latitude and longitude, so his answers are essentially identical to mine.
@VictorI
Your remark:
“Remember, the log-on and the 0.65g acceleration down occurred about a minute after the APU started, not at the same time.”
That’s why I suggested the APU start-up triggered the 0.65 acceleration/8 sec. descent rate. Not at the same time but one minute after, when the APU start-up was completed and provided power to SDU again which then started the log-on sequence.
The plane was already descending at a ~4500ft/min rate at the time the 00:19 log-on started, increasing to ~15.000ft/min in those 8 sec.
This relates to me as a fast (4500ft/min) descent initiated at the moment the APU kicked in going to a high speed descent in the minute after.
This all happening one minute after second engine flame-out does not fit the ATSB simulations imo. And it does not fit the corrections to the hydraulics to the left wing flaperon and maybe other corrections (rudder trim?).
It does fit though a pilot induced descent the moment (or even before) the APU started imo.
Regarding the new Holland Paper and “secondary Doppler Compensation Mode”…
The paper states on Page 6:
For log-ons 1 to 6, the aircraft was on the ground at
an airport. It has been established that the most common
scenario for an on-ground-at-airport log-on after sustained
power outage of the SDU is for the AES to use a different self-
Doppler compensation method than was in use for log-on 7
and the log-on at 00:19Z. This is due to a lack of navigational
input data from the aircraft’s inertial navigation unit in the few
minutes after power-up, and is referred to as the closed-loop
Doppler compensation approach.
The standard AES Doppler compensation mode when navigational
data is available to the SDU is the mode used for
log-on 7 and the log-on at 00:19Z. This approach was briefly
described in Sec. IV-A. It used the aircraft’s own estimates of
its position and horizontal velocity to predict and correct the
AES contributed Doppler shift. This is referred to as open-loop
Doppler compensation.
I think Ian’s description is confusing, and a little misleading (unintentionally). I recently met with several people from Inmarsat and one from Thales. We discussed several issues that have been confusing for me. One was the packet collision at 18:25:27 which I previously discussed here. Another was the question of how the OCXO is periodically calibrated. According to those Inmarsat and Thales people, the “Rockwell closed loop method” is never used in flight. Instead, a modified version of the closed loop method is used to calibrate the OCXO while the SDU is stationary on the ground. This is what happened at KL at 7/03/2014 16:00:23.407 when the OCXO was adjusted by 16 Hz. I have been asking ATSB to let us know how the the OCXO is periodically calibrated, and they tried to pry that info from the vendors, but it was never forthcoming, until my meeting last month. That was a real urika moment for me. Any way, the SDU uses the outbound P channel carrier as a reference frequency, and knowing only that the plane is not moving, an OCXO long term drift bias can be calculated. That bias is then applied to all frequency calculations until the next ground calibration. Between these calibration events, which are always at least 26 hours after the previous calibration, the SDU uses the open loop method that relies on the availability of nav data via 429 bus. Since we know the OCXO was calibrated at 16:00:23.407, and it can’t be calibrated in the air (or earlier than 18:00 the next day), we know the SDU was using the nav data to calculate the Doppler compensation.
@ALSM: We also know the log-ons at 18:25 and 00:19 were Class 3 on the HGA, requiring navigational data.
@ALSM
Any way, the SDU uses the outbound P channel carrier as a reference frequency, and knowing only that the plane is not moving, an OCXO long term drift bias can be calculated.
I am still confused. The fact that the plane is not moving does not alter the fact that the satellite is moving. Does the AES have satellite ephemeris and the ability to adjust for it? The AES would also have to know its position to do this.
Dennis: Good questions. Yes, the AES has the satellite ephemeris from the P channel broadcast messages (although it is not exactly correct as we have discussed; assumed 0 degree latitude). But the EAFC at Perth applies the C band Doppler compensation needed for the the outbound downlink L band P channel to be (theoretically) exactly on frequency. If the EAFC worked correctly (which is did not due to the latitude issue) and the broadcast ephemeris was correct (which it was not), then all OCXO long term error could be compensated exactly. The fact that the EAFC leaves as much as 0-40 Hz of error and the fact that the assumed 0 degree satellite latitude results in some additional P channel error as received at the AES means that the calibration is not perfect, but that does not matter for the purposes of the communication system or our purposes here. It just means that the OCXO is never exactly right, but over the life of the AES, it is never off by more than 100 Hz or so (at L band), which is fine for long term drift control. For MH370, this imperfection has no significant relevance. As long as the OCXO is not adjusted over the time of the flight, no problem.
@ALSM
Thanks. My conclusion is that the BFO calibration we did for MH370 at KL along with the fact that in flight compensation only uses Nav data means that we are as good as we can be relative to BFO i.e. ocxo drift is the only source of large potential errors.
@ALSM. For now, this BFO dual method (open/closed loop) is becoming more complicated, not less.
Are you suggesting that the closed loop is only used for periodic correction of drift (rather than being used for frequency control/compensation at log-on)? That’s not my reading of what Holland is saying. He appears to be saying that closed loop is used for stationary logon on the ground, with nav data assumed unavailable but that open loop, using nav data was used at 1825 and 0019. How can we know that either of these assertions is correct?
Even if it IS correct, I cannot understand how you expect to see a similar thermal transient shape/timing/amplitude with two different control loops (or know that one is “double” the other). Among other considerations, the open loop would be affected by the “anti-Perth” anomaly and the closed loop not.
I hope that we are not going to discover that the timing control (thus BTO) also has two modalities. That would really put the cat among the pigeons.
Paul Smithson:
“Are you suggesting that the closed loop is only used for periodic correction of drift (rather than being used for frequency control/compensation at log-on)?” Yes, that is what Thales said. It is used to calibrate the long term OCXO drift only, and only on the ground. Note that a logon occurring on the ground following a long period with the SDU power off will be preceded by an OCXO calibration. However, a calibration is not always performed before a ground logon. It is only performed after at least 26 hours has passed from the previous calibration.
Regarding the thermal transient, there is always a delay after power on before the calibration is performed. I forgot to note the time, but it is long enough to reach nominal thermal stability. Thus, there is no transient error at the time of the closed loop calibration.
It should be noted that use of the closed loop method for Doppler compensation would imply patent infringement. That is not what the MCS6000 does. Insted, it calibrates the OCXO offset by assuming the Doppler is known (or computed). IOW…Rockwell assumes the Doppler is unknown and effectively measures the unknown Doppler using their patented close loop method. Honeywell assumes the Doppler is known, and measures the frequency offset it attributes to the OCXO long term drift error.
@sk999
My apologies for the delay in trying to answer your most excellent questions:
1. 4th Jan “Can someone explain the fascination with an ending latitude of S30 as being deserving of a higher priority than any other latitude in the general search area?”
30°S only has a probability of 10.3% in my view and is therefore only slightly more probable than other possible MH370 end points.
(a) It matches the satellite data given meteorological data en route.
(b) It matches the fuel range and endurance.
(c) It matches the transoceanic drift analysis for all 25 debris finds.
(d) It matches the debris fields spotted by the RNZAF as per Victor’s previous post.
2. 5th Jan “has there been any serious attempt to intercompare them to determine if they are reliable? ”
Yes, I have compared drogued and undrogued GDP buoys. My database allows you to select either. I am not convinced that the difference is as dramatic as some would have you believe. In any case, the drift modelling is only accurate to determining a MH370 end point of 30°S +/-1°. I recognise, that there are many other drift models with quite different results.
3. 6th Jan “I am probably the only one who modeled the POR data”.
No, I have modelled the POR data from MH371.
4. 8th Jan “How accurate are the BTO calculations? ”
Glad to see we are in agreement to 0.4 NM.
5. 8th Jan “Suppose an inclination, node, and latitude were downloaded to the Honeywell/Thales AES, will the AES correct for the first effect, the second, or both?”
In my MH370 Flight Path Model, I have assumed an inclination of zero for the satellite reference altitude above the nominal sub-satellite point of 0°N 64.5°E.
6. 8th Jan “Conclusion: everyone is using a spheroidal Earth model.”
No, I have a correction for the non-spheroidal Earth model in my MH370 Flight Path Model.
@ALSM. I’ve read and understood what you have written. But this differs from what I understand Holland to be saying. He appears to say that the logon 1-6 used closed loop freqency control and logons 1825 and 0019 used open loop. If that is true then I’d expect that it presents a problem – because the shape and nature of those thermal transients should be expected to differ depending on the control loop in use. And the second problem is even having confidence to asser which control loop algorithm was in operation in which instance(s).
I agree that Holland’s description is confusing, but I believe Thales knows how it works and I believe their description makes much more sense. In any event, we know from multiple sources now that there was no calibration in flight, and we know the logons at 1825 and 0019 were both class 3, so nav data was required.
And if nav data inaccurate or unsettled (as is the case during establishment/refinement of GPS position)? Is there a possibility of anomalies due to inferred movement from one inaccurate fix to a slightly more accurate one? I have certainly seen GPS-based systems doing this – and telling you that your boat [in my case] spent a few seconds jumping at a ridiculous speed from false-position A to more-accurate-position B.
@Victor
@Richard
I took your 18 Great Circle flight paths (thank you!), and ranked them.
I assumed for the 2-hr time period 20:41 (Arc3) to 22:41 (Arc5) that there were no maneuvers and there should be minimal Bias drift during that short period. Thus I am defining “best” path has no BFO Bias drift. I don’t care what the Bias is, as long as it does not drift.
BFO Bias Drift is defined as Abs((Arc5 BFO error)-(Arc3 BFO error)) and I am looking for a best path with zero (0) change.
If a path shows a significant Bias drift during this short 2-hr snapshot, it suggests either the proposed path/heading is faulty, but it could suggest that the aircraft/SDU was not at steady state during this time period. I come back to this latter possibility.
Here is my ranking of the 18 paths from “Best” to “Worst”
Track__Lat-S__ BIAS DRIFT
185.0__37.2___ 0.8 (Best = Least Drift)
184.0__36.7___1.1
183.0__36.3___1.6
182.0__35.6___2.0
181.0__35.1___2.5
180.0__34.3___3.0 (S-Pole path)
179.0__33.7___3.3
178.0__33.0___3.7
177.0__32.5___3.2
176.0__31.9___3.6
175.0__31.3___4.4
174.0__30.6___5.5
173.0__30.0___6.4 (YWKS path)
172.0__29.3___6.9
171.0__29.0___7.5
170.0__28.1___8.6
169.0__27.5___9.3
168.0__26.8___9.6 (NZPG – Worst unless there is no steady state)
Whereas: Lat-S is the Arc7 latitude end point
This ranking suggests either (1) the 177-185 deg True South paths fit the data much better, or (2) maybe BFO was in fact drifting in those 2 hours (due to say flight maneuvers, depressurization, bleed air outage, other on-going upset) in which case we could indeed have a NZPG waypoint. The latter hypothesis could possibly be tested if we knew how a depressurization and/or other events impacted the BFO read-out of the SDU (Satellite Data Unit).
Important Note- I am *not* saying 185 Deg S is the true path, because any scatter in the raw BFO data will shift this analysis towards the 180S paths. Also I still tend to favor CTH or CMH paths, which could skew the sweet point towards 180S. This is a preliminary ranking tool not intended for heading determination (as yet).
After rereading, reading more of the Holland paper (which I should have done first), I am now in agreement that something does not add up. It is more than just confusing language. Holland’s description of when and how the closed loop technology is used in the MCS6000 is completely different from what Thales told me last month. Holland says (bottom pg 6) his analysis assumes (note is was an assumption) that closed loop Doppler compensation was used for 1-6, all while on the ground. He then assumes open loop compensation at 1825 and 0019. That is definitely not the way it was described to me by Thales last month. Thales explained that the closed loop method was ONLY used to calibrate the OCXO bias, and after the calibration was complete, a new logon follows immediately using the open loop Doppler compensation. That is exactly what we observe in the log at 1600. Frankly, I would be shocked to learn that Honeywell was using the P channel for (patented) Doppler compensation at any time, on the ground or in the air.
@Paul Smithson
The AES receives position and attitude data sourced from the ADIRU. GPS data, from the MMRs, is not directly provided to the AES.
@ALSM
Thanks. I was really struggling trying to reconcile Holland’s narrative with my previous understanding and your feedback.
@ALSM. Quite. And now I shall stop commenting in the hope that those more skilled in the art can figure out what Holland is talking about…
@Victor,
@TBill
@Richard
A correction: My comment at January 4, 2018 at 2:17 pm “Dordrecht Hole is apparently beyond fuel range, and the path crosses the 7th arc a bit north from the planned search {28.3s104}.”
Should be 28.3s99.6E, (as per the path to 45s104 “Possible MH370 Paths Along Great Circles. Victor Iannello, 10/22/17).
If the plane is not found by 29S, 28.3S would assume top billing in @TBill’s Bias Drift ranking!
It would be frustrating if the new search treats the simulator evidence as though it doesn’t exist and stopped at 29°.
@TBill
As soon as someone i.e Holland makes a histogram of BFO errors (Figure 2 of his latest publication), I shake my head. I would challenge you or anyone else to display a manufacturer’s specs on oscillator performance that use mean and standard deviation. It is just stupid. When I see that in Holland’s paper my brain shuts down.
The reality is that ocxo drift is not predictable, and it will corrupt any analytics we do.
@ALSM,
I have sent some questions to Dr. Holland to clarify his modeling of closed-loop frequency adjustments for ground-based log-ons. I’ll pass on whatever I learn. My initial reaction is skepticism, since that process produces 2*130 = 260 Hz peak transient corrections for in-flight warm-ups (like 18:25 and 00:19). The contributions to the BFOs at 18:25 and at 18:27 from maneuvers would be large and negative, and you can’t fit both the BTOs and BFOs with turns only. In addition to a SLOP, you would need a high ROD, too, prior to 18:28, that goes to zero by 18:28. Fitting both BTOs and BFOs with twice the transient BFO error gets very complicated.
DennisW,
As far as drawing rings on a map, I haven’t done that yet with the precision we would like here. (Like everyone else, I use interpolation when accuracy is desired.) However, I made the silly mistake of thinking that Fig 14 of the ATSB’s “First Principles Review” of 20 Dec 2016 would be pretty close. What a botch job! The coordinate axes are only loosely correlated with the positions of features on the actual map, and when I try using Google maps to better define the coordinate system, the 7th arc is not particular close to where I would draw it. Hopefully these charts were generated solely for publicity purposes and are not being fed to the navigation system of Seabed Constructor.
Richard,
Thanks for the comments. Hopefully I represented your work correctly. R.E. terminology, by spheroid, I meant oblate spheroid, i.e., not a sphere. The term ellipsoid to me means a triaxial shape of three unequal axes, which is more general than what is needed to describe the reference surface of WGS84.
@DrB
Holland’s response to my query relative to BFO characterization.
We are aware that the oscillator behaviour is strictly speaking, not stationary and ergodic. Fig. 5.4 from the DST Group book indeed indicates this. The first paragraph below that figure states “The mean bias is different between flights and even within a single flight there is evidence of structured variation.” The next paragraph in the book explains that the structured bias variations happen over a timescale of minutes rather than hours, but for MH370 the values are only available approximately hourly, and that is why we did not use a coloured BFO noise model in our trajectory analysis.
You are wasting your time.
@ALSM,
Oops! I put the factor of 2X in incorrectly in my post above. Dr. Holland’s proposed compensation method results in the ground-based log-on displaying twice the true OCXO transient error, if I understand it correctly now. Thus, since those ground-based log-ons are best fit with a 130 Hz peak error, the true OCXO peak error would, in his scheme, be half of that amount, or 65 Hz peak error, and this is what would be seen at 18:25 and 00:19.
So the obvious question is, how does a 65 Hz peak transient error curve match up to the observed 18:25-18:28 BFOs? The answer is, quite poorly.
Consider the 18:25:34 BFO. It is 130 Hz higher than the 18:28 BFOs (for which the transient error will be very small because of their long decay time). So, the 18:25:34 BFO (the infamous 273 Hz) is close to 130 Hz higher than the 18:28 BFOs (when the transient has died out). That 130 Hz comprises OCXO transient error and maneuvers. Well, the transient error cannot be larger than 65 Hz, according to Holland’s method, even if the 18:25:34 BFO occurred exactly at the peak of the transient error. Now we are left with at least +130 – 65 = +65 Hz for maneuvers. To achieve that, one needs either a ROC of +3,250 fpm or a turn to due North (+20 Hz) and a climb at ROC = 2,250 fpm.
Taking the smaller ROC of 2,250 fpm, I believe even that is well beyond the capability of a B777-200 already flying at FL340 at M0.84 at +10C. I think the maximum sustainable ROC at this condition would be about 1,000-1,200 fpm, or half of what is needed to make the 18:25:34 BFO match up.
My conclusion is that Holland’s (proposed?) method immediately runs into a severe difficulty in explaining the 273 Hz. Maybe Thales got it right, and there is no (closed-loop) feedback from the received frequency that directly affects the transmitted frequency (except during the automatic bias calibration procedures), regardless of whether you are on the ground or in flight.
DrB: I am also seeking clarification. As I understand Holland’s new assumptions about logons 1-6, he assumes that the closed loop (Rockwell) Doppler compensation was used for the few minutes until the IRUs were calibrated. I checked the #6 logon at 1250 (for which we have the full unredacted data) and it shows that the LGA was used from 12:50:19 to 12:57:50, after which the HGA was used until 00:19:37. That means no HGA pointing information was required during that 12:50-12:57 time peroid. In theory, that means the Rockwell Doppler correction technique would have been technically possible since the nav data was not needed for the LGA.
Thinking back on the conversation with Thales, they ONLY described the use of the closed loop technique to calibrate the OCXO long term drift error. They said nothing about the use of the closed loop method for Doppler correction at any time. And they said that the open loop method was used for all Doppler corrections. But maybe what they meant was that the open loop method was used for all MH370 Doppler corrections, while remaining silent on the use of the closed loop technique for the special cases of cold starts on the ground via LGA prior to IRU calibration, afterwhich the SDU switches to open loop comp and the HGA. That’s pure speculation on my part. But it is the only scenario that I can think of that makes the Thales and Holland reports consistent. That said, I have a hard time understanding why they would design the SDU to use the pattened Rockwell technique for such a limited purpose, and then switch to open loop compensation for 99% of the transmissions. I hope there is an answer forthcoming soon from Thales.
@ALSM
I can only close my eyes and shake my head. Someone needs to open a big can of “kick ass” with these people. It gets back to my complaint of the lack of a “world class” leader. Sad really. It could have been so much better.
https://tinyurl.com/straitstimes-20180110
Malaysian govt appoints US-based exploration firm Ocean Infinity to hunt for Malaysia Airlines MH370 on no find, no fee basis
KUALA LUMPUR – The Malaysian government on Wednesday (Jan 10) signed an agreement with United States-based seabed exploration firm Ocean Infinity to locate the missing Malaysia Airlines MH370 aircraft, on a “no cure no fee” basis.
The search for the debris and black boxes will begin from the middle of this month, covering an area of 25,000 sq km in the priority search area in the southern Indian Ocean under a 90-day timeframe, the Transport Ministry announced at a press conference.
The fee for the search starts from US$20 million (S$26.7 million) if it is found within 5000 sq km and can go up to US$70 million, depending on where the debris field or black boxes are found and only if they are found within 90 days.
The payment for the search will be borne by the Malaysian government.
Norwegian research vessel Seabed Constructor, which is being leased by Ocean Infinity for the hunt, has set off from South Africa and is aiming to arrive in the search area by Jan 17, the firm’s CEO Oliver Plunkett told reporters.
The vessel is carrying 65 crew on board, including two personnel from the Royal Malaysian Navy. Four of the crewmen had been involved in the search for Air France Flight 447 which was travelling to Paris from Rio de Janeiro when it crashed off the coast of Brazil in 2009, killing all 228 people on board. The flight’s black box recorders were recovered after a multi-million-dollar search that spanned more than two years.
Malaysia’s Transport Minister Liow Tiong Lai said there was an 85 per cent chance that the wreckage would be found within the 25,000 sq km area.
The next-of kin of the victims and their families will be updated via text messages and e-mails as and when new information on the search is available.
The location of Flight MH370 has become a mystery since the plane carrying 239 people on board disappeared enroute to Beijing from Kuala Lumpur on March 8, 2014. Most of the passengers were Chinese.
The plane is thought to have been diverted thousands of kilometres off course out over the southern Indian Ocean before crashing off the coast of Western Australia.
The governments of Malaysia, Australia and China called off the search in January last year despite protests from the passengers’ families, after the A$200 million (S$213 million) hunt which was carried out for more than two years failed to bear fruit.
Several pieces of debris confirmed to have been from the missing aircraft have been found by members of the public on the African coast and islands in the Indian Ocean.
The Seabed Constructor is currently located near 37°42’S 036°45’E, apparently undertaking sea-trials of its UAV fleet in water depths around 17,000 feet.
ALSM wrote “shows that the LGA was used from 12:50:19 to 12:57:50, after which the HGA was used until 00:19:37”
The log shows that the LGA signal path was active until a Log On renewal, exploiting the HGA, occurred at 15:57:49.
The sequence of SATCOM activity from 15:55:57 thru 15:59:55 is consistent with the MH370 route being activated in the FMC and completion of the ADIRU alignment.
A more detailed breakdown of the reward.
http://chinaplus.cri.cn/news/world/10/20180110/76162.html
Thanks for the correction Don. I misread the log. 12:57:50 should be 15:57:50, according to the AES Process log Tab. (Not sure why your time is 1 second different.) The point is the same. The AES was using the LGA for the initial cold logon at 12:50, #6, (and thus could have used the closed loop Doppler correction technique). But the fact that it continued to use the LGA until 15:57:50 and the ADIRU alignment did not happen until 15:55:57 thru 15:59:55 suggests that the closed loop technique may have been in use for the entire period from 12:50 to 15:57:50, not just the original power on logon. Requests for clarification have been sent to Holland and Thales.
This in The Australian this evening.
Ocean Infinity CEO believes in realistic chance to find MH370
The Australian 6:19PM January 10, 2018
Ean Higgins Amanda Hodge
The search for Malaysia Airlines flight MH370 is due to be back on a week from today, with the head of the underwater survey company conducting it saying it has “a realistic prospect of finding it.”
In an exclusive interview with The Australian ahead of the signing tonight of a contract with the Malaysian government, Ocean Infinity chief executive Oliver Plunkett said the ship Seabed Constructor was expected to be on station in the search zone in the southern Indian Ocean on January 17.
“It will launch almost instantly,” Mr Plunkett said, referring to the eight Autonomous Underwater Vehicles, or torpedo-like unmanned mini-submarines with side-scan sonar which Ocean Infinity will use to try to locate wreckage of the Boeing 777 at great depths.
On the way from its last port of call in Durban, Mr Plunkett said, the crew and scientists had conducted some trial dives of the AUVs.
“We have had some pretty good results, as far as I can tell, a dive down to 5,800 metres, which is pretty cool.”
Mr Plunkett said the search would first concentrate on areas identified by Australian authorities and scientists as the most likely based on their “ghost flight” and “death dive” theory that the plane’s pilots were incapacitated at the end of the flight and it crashed down rapidly after fuel exhaustion.
But if it is not found there, Mr Plunkett said, it was possible to divert the Seabed Constructor to another target zone preferred by international airline captains Byron Bailey and Simon Hardy, who believe a rogue pilot hijacked his own aircraft and flew it to the end, ditching it. [my bolding]
“I wouldn’t rule it out,” Mr Plunkett said.
“We are ultimately providing a service to the government of Malaysia.”
MH370 disappeared on March 8, 2014 on a scheduled flight from Kuala Lumpur to Beijing with 239 people on board, mainly Chinese nationals.
Radio contact ended about 40 minutes into the flight, when the aircraft turned around and headed back over Malaysia to the Andaman Sea, then on a long track south, with its radar transponder turned off.
A two-year, $200 million search led by the Australian Transport Safety Bureau at the request of Malaysia failed to find a trace of the aircraft in a 120,000 sq km search zone, and that survey ended a year ago.
Ocean Infinity, a Houston, Texas-based ocean survey outfit owned by a number of British investors which uses cutting edge technology, made an offer to the Malaysian government late last year that it would launch a new search on a “no find, no fee” basis.
Mr Plunkett signed the contract, valued at between $US20 million to $US70 million according to where the wreckage is found, with Malaysian Transport Minister Liow Tiong Lai at a ceremony in Putrajaya, Malaysia’s administrative capital.
From the day the search starts, under the contract Ocean Infinity has 90 days to find the aircraft, collecting a fee only if it is successful.
“I think assuming all of the aviation analysis is right, then I think we have a realistic prospect of finding it,” Mr Plunkett said.
The ocean explorer outfit has leased a Norwegian search vessel owned by the Swire group, and equipped it with new “swarm search” technology, simultaneously using eight AUVs to scour the seabed at depths of up to 4,000 metres.
Mr Plunkett said the mission had to move fast to complete the search before bad weather sets in late autumn.
Ocean Infinity will concentrate its efforts on a 25,000 sq km area just north of the original search zone, following breakthrough CSIRO research last August that significantly narrowed down the likeliest location of the plane to three possible crash locations within a 5000sq km box.
CSIRO scientists say the research was the result of methodical investigation _ including drift analysis of evidence as it washed up on distant shores: first a sheered MH370 flaperon on Reunion Island and then a piece of the aircraft wing in Tanzania.
Ocean Infinity has a staff of 45, of which nearly half are on the Seabed Constructor.
@Victor
Now that I’ve stopped laughing, from you what you understand of OI’s thinking do you think that they’re going to swing down to 40°10′S, 86°30′E to check out Captain Hardy’s preferred site? Has Scoop Higgins got this right?
@Mick Gilbert,
Higgins doesn’t give up does he? He has the knack of using verbatim verbal or email quotes, selectively editing, and adding some narrative to (re)introduce and reinforce his personal agenda. Note how he brings back his ‘ghost flight’ and ‘death dive’ devices, I’d expect those words were not used in his conversation with Mr Plunkett.
@Don Thompson
It’s astounding. As though with a ticking clock, OI are going to sail 400 nm in the wrong direction to check out a site that the airplane probably couldn’t have reached anyway before turning around and sailing 400 nm back so that they can proceed north.
But, of course, Joe and Joanne Public reading the article think it’s genuine and accurate.
@Mick Gilbert: I would be quite surprised if OI went south after the first 25,000 sq km. I don’t trust Higgins’ reporting on this. OI was being deliberately ambiguous, which Higgins latched on to.
@ Good job OI, Ean Higgins and Byron Bailey aren’t as blinkered as the rest of you.
😁
Kind Regards
Rob
@Rob,
Higgins, Bailey (and Hardy, you omitted mention of him) would have no need for blinkers, their comments demonstrate a natural myopia.
It is entirely wrong to conflate anything of OI’s intentions with those individuals.
Again, Higgins is arguing a baseless case on behalf of his pundits’ bruised egos.
Update on Jan 10, 2018. As widely reported, the agreement between Ocean Infinity has been finalized in a signing ceremony. The tiered payment terms are linked to where the debris field is found, and ranges from $20 million if found in the highest priority, 5,000 sq km area, to $70 million if found beyond the 25,000 sq km area. The complete statement from Minister of Transport Liow is found above.
@Mick
From David’s link above:
He (Plunkett) also noted that the company has in their team people involved with experience in Air France crash in north Atlantic in 2009.
I wonder if he is referring to Metron or simply workers on Seabed Constructor.
As far as the 40S notion is concerned, I would not dismiss it casually. The “science” team here and in the ATSB (and subcontractors such as DSTG) have not been confidence inspiring. There was even a recent paper extolling a CI trajectory. When people I know, competent management types, look at the track record they don’t have positive things to say i.e. “how many chances do you want to find the aircraft?”. Let’s be realistic here. Hardly a day goes by that we don’t have PhD types tossing spit balls at each other. We take it in stride which is good. An outsider looking in is a bit taken aback by it.
@DennisW asked: I wonder if he is referring to Metron or simply workers on Seabed Constructor.
I don’t think Metron has anything to do with the OI’s search for MH370. There are other individuals and organizations on OI’s team that have vast experience in underwater searches, including aircraft debris fields, including but not limited to AF447.
It seems to me the conditions of the agreement probably (hopefully) leave enough time to search beyond the 25.000km2.
Some set-backs included (weather, equipement failure etc.) it should be possible to cover the 25.000km2 within 30 days.
If necessary they have another 60 days and ~50.000km2 to search outside the ATSB-area.
From ~32.5S up to ~30S +/- 25Nm would take them another ~25.000km2 and another 30 days.
When still nothing is found they have an additional ~25.000km2 and 30 days to search locations further from the 7th arc above ~32.5S.
I would also be suprised when they go south again. But you never know.
Maybe they know something we are not aware of yet.
Anyway it’s awesome the search is on again and mr. Plunkett seems to leave the two basic scenarios open; uncontrolled (or controlled) high speed impact near the 7th arc or a controlled ditch-like impact outside the +/- 25Nm 7th arc area.
This shows to me he and OI are not confirmation-biased in this regard.
This attitude and the conditions given, give the best possible opportunities for succes imo.
I wish them all the luck. It’s in their hands now.
I’m sure they know their actions will be followed closely by all of us here and many others. And many will keep trying to support them by providing their best knowledge and insights.
And I’m sure the crew onboard will read @VictorI’s blog at least in their spare time 😉
From a new WSJ story:
Malaysian Transport Minister Liow Tiong Lai sounded confident after the government signed the deal with Ocean Infinity on Wednesday to continue the search.
“After looking into their expert’s view, there’s an 85% probability of finding the wreckage at this new [25,000 sq km] area,” Mr. Liow told reporters after a signing ceremony in Putrajaya, Malaysia.
@Damien
“A correction: My comment at January 4, 2018 at 2:17 pm “Dordrecht Hole is apparently beyond fuel range, and the path crosses the 7th arc a bit north from the planned search {28.3s104}.”
Should be 28.3s99.6E, (as per the path to 45s104 “Possible MH370 Paths Along Great Circles. Victor Iannello, 10/22/17).”
Yes that correction makes more sense thank you. But actually my preliminary BFO Bias analysis tends to favor 32.5 to 36-South, assuming BFO Bias was not drifting during the flight.
@VictorI
“After looking into their expert’s view, there’s an 85% probability of finding the wreckage at this new [25,000 sq km] area,” Mr. Liow told reporters after a signing ceremony in Putrajaya, Malaysia.
Wow. That is not something I would expect an official to say – a terrific way to look bad in 90 days.
Thinking more about the 90 days I can make sense of it from two persectives:
1> OI probably asked for exclusivity (I would have) and that was Malaysia’s response. You don’t want to grant exclusivity without a time bound.
and/or
2> Accountants (finance ministers…) don’t like to have an unbounded liability on the books. It entails a “finance footnote” going forward.
The link shows a quick summary of an explicit path calculation I’ve made from 18:25 to 00:19 assuming very basic interpolated residual Doppler (D(t)) and distance (r(t)) functions. To make the interpolations I used the 18:25:27 BTO value and a 18:39:58 BFO value of 88 Hz to extend the usual graphs to pre-19:41. I did not attempt to model a major turn, instead I was interested in the path crossing route N571. The outcome suggests a typical two-turn scenario, including a dip in the groundspeed between 18:40 and 19:41. As you can see the time of crossing N571 is around 18:42, which is late (should be pre-18:40). Given the simplicity of the approach of course not unexpected that there can be slight inaccuracies in the results. Nevertheless I hope a useful illustration of a possible scenario.
These results are based on little or no BFO bias drift. In case of considerable bias drift, the end point can significantly deviate from the predicted 34.8 degrees south.
https://www.dropbox.com/s/ghm9w3t5yns6v2x/Path1825_0019_v3.pdf?dl=0
@Victor @all
Re: Digital Flight Data Recorder
When MH370 crashed, I believe that former NTSB Greg Feith said the FDR data would be interesting even it was turned off in the EEBay, because we would at least know the data up to that point. But is there some way to erase it, like landing and starting a new flight?
Notice I am starting to get more optimistic about finding the aircraft, but I am not too hopeful we can ever answer the questions due to political sensitivities of the accident.
@DennisW said: Wow. That is not something I would expect an official to say – a terrific way to look bad in 90 days.
More like 30 days. The first 25,000 sq km will probably (depending on when the search begins) be completed during the first swing.
From Hal Hodson to Jeff Wise:
“Dear Jeff
Jeff, now that the contract has been signed and my reporting shown to be completely accurate, as I knew it would (but could not write about at the time), I wonder whether you’d consider removing or updating this post about my piece, which is completely inaccurate
http://edition.cnn.com/2018/01/10/asia/mh370-ocean-infinity-search-intl/index.html“
@Richard
I’ve read it. No reaction yet from Jeff. He must be thinking.
It can be hard to be confronted with undenialable facts we all know..
Hope he shows up to be the man of character that he has been for several years in this endavour.
The 90 days time frame does not seem to be a very strict one, this is just to ensure the search does not go on forever. There seems to be quite a bit of extra allowance based on progress of the search.
‘Its priority is to locate the wreckage or the flight and cockpit recorders, and present credible evidence to confirm their location within 90 days, Liow added.
“They cannot take forever or drag it on for another six months or a year.”’
https://www.reuters.com/article/us-malaysia-airlines-mh370/malaysia-to-pay-u-s-firm-up-to-70-million-if-it-finds-missing-mh370-idUSKBN1EZ0OA
@Ge Rijn
It can be hard to be confronted with undenialable facts we all know..
I don’t know about that. The flaperon did not give him much pause.
Official statement from the OI website:
https://oceaninfinity.com/ocean-infinity-continue-search-missing-malaysian-airlines-flight-mh370/
@Ge Rijn
Delete.your.account.
@DennisW
The flaperon was the first piece found and surrounded with barnacles. Jeff bit in to this and rightly so for it was an unsolved problem.
But all later findings including the RR-piece reduced the questions on the flaperon barnacles to a non-issue in the whole context of findings.
It became clear several scenarios could explain the barnacle growth on objects and also the rather fast disapperance of barnacles after debris had beached.
It seems Jeff got stuck on the flaperon and never considered the ~25 pieces that were found afterwards seriously in this regard.
But this is up to him ofcourse to answer.
@Don Thompson
I won’t give you the pleasure.
This is up to VictorI.
Please stop your narcissitic devaluations.
I’m no threat to your devine superiority.
@TBill
Re: ”the FDR data would be interesting even it was turned off in the EEBay, because we would at least know the data up to that point. But is there some way to erase it, like landing and starting a new flight?”
No, there is no easy way to erase the data.
@TBill
I know we have talked about this before, and concluded that it is likely not a problem for the low temperatures at the black box locations, but data loss over time is still a small concern.
https://www.anandtech.com/show/9248/the-truth-about-ssd-data-retention
I’ve now created a .kml file with my rendition of the 7th arc and created a simple Google Map version, which you can find here (top link):
https://docs.google.com/document/d/14hleZyx1pUPL44yaeHKt6jnSQ3DbgRq2zibbKkFLq2c/edit?pref=2&pli=1#
Almost none of the maps and figures in the various ATSB reports has this drawn exactly correct. (The DSTG book does seem to get it right.) In First Principles (Dec 16, 2016), Fig 14, which shows the 25,000 sq. km. search zone, the coordinate axis annotations are just wrong, although the 7th arc is close to my version when matched to the seafloor features. It looks like the search zone reaches to -32.6 latitude on the arc itself; that should encompass virtually all autopilot routes, but it doesn’t quite reach to the best-fitting data driven route (i.e., a route that crosses the arcs at the right time and best fits the BFO data.) The probability distribution for the latter peaks at around -32.4 and extends Northward with no well-defined cutoff.
Unfortunately, -32 falls in the region of the Diamantina Fracture Zone and Ob’ Trench leading up to Broken Ridge, which might be difficult for the AUVs to navigate. One hopes that OI is willing to try.
sk999: please post a link to your 7th arc .kml
For those interested interested in the story of the search for the Central America (where Larry Stone was a major player in defining the search zone), here is a link to the first Appellate Court ruling, which gives a lot of background information about the search:
https://coast.noaa.gov/data/Documents/OceanLawSearch/Columbus-America%20Discovery%20Group%20v.%20Atlantic%20Mutual%20Ins.%20Co.,%20974%20F.2d%20450%20(4th%20Cir.%201992).pdf
It is quite an entertaining read, involving, among other things, a discussion of the “law of finds” v. the “law of salvage,” plus the story of how an Order of Catholic Monks (the “Capuchins”) came to be positioned to receive a share of the gold.
@DennisW
The subject of solid state storage employed in the memory modules of the DFR and CVR has been previously discussed: NAND or NOR.
ATSB recently posted information on a Loral data recorder memory module from the same ‘era’ as 9M-MRO. Fig.5 in the linked doc shows Intel ‘Flashfile” ICs, type F28F008SA-120, in a TSOP package.
9M-MRO was equipped with a Honeywell 980-4700 type DFDR, same type as F-GZCP. BEA’s strip down of F-GZCP’s recovered Honeywell recorders shows similar TSOP packaged ICs.
Perhaps tenuously, the Honeywell devices appear similar to the Loral unit, the TSOP IC packaging is typically used for memory devices.
The Loral’s Flashfile devices are NOR devices manufactured using Intel’s ETOX process – EPROM Tunnel Oxide. So, NOR, not NAND. NOR devices have the advantage of byte-by-byte writes whereas the NAND devices must be written in page increments which implies the system design must buffer writes before commiting commiting to persistent storage. Buffering is not a good practice for a forensic data recorder.
ECC and mirroring would be sensible design mitigations for the data recorder memory module reliability.
I am confident that the DFDR/DCVR memory module data will be readable.
@sk999 @Victor
Yes I hope they give decent attempt at Broken Ridge as that’s where it could be, if they cannot find it south of 32.5 South.
Victor sk999 makes a good point that data possibly points to 32.4-South and that still needs a margin of error tacked on. Broken Ridge unfort is the margin.
I would take a similar flight path view but suggest there could be some maneuvers at the end getting a little more easterly.
@Don
Thanks. Great input.
ALSM,
I added a more direct link to the file, but Google still insists on showing you a map first. Here it is:
https://drive.google.com/file/d/1-HW09ZseiRAYiNxq5B-KK8dltSv3zpCS/view
This version has a more obvious download button at the top. (The previous version hid it behind an inscrutable icon).
@DennisW
Re: “As far as the 40S notion is concerned, I would not dismiss it casually.”
G’day Dennis, my reasoning for thinking that Simon Hardy’s calculations are probably not correct are:
1. He uses a simple approach that assumes a constant ground speed over five hours. Given the possible airplane flight modes and winds en route, that’s nigh impossible.
2. He doesn’t appear to have given fuel consumption any thought at all. Unless the airplane turned south-west just after 1825 and overflew the western end of Sumatra it probably couldn’t have reached 40°S.
3. The absence of any debris on Western Australian shorelines seems to exclude impacts around 40°S.
My good mate Byron (who despite having flown B777s has on two occasions stated publicly that they are fitted with three transponders) doesn’t appear to have performed any calculations whatsoever to determine his nominated search zone, he just waved his finger over the map “… at least 400 kilometres south and west …” of the ATSB 120,000 km² search area. Forget fuel consumption, the airplane simply couldn’t have flown fast enough to get there by 0019.
Never say never, right? but with 90 days on the clock I can’t see OI allocating the best part of a week for an excursion and search campaign around 40°S.
@Ge Rijn
I seem to recall you professing to some expertise in the field of psychology? Or was that simply another professed, but unfounded, opinion.
It has been noticed that the frequency of your posts at this blog is among the highest of all commenters, but it’s also noted that the substance of those comments is typically unsubstantiated opinion, and even when challenged, the opinion is simply reiterated.
The absence of any reinforcement for an opinion, merely reiterating it, is truly symptomatic of a narcissist: here’s my belief, it is sacrosanct, it is undeniable!! Throw in a dash of psychological projection…
You misconstrue my ire: I can chose to ignore your junk for what it is, but readers may perceive that any and all views espoused here have equal value. Your opinions have been challenged and the responses are inadequate.
I would have no objection to an accusation of pedantism. I strive to be factual and present authorative information, sometimes I’m wrong, but I forthrightly reject your slur of narcissism.
Refer to earlier comment. Isn’t there another forum somewhere…?
@sk999
Thanks. I did not actually do a 7th arc (but I will in bit) since I had no interest in the 00:19:xx data. However, it looks like my 6th arc is just about perfect based on ~450 knots and ~180.
@SK999,
Just download the .klm file from the link on your site, and then drag it into google earth. Works fine. Your arc seem to be displaced slightly [perhaps 500metres] from the rather old arcs that I have, which I think originated from ALSM, or perhaps Barry {Aqqa]. Close enough for all practical purposes though.
Brian
I would be interested in seeing DennisW’s 6th Arc kml if it is linked somewhere, and his 7th when he does it.
Just going back to the “refraction” issue of a few days ago, thanks for the detailed explanations of how minor and thus insignificant it is for our purposes.
@sk999
Do you have a write-up on the 32.4 South best fit work?
Brian Anderson,
Great. At the level of 500 m, choice of altitude matters (I used 6 km, ~20,000 feet.)
TBill,
If you follow the first link I posted and scroll down to my report from 2016-Jan-18, it is in there.
@Brian, sk999
The ‘aqqa’ arc generation XLS workbook permits an arc to be generated at any altitude & BTO bias. On my list of ‘interesting things to do’ there’s been the idea of creating a 3D ‘shell’ of the 7th arc in KML so as to readily visualise its true appearance as altitude decreases.
@ventus45,
At the same time as you were having the discussion over refraction, the following article relating to a couple of upcoming NASA missions was published.
https://tinyurl.com/nasa-gold-icon
Just to prove that there are still some unknowns in the earth’s surface to space zone.
@Brian Anderson
Have you had a look at the new Christmas Island path paper?
I see they reference your papers in the back, but I did not see a reconciliation with your speed calc. They are going SLOW but straight. Would you rule that out?
I do like some of their ideas up to 19:41, but I’d probably speed back up by 19:41 and head south.
@sk999,
Yes, of course. I was comparing it to the arc at FL350, so that would account for the difference.
Brian
@Ventus
I will link both (6th and 7th arc kml) when I get around to doing the 7th arc. This is good stuff IMO. Thanks again sk999.
sk999: Here’s a comparison of your 7th arc to the arcs I have been using for the last 3 years, all generated with Barry’s spreadsheet. https://goo.gl/B9x3Gb
@sk999: Likely differences are from the satellite position. What exact position did you use?
@TBill,
Yes, I have looked at the Jean-Luc paper, and I have commented to them and received a reply.
I cannot agree with their track immediately before and after 19:41. When looking at their scenario it seems that the path between 19:41 and 20:41 is relatively straight, however the path is crossing the 19:41 arc from inside to outside at that time. This means that the minimum LOS distance to the satellite must have occurred earlier than 19:41. Indeed the diagram illustrating the flight path illustrates this.
I think the the BTOs, and hence the derived LOS distances can demonstrate that this is not correct. I believe that there is a clear minimum in the BTO and the LOS distance from the satellite after 19:41, and not before. That means that the ground speed at that time cannot be less than about 395 knots. A number of us have independently calculated that the minimum LOS occurs at about 19:54, and that being the case the ground speed is in the region of 485 knots.
The response from Jean-Luc tried to justify the lower speed and lower altitude on the basis of differing measures of the arc radius, and hence BTO, at a different altitude. To me the argument was rather circular and didn’t make sense. Using the derived LOS distance [in the region of 37,000km] effectively consigns altitude variations to the insignificant category, and also avoids any debate over the radius and shape of the earth.
Brian
@sk999 and ALSM,
Woops, confused myself again. 500m from the original 0ft 7th arc, not the FL350 arc.
When I check the arcs in Mike’s latest post against the old ones I have been using there is absolutely no difference. Hence I’m pretty sure they were originally derived from Barry’s spreadsheet.
Brian
ALSM,
What time and BTO does he use? I took the average of the last 2 BTOs (18390) and used a time of 18:29:37. The yellow track in your figure is about 1 arcmin left of where I would draw it – not clear what is going on. Finally, which ATSB arc were you using when you compared with the spreadsheet?
Victor,
Here’s my satellite position (ECEF):
18178.570 38050.820 391.836
From my notes, there is a maximum 1 km difference relative to Inmarsat’s tablulated positions in my satellite model.
sk999:
PAR5 vectors at 00:19:37
0:19:37 87577 28177 18178.339 38050.821 392.398 0.0014600 -0.0015550 -0.0831630
Looks very close to yours.
sk999: The spreadsheet was developed by Barry, but I ran all the numbers.
WGS84 PARAMETERS Constants
a (m) 6378137 Bias K (s) -0.495679
b (m) 6356752 c (m s^-1) 299792458
e^2 6.69438E-03 Diagnostic OK
PERTH LES POSITION
ϕ deg, LAT λ deg, LON h (m), ALT
-31.80463889 115.8874444 46 Position as agreed with Bill Holland;
X (m) Y (m) Z (m) but with our ellipsoid height.
-2368818.369 4881110.799 -3342065.185
sk999: I have not found thew BTO value I used, but from memory, I’m pretty sure it was the 00:19:29 value (23000-4600=18400 usec).
ALSM,
Same Perth position. X, Y, Z differ by tens of meters. I don’t include altitude (46 m), which would explain. Just over half the length of a B-777. (Plus constant K is tweaked to give correct BTO at gate C1, which largely absorbs the altitude). Same deal for Burum. Need to look elsewhere.
sk999: I found one difference you might want to look at. The average BTO value you used does not agree with the info I have. According to my records, the 00:19:29 value is 23000-4600=18400. But the second value at 00:19:37 is
49660-5*(7812.5)=18410. How did you get 18390 usec?
@sk999
When I run your model (which I coded up in Python) the results I get at 00:20:00 (the listed ATSB table time) I get:
x = 18180.6 km
y = 38055.6 km
z = 390.2 km
ATSB Table 3 values
x = 18178.4 km
y = 38050.8 km
z = 390.5 km
“49660-5*(7812.5)=18410” should read “49660-4*(7812.5)=18410”.
@TBill
Yes, I was just putting a positive spin on 28.3s being down with the worst on your list. That if wreckage hasn’t been found south of 29s,(and with any luck it will be), that 28.3 is not that much further (comparatively).
I regard the simulator evidence as primary information from the perpetrator and I hope the new search takes it into account if they arrive at 29s empty handed.
ALSM,
I used 7820 from Bayesian Methods, where you use 7812.5.
The first BTO value is an R600 channel BTO, and we saw from MH371 and the 18:25 logon for MH370 that those could have errors of 100 microsec. The second BTO is anomalous but correctable, as we both did. It could be that the “empirical correction” of 7820 was skewed by the errors in the adjacent R600 BTOs. More work might narrow it down.
Another things that needs to be checked is the latitude/longitude used for KL. Most of the calibration BTOs came when the plane was at gate C1; some of the later BTOs came after pushback and possibly the early stages of taxi. I don’t remember now what I picked; some of the BTOs (and ACARS data and FR24 data) came as the plane was taking off, so lots of positions one or more NM apart were involved.
DennisW – you might have a slight coding error. Here’s what I get for 00:20
x = 18178.6
y = 38050.8
z = 390.0
which misses by 0.2, 0.0, and 0.5 km.
@Damien
“I regard the simulator evidence as primary information”
You and me both to the extent I have done quite a bit of FS9 simulator work now, the only reason I did that was MH370. My first favorite path was of course McMurdo, but when DrB came out with his path, it fit the BFO more like what I was looking for. DrB’s path gave me the idea that 180 South was a good base case to start with, and that’s as far as I got. But none of us knows for sure.
Update –
I am using the coordinates of gate C1 for the BTO calibration:
Lat = 2.7468, Long = 101.7125 (should be 101.7126 – sorry about that)
Note that a Boeing 777-200ER is .0005 degrees of arc in length.
sk999: Re “I used 7820 from Bayesian Methods, where you use 7812.5.”, 7820 was an early approximation. Later, it was noted that one of the clocks in the SDU has a period of 7812.5 usec, and that clock is suspected as being the source of the variable additional delay. We know the anomaly can be N*7812.5 usec, where N is a small integer, 4 in this case.
From the Dec 3, 2015 ATSB update:
“The manufacturers of the SDU and the ground station equipment were unable to determine a specific reason for the offset, however they did note that the reference clock used for the time slots was 128 Hz, which equates to 7812.5 μs.”
@TBill
I’m not an aviator, so I’m not able to determine the technicalities myself, my main point is that the simulator evidence be utilized. It’s for others to work out how best to utilize it. I was just working on the logic of trying the path to 45s104 as it’s a piece of evidence to work with. There might well be better options, but as long as it’s all considered.
@DennisW. Your interest in the business side.
In the Nederland Reuters link above, “The company’s shareholders would bear the upfront costs of the search, Plunkett added.”
Can we suppose that there a controlling (and possibly sole) shareholder who did the agreeing, since canvassing all in many businesses would be impractical?
In other words, he/she would take the bulk if not all of the risk; and would be adventurous or philanthopic but either way would be well-heeled?
@Don Thompson
Your personal opinion on me and my writing is of no concern here. You make repeating devaluating comments to me and not only to me. You permit yourself to judge on who should write on this blog and who’s account should be deleted. With this behaviour you put yourself high above others and also above the moderator. You won’t silence me. I won’t let anyone who shows narcissistic behaviour silence me.
The only one who can silence me here is @VictorI.
I suggest to put this way of commenting to rest from now on.
@Mick
Simon Hardy’s YouTube explanation was very instructive for me when I got started. So I went back to see why I departed from his example. He picked Arc5 to Arc6 as a steady state period, and he ignored Arc3 to Arc4. I use Arc3 to Arc4 to Arc5 as the only possible steady state periods. I do not think of Arc65 to Arc6 as a steady state period. Also he using BTO no BFOs.
Where does he get that wall sized SIO map? I could use a few of those. But I do the same lines on Google Earth that he does.
@TBill
G’day Bill,
Captain Hardy’s videos are interesting, for sure. I think you’ll find that like you he uses arcs 4, 5 and 6 for his calculations. He assumes a constant (ground) speed over the corresponding two and a half hour period (I had mistakenly said 5 hours earlier). Given the changing winds (from mild easterlies around the 4th arc to strengthening westerlies around the 6th arc) how likely do you think a constant ground speed is going to be?
Given that this bloke is meant to have been a B777 pilot I’m somewhat astounded by his failure to consider endurance amd range given the fuel load. In order to get to where he thinks the airplane might be in the known timeframe it would have had to have turned south-west pretty promptly after 1825 and flown at MRC. He must have known that that means that the airplane would be slowing down as it went along. That’s another nail in his constant ground speed coffin.
If he had have ended up saying, “Look, given the limitations of these calculations we’re going to end up with an area of x km² that the airplane may have come down in” I might have given his work further consideration but he doesn’t, he gives us an X marks the spot outcome with a degree of precision that is well beyond that possible given the data and his method.
Like you, I do like his big SIO maps complete with the arcs. PrintExpress?
@Victor. Despite my (at least) faith that the First Principles Review would have got it right in endorsing the ATSB/Boeing conclusion that simulations were consistent with the final BFOs, my sub-conscious nags me still.
With the APU running the roll should decrease if anything, as I have mentioned, leaving the descents unexplained unless there were residual engine thrust inducing yaw and roll. Yet the Boeing simulations did not include either engine or APU starts.
Thus even if engine run time and thrust were higher than expected, obviously that would not lead to bank in those simulations. Also, while a pilot might have been responsible, as has been discussed, clearly this was not what led to the simulation replication of the descents in the ATSB/Boeing scenario.
That leaves the lack of APU start in the simulation.
Theoretically the RAT induced roll would continue if that were not included.
If Boeing/ATSB /SSWG have overlooked the possibility that their simulation outcomes would be falsified by that, we might be left with a pilot as the only possibility. I am sure @Ge Rijn will be following.
@ Ge Rijn
You have made over 250 comments on this website in the last 3 months, including how other people have made “serious basic mistakes” and got it all “wrong”. I have requested to see your flight path reconstructions, drift model or debris analysis. You have not shared any of your work on these subjects. However, you set yourself up in judgement over others and openly admit your aim is just to criticise. When others answer your criticisms, you ignore the answers and repeat your criticisms. You are a one way street.
Other commenters here share their work, for example, DrB, TBill, ALSM, Don Thompson, David, Andrew, sk999, Mick Gilbert, Kenyon, Brian Anderson, DennisW, Richard Cole, Niels, Barry Carlson, Sid Bennett, etc. We do not always agree, but then we have a chance to see where we differ in our analyses and gain insights that allow our collective knowledge to move forward. This is two way street.
As Don has suggested, there are other websites, which are more appropriate to your aims. My problem is not that you disagree with me and others, my problem is that you are not willing to engage in a constructive dialogue.
@Richard
I study this mystery from the beginning from all possible angles within my capabilities. I share my thoughts, interpretations and ideas based on this.
I always try to contribute in a possitive way. Sometimes my contributions have also led to better/wider insights in some ways. I know I have an unorthodox approach and diplomacy is not my strongest asset. I don’t produce papers like many of you which I all respect a lot and often mention this, including you. And I study them all with great interest.
My choice of words is not always correct and ‘tactical’ I know. Like my pointing out a ‘basic mistake’ in your drift-analysis. I did not mean to put it that strong. I just noticed the difference between drifting speeds of drogued and undrogued buoys was huge. And since your model seems to be basically based on drogued buoys I felt it could be important to bring this forward. I never said your drift-analyzis were completely wrong. This would be total nonsence.
I think you over-reacted by immediately calling out to ban me from this site. Instead of engaging the point I made in a constructive way on it’s merits. I understand your initial reaction though. The way I put my critique on your model was not ‘nice’. Excusses offered.
The rest is up to others to decide. I’m not aware and don’t recognise that I’m unwilling to engage in constructive dialogue.
On the contrary in my own perception.
But if this is also the opinion of @VictorI I will consider your suggestion.
@Ge Rijn
There you go again with your false accusations! You can’t help yourself!
The difference between drogued and undrogued is NOT huge.
I did not over-react to your comments. I reacted to the 21st comment aimed @Richard.
I tried patiently to explain my drift papers and answer your criticism.
You ignore my answers and repeat repeatedly the same criticism.
I am not alone with my views.
@ALSM
DennisW – you might have a slight coding error. Here’s what I get for 00:20
x = 18178.6
y = 38050.8
z = 390.0
which misses by 0.2, 0.0, and 0.5 km.
What value are you using for rs? I was using 42170 km.
Googling shows 42164 km. When I use that value I get what is listed below. Not sure where I got the 42170 number from.
x = 18178.0
y = 38050.2
z = 390.2
|delta_x| = 0.4 km
|delta_y| = 0.6 km
|delta_z| = 0.3 km
@ALSM
Sorry, I must have you on my brain. I meant @sk999 above.
@David
Can we suppose that there a controlling (and possibly sole) shareholder who did the agreeing, since canvassing all in many businesses would be impractical?
That would be my assumption. I have not been following any discussions on the OI pedigree.
@Ge Rijn said: But if this is also the opinion of @VictorI I will consider your suggestion.
I suggest that if you re-phrase some of your strong assertions as opinions and questions, all will be fine going forward.
@Richard
Yes, you explained your model takes both drogued and undrogued buoys into account. You explained you can switch between them in your database etc.
That was more clarifying to me but still left unanswered the point I made about difference in drifting speed between drogued and undrogued buoys.
You state a difference in drifting speed of 0.7cm/sec(drogued) and 8.6cm/sec(undrogued) with 10m/sec wind is not huge. I suggested the difference over a considerable lenght of time (I took 500 days as an example) is huge.
You never explained why my suggestion was wrong (if it was wrong).
But let’s leave it here. It’s probably not worth the hassle anyway.
@VictorI
Thanks. I’ll sure keep this mind and try to improve on it.
I see my part of the problem better now.
@sk999, @ALSM: You raise some interesting points. First, I don’t there is justification for a correction of N*7820 μs, but there is for N*7812.5 μs, which is based on the design of hardware. Second, we have two BTO values. According to the DSTG statistical analysis, the first at 00:19:29 is from a population with a standard deviation of 62 μs. The second at 00:19:37 is from a population with a standard deviation of 43 μs. I have considered weighting the two values by the inverse of their variances. However, that assumes the two values are not correlated. In fact, we know that a lot of the BTO error arises from the 24 μs sampling period that is used by the AES to synchronize its output to the received P-channel signal. As the 8 second time period between the two BTO samples is relatively short, and the satellite and the aircraft have moved a short distance, the two BTO samples are strictly speaking not independent.
In a nutshell, I propose that the 7th arc be calculated using the value at 00:19:37, and be corrected as Mike suggested, i.e., 49660-4*(7812.5)=18410 μs, with a standard deviation of 43 μs. Do we have reason to believe this standard deviation is incorrect?
@David: If Boeing/ATSB /SSWG have overlooked the possibility that their simulation outcomes would be falsified by that, we might be left with a pilot as the only possibility.
I agree that the Boeing simulations have some unexplained characteristics and did not capture all of the system interactions (such as the engine restarts, APU start up, RAT-powered flaperon asymmetry) that we know may influence the trajectory. We disagree on the possibility of an engine restart and the effect on the trajectory. In the Level D simulations that Mike witnessed, the (failed) restart of the left engine occurred and affected the roll rate. I don’t rule out that this occurred for MH370, despite your belief that there was not enough residual fuel available. If it did not occur, I agree that pilot input becomes more likely.
But again, considering the high value of downward acceleration, independent of the cause of that acceleration, I think it is likely the plane will be found close to the 7th arc. In the event of pilot input, I think it is unlikely (but not impossible) that a steep descent was followed by a skilled recovery and glide.
@Victor
Where does the 43us value come from? Thx.
@DennisW: On page 25 of Section 5.2, the DSTG advises that for anomalous R1200 messages, a standard deviation of 43 μs be used.
@Victor
On page 27 of the DSTG Book it is stated:
For the R1200 messages, the empirically derived standard deviation of the measurement noise wBTOk is 29µs, and for R600 messages, 62 µs. For anomalous R1200 messages a standard deviation of 43µs was used.
Got it. The key being “anomalous”. I’ve been using 29us value of the larger population.
@DennisW: Yes. I was speaking about the BTO value at 00:19:37 (and the one at 18:25:34).
Seabed Constructor is headed east at 14 kts, now east of the recent test area. Blaine confirmed they are headed to the test area now.
@Ge Rijn said: Hope he shows up to be the man of character that he has been for several years in this endavour.
Have you been paying attention? Anybody that presents evidence of a flight to the SIO is accused of Russian collusion. Suggesting that Blaine Gibson is a re-activated Russian agent shows character? Rather than acknowledging the fine reporting of Hal Hodson, my guess he will continue to search for evidence to smear the reputation of Ocean Infinity and to link them to Russia. In doing so, he’ll have a lot of cheerleaders on his blog.
@Mick
“Captain Hardy’s videos are interesting, for sure. I think you’ll find that like you he uses arcs 4, 5 and 6 for his calculations.”
No I use Arc 3, 4, and 5. As you say the mighty wind between Arc 5 and 6 is one problem using that ring for data fit. Also of course I envision descent and slow down could have started before Arc6, therefore I am not ruling out paths where Arc6 has the appearance of being too close to Arc5.
@VictorI @David
Just an assumption/opinion (yes I’m trying to be carefull..) on a possible sequence just after second engine flame-out without pilot input:
-A/P disengages and the drag of the already windmilling right wing engine tends to pull the plane in a right bank/roll and yaw direction while the left engine is spooling down.
-the RAT is deployed on the right side of the fuselage and adds to this right bank and yaw direction.
-with loosing hydraulic pressure in the meantime the leftwing flaperon goes to his upward position correcting the right bank angle/roll and yaw to wings level again together with the now also fully windmilling left engine causing a slow left bank/roll.
-then the APU sets in (~one minute after second engine flame-out) which provides hydraulic power to the left wing flaperon again forcing it in its neutral position. The (left) bank/roll gets corrected again to a certain degree towards wings level. Only infuence now is the RAT that causes some right direction yaw and some right bank maybe (I don’t consider the rudder-trim in this thought-experiment)
-a brief restart of the left engine would at least relieve some of the windmilling drag pushing the plane further to wings level or to a right bank/roll again. Certainly if the left engine produced some thrust also depending how long the thrust continued.
-result would be IMO! a spiral dive in a turn to the right.
Problem here though IMO! a sequence like this (and the ones known to me) can not explain the final BFO’s and high rates of descent measured just after the minute the APU was fully operative and the SDU started to log-on, and the ‘dive-descent’ just happened to be precisely during this final log-on sequence.
I cann’t comprehend this yet.
Maybe someone can clear this up for me? (Yes Victor I’ll try to end with a question if appropriate..)
@Brian
Thank you…I hope Jean-Luc would attempt to restructure his pre-1941 path to meet your inside-out criterion. After that I think I know what to with it.
@sk999
Your Jan_2016 report is great. Interesting that the ACARS data is “injested” by the weather mapping services. Of course as you say not too many flights in the SIO. I wonder if an update version might be important if to define the search area.
@VictorI
I know but I have seen different times with Jeff Wise long time ago.
Still wished he’ll come back to reality. It’s not for nothing that I switched over to your blog when you started. I’ve been very critical to Jeff on his blog for the reasons you mention, till I gave up.
Still I think it’s a waste. Jeff has produced some good and daring articles in the past IMO.
With his latest on Hal Hodson’s OI reporting he missed the plank completely again.
Just had to assert this statement to him for I still think it’s a waste he has fallen to this level without giving him a change to redeem himself.
IMO he also made important contributions in the past.
@Ge Rijn: Regardless of what’s said other places, we should be very happy about the current state of affairs. Contributors here have expressed a surprising amount of consensus and optimism about the re-started search. For some months, I have been encouraging contributors to re-examine the evidence and propose where OI should search if the initial 25,000 sq km is unsuccessful. I think that effort has been productive and useful.
@Ge Rijn
The APU start is not harmoniously restorative.
After the second engine fails, the transfer buses lose power, and 15 seconds elapse before the RAT deploys and the APU start is initiated.
The PFCS degrades to ‘secondary’ mode, it does not automatically restore to ‘normal’ even with the APU running.
The RAT takes only a few seconds to extend and spin-up to deliver limited electrical and hydraulic power. The APU requires at least 1 minute before it is generating power. The APU takes 1 minute to start on the ground, likely slightly longer at altitude after a cold-soak, but 1 minute is OK.
You mention the left flaperon issue experienced during RAT operation. While the operation of the APU ‘fixes’ the left flaperon issue, the PFCS remains in ‘secondary’ mode.
A significant characteristic of the PFCS involves processing of the elevator control law, in ‘secondary’ mode processing of this control law is degraded.
Summary at 75 seconds: a left wing down roll has been initiated, its pitch stability is compromised.
Summary after 75 seconds: the roll input has been neutralised, pitch stability remains compromised.
Since you’re a ‘ditcher’, you may be interested to learn that the PFCS ‘secondary’ mode behaviour for flaperon control when the flaps are not retracted is for a static 20deg down ‘droop’. There is no gradual blend as flaps extend through the lever positions. So, to consider flaps not retracted prior to fuel exhaustion (or during APU run) implies that transition to RAT only power results in the left wing down roll momentum even stronger than with flaps retracted.
There’s more detail to this, but that’ll do for now.
(So, I took the bait… over to you)
@Victor said: “Frankly, I am open to hearing what recommendations others here might have.” and “Where should we search next?” and “So, let’s re-visit a question that’s fundamental to guiding the new search: What assumptions (if any) were incorrect in the DSTG’s Bayesian analysis that led to a posterior distribution that most of us now believe is incorrect, or at least has been modified by the null search results to date?”
[S25: I am removing a long portion of your comment for brevity. Contributors here typically submit longer comments and material as linked documents. It would be helpful if you did the same.]
The YBR 00:19:37 terminal point area (31.400S. 96.940E) would be my priority choice of location to conduct an ocean bottom search for MH370 along the 7th arc. In addition, search conditions in the SIO area north of the Broken Ridge may also be optimal in terms of ocean floor environment, ocean depth, and seasonal weather. The ATSB never provided the supposed locations of the “hard objects” mentioned in a press release during the bathymetry survey conducted north of the Broken Ridge Trench in the Autumn of 2014. The shallower water in that area may have allowed the survey sonar to respond to large aircraft fuselage, wing or engine pieces to give the “hard objects” indication. Therefore, the recorded hard objects may have been associated with MH370 debris rather than being stray sonar artifacts.
I wonder if the Malaysians and/or the ATSB will select some of the quality independent inputs that we have submitted to them during the past three years and provide this data/information to the Ocean Infinity team so that these offerings can be factored into their search strategy?
S25
@All,
If anyone wants to compare BTO models, here is mine:
Time 00:19:37 UTC
BTO 18,410 +/- 86 (2-sigma) microseconds
Geometric altitude 20,000 feet
Latitude 34.00000 S
Latitude 94.13138 E
I use the following constants:
BTO bias -495,679 microseconds
WGS-84 Earth dimensions
I do bi-cubic interpolation of the satellite ECEF XYZ over time from ATSB’s Table 3.
In the BFO calculation, I use 42,588 km for the nominal distance of the satellite from the center of the Earth.
@VictorI
You should know that although I’m a ‘ditcher’ I’m also very happy about the current state of affairs. I’ve put my cent in the basket just outside the new 25.000km2 search zone at ~32.4S/~97E hinting on a recovery and ‘short’ glide after a steep descent.
If the plane gets found anywhere close to the 7th arc I’ll be just as happy as anyone.
@S25: Welcome to the blog. Please submit another comment with a link to your full scenario so that others can see it.
You said: I wonder if the Malaysians and/or the ATSB will select some of the quality independent inputs that we have submitted to them during the past three years and provide this data/information to the Ocean Infinity team so that these offerings can be factored into their search strategy?
OI is aware of the comments on this blog. I am certain the strategy to continue further north along the 7th arc was influenced by the comments and analyses of contributors here. I would be surprised if 31.4S latitude was not searched if not found before then.
@DrB: Using the same BTO bias and altitude as yours, and using Henrik’s PAR5 orbit approximation, for a BTO value of 18410 μs, at 00:19:37 I calculate a position of (-34, 94.13286), which is about 138m from your position.
@Don Thompson
Thanks. I’ll try to reply on your comment as good as possible but I first have to know what you mean exactly with ‘PFCS’. This one is new to me.
So I start with this question..
@Ge Rijn: PFCs = Primary Flight Computers
@Victor,
Thanks for the BTO calculation. That’s pretty good agreement in my book. The separation of our arcs is only about 86 meters.
Anybody else want to compare?
@sk999
So I sorted out the issue in the satellite orbit using your model. I simply had an incorrect orbit radius. I don’t know where I got the incorrect radius. There was no note in the code (my bad).
So I went ahead and plotted the 7th arc using a very simple model – sub-satellite point and great circle radius. Earth radius used to calculate great circle distance was 6371 km. Negligible difference (my opinion) with your arc.
https://docs.google.com/document/d/1WV2Susje4Qvo-2ZMMdXotpEA4m-co1pSXez3a-3PPHI/edit?usp=sharing
@Victor, @Ge Rijn
PFCS is the Primary Flight Control System, comprising the rudder (yaw), the elevators & horizontal stabiliser (pitch), the ailerons and flaperons (roll), and the speedbrakes. The FBW augmentation & control functions are part of the system, the PFCs and the ACE; Primary Flight Computers and Actuator Control Electronics
@Don Thompson: Thanks. There are too many acronyms, some very similar. PFCs are known, as there is a cockpit switch for this, which must cycled to return to normal flight control mode after entering secondary mode.
@DrB, @Victor: Using the same BTO bias and altitude as yours, and using Duncan’s orbit approximation, for a BTO value of 18410 μs, at 00:19:37 I calculate a position of (-34, 94.13646), which is about 332m from Victor’s position.
@Google Earth users
Dumb question. I added a link to sk999’s kml file, and now our arcs only differ by 0.3 nm when I invoke Google Earth. Apparently Earth reads the link to files on initialization? sk999 does not appear to be on the forum today, but my suspicion is that he tweaked his kml file slightly. Is this interpretation correct?
@DennisW: I don’t know how you are measuring distance, but be careful about whether the arc is shown at altitude or clamped to the ground. When viewing at low heights, an arc in GE at altitude could appear as displaced relative to one clamped to the ground.
@Victor
Thanks. I’ll check that out.
@All: Here is an update I received from Ocean Infinity:
We planned for two trial sites when we left Durban, respectively “test site 1” and “test site 2”. Test site 1 was to validate and document our ability to detect targets and debris similar to the suggested size of the MH370 debris. We fabricated two targets identical to the last search, a cross and a box. We placed these in 1000-1200m water depth and ran lines progressively close and far from the targets at different vehicle speeds and settings. We documented all of the results proving our detection capability at various ranges and altitudes. We also found a munitions dump in the area with lots of small ~50cm cylinders which we detected as a scattered debris field on the SSS. Very good results all around. This work was completed within about 2 days.
For test site 2 we selected an area with difficult terrain and deeper water ~5800+ meters. We had a few goals for this site, including a depth trial of our two new HUGINs, which in practice are never fully depth trialed when they leave Kongsberg. Secondly we had a major repair to one of our AUVs and needed to check ballasting, calibrate the DVL and perform a patch test on the MBES. Finally, we needed to run a full cycle of all 8 AUVs simultaneously, as we previously have not operated with more than 6. Essentially we needed to put everything through the paces before we got to the MH370 site. We’ve found and corrected a few issues, which makes this well worth the time spent. This testing is due to be wrapped up late tonight or early tomorrow morning, we will then steam to the ATSB box and hit the ground running.
@Victor,
A very succinct description of what has been happening, and sounds promising.
At 2130UTC the Seabed Constructor was stilling operating in the Test Mode near 37°15’S 036°37’E.
GC distance to 35.6°S 92.8°E is about 2680NM, which will take 8 days at 14 knots
@TBill
Re: “No I use Arc 3, 4, and 5.”
Gotcha. Thanks for that clarification.
@All,
Here are some comments on Dr. Holland’s proposed method of correcting the in-flight log-ons (at 18:25 and 00:19) using ½ of the ground-based log-on frequency decay curves.
First, Dr. Holland says Thales and Inmarsat are in agreement that this is how the SDU works.
However, accepting that, an issue still remains as to how to explain the 273 Hz BFO at 18:25:34.
Using Dr. Holland’s latest method, the largest possible contribution from the in-flight OCXO transient error to that BFO is essentially 1/2 the largest observed ground-based transient error. The largest ground-based transient errors are 125 Hz from Log-on #1 and 127 Hz from Log-on #6. Using their average (126 Hz) and taking half of that, we get 63 Hz for the largest in-flight transient error correction. Taking the measured 273 Hz at 18:25:34 and subtracting the maximum 63 Hz in-flight OCXO transient error correction, we get at least 210 Hz with the transient error removed, and this is the (lowest) BFO that would have been measured with no transient error present.
My BFO model, assuming a course parallel to N571 and no speed change, gives a predicted BFO of 143 Hz at the approximate position of the aircraft at 18:25. Thus, the difference (210 – 143) = 67 Hz can only be caused by a maneuver at 18:25.
The highest BFO due to a course change occurs at due North (0 degrees true), but this only raises the predicted BFO from 143 Hz at 296 degrees to 167 Hz at 0 degrees true course. The shortfall of 167 Hz from the calibrated BFO value of 210 Hz is +43 Hz. That additional BFO increase can only come as a result of an ongoing climb with a rate of climb of 1,900 fpm.
This 1,900 fpm ROC is somewhat larger than the performance limit for 9M-MRO. According to Rolls Royce, the maximum rates of climb at FL340 are 325 fpm at 298 tonnes and 1300 fpm at 235 tonnes for a B777-200 with Trent 892 engines. Note this is for the -200 series, not the -200ER, but I don’t think the rates of climb will be very different for the two configurations. If one linearly extrapolates the maximum climb rate to W = 209 tonnes predicted weight (using my fuel/endurance model) at 18:25, you get 1,700 fpm as the maximum possible ROC. Thus, the required 1,900 fpm is somewhat greater than the estimated maximum performance climb rate of 1,700 fpm, and therefore it is likely not achievable. I will note here that the maximum-performance ROC is a rather strong function of altitude, and one might eke out 1,900 fpm at FL310 (but interestingly, the max ROC then declines as you go lower in altitude until you get below FL250, where it increases again). At FL310 the Fuel Flow is about 10% higher than at FL340 (at M0.84 +10C).
What about the BFOs at 18:27 and at 18:28? Let’s consider the 18:28 case first. We know the contribution to the BFOs then from OCXO transient error is almost zero (less than a couple of Hz at most) because of the almost complete decay over 3 minutes of time. Thus, we now have two BFOs corrected using the method proposed by Dr. Holland : 210 Hz at 18:25:34 and 143 Hz for the average of the 18:28 BFOs. In order to explain the 210 Hz, one needs a greater-than-maximum performance rate of climb at FL340. If that climb continued until 18:28, the BFOs then would also be approximately the same 210 Hz. However, the BFO at 18:28 is only 143 Hz, and that means that a climb could not have continued from 18:25-18:28. If a climb was underway at 18:25, it must have stopped before 18:28. In fact, it also must have stopped by 18:27, since the transient correction then, corrected using Dr. Holland’s proposed method, would be a reduction of ½ of 10 Hz, or 5 Hz. So, the average 18:27 BFO would be the measured 175 Hz less the 5 Hz transient correction, or 170 Hz. That is still consistent with a right SLOP, but certainly not with a significant ROC at 18:27.
In summary, the implications of using Dr. Holland’s method of transient correction are that the measured BFOs at 18:25/27/28 of 273/175/143 Hz (after correction for the channel bias differences) have transient corrections of -63/-5/0 Hz, resulting in calibrated BFOs of 210/170/143 Hz. The 18:27 and 18:28 BFOs are consistent with a 15 NM right SLOP and no ROC, but the 18:25 CBFO requires an ongoing greater-than-maximum-rate of climb at FL340 to match. That combination of events seems rather unlikely to me, and that is why I am reluctant to accept applying only ½ the ground-based OCXO error when in flight.
On the other hand, applying the full OCXO transient correction (not ½) gives transient corrections of -130/-11/-0 Hz for calibrated BFOs of 143/164/143 Hz. These are compatible with a SLOP only and no climb.
Can anyone explain how you can get a calibrated BFO of 210 Hz at 18:25:34 that is within the performance capabilities of a B777-200ER with Trent engines? The only scenario I can find that might barely work is a maximum performance climb at M0.84 at FL310. The penalty for that will be on the order of 10 minutes of reduced endurance, causing even greater difficulty for the extended loiter scenario (which, even without this additional fuel penalty, has to have the air packs turned off to match MEFE). Even if the 1,900 fpm ROC were possible to achieve, at FL310, you have to wonder why the ROC would be so high at 18:25 but then go to zero before 18:27.
Bobby Ulich,
Using your position, altitude, and bias offset, I compute a bto of 18411. My KML file used a bias that is 10 microsec lower. That might be the origin of the discrepancy with the Barry Martin spreadsheet. Nice that we are all reproducing one another’s results.
ALSM,
Regarding the anomalous BTOs, while it might seem logical to use 7812.5 microsec as the unit of correction, the data generally prefer a larger value. There are 20 anomalous BTOs during the MH371 flight, half to the IOR satellite and half to the POR satellite. I experimented by comparing the corrected BTO to the average of R1200 values in the succeeding 4 minutes. For this exercise, it was necessary to correct for the change in BTO due to the flight path. The IOR data seemed to prefer a value around 7816, while the POR data wanted 7820. The number of units of correction varies from 2 to 5. There was enough noise (about 25 microsec rms) that these values are not well determined; however, 7812.5 was disfavored. Go figure. Given that the DSTG had access to more data than we have (and was the first to comment on the 7812.5 interrupt period) I tend to favor using its value.
@dr b thanks for your thoughtful examination of the implications of Dr Holland’s latest. I agree that something here must be awry. I’m puzzled that you might even expect an open and closed loop transient to look comparable. Doesn’t the closed loop by definition “bake everything in” to its correction? In fact, would a closed loop algorythm not assume a stationary aircraft?
@DrB
Re: ”I will note here that the maximum-performance ROC is a rather strong function of altitude, and one might eke out 1,900 fpm at FL310 (but interestingly, the max ROC then declines as you go lower in altitude until you get below FL250, where it increases again).”
I’d say it’s related to the speed schedule used during the climb. The ROC typically drops off between about FL250 and FL310, while the aircraft is climbing at a constant IAS. The climb rate increases again above the crossover altitude, when the aircraft continues climbing at a constant Mach. May I ask which graph or table was used to derive the ROC figures?
@sk999
The 7812.5us may appear disfavoured, statistically, but it equates to one T-ch slot period. The AES transmit window, sync’ed with the forward channel 500ms wide P-ch frame, comprises 64 T-ch slots. It appears the SDU will erroneously initiate a R-ch burst delayed from the beginning of the transmit window by n x T-ch slots (where n is 2 to 5), strictly to protocol there should be no delay.
Thus, the difference (210 – 143) = 67 Hz can only be caused by a maneuver at 18:25.
I’d say that’s one possible explanation. You are comparing power ups on the ground, all probably with similar thermal boundary conditions, with a power up in the air, probably with different thermal conditions. Different Doppler compensation algorithms for the ground-based and inflight power ups, if true, is also problematic, even if with the 2x correction. Without data from other inflight power ups, we can only guess that the behaviors are the same.
If one linearly extrapolates the maximum climb rate to W = 209 tonnes predicted weight (using my fuel/endurance model) at 18:25, you get 1,700 fpm as the maximum possible ROC.
That’s for a sustained climb. It is possible to achieve very high RoCs for short periods of time, although the steep climb would have to be timed with the start of the log-on, which is unlikely.
Don Thompson,
I cannot argue with your reasoning. It is well grounded. I would argue in support any time. However, the BTO values that fell off the A/D converters in Perth seem to have had a mind of their own. When the data disagree with “theory” and your job is to find the plane, which way do you go?
@Victor. The main issue I was addressing was how it could be that Boeing simulations were consistent with the last BFOs and such as ourselves could find no explanation, unmanned, which did not include an engine relight; yet those simulations included no relight, as the ATSB related.
So what else could explain those results? The main alternative is a rolling moment from RAT deployment. In reality though as per recent discussions the roll which resulted from that would have ceased when the APU came on line. At that point bank increase would cease and indeed, due to aircraft basic stability, start decreasing.
Supposition is that the high descent rates would be the result of increasing bank yet with APU start bank maximum would be a minute before the aircraft was in fact in its final sharply accelerating descent. So APU start would rule out RAT induced roll as the cause.
What I aimed to point out is that the Boeing simulations would not have reflected the above constraint on the RAT being the cause of the BFOs if APU start was not included in them, which indeed it wasn’t (like the relight), even though it was what led to the 7th arc log-on. Thus in those simulations the RAT could continue to impart bank increase right to the point of the log-on, providing consistency with those BFOs. However that be so only because APU auto-start was not within the simulator’s capability.
I raised a question for ALSM some weeks ago about whether lack of simulation of engine relight and APU auto-start invalidated the Boeing results. This is a possible example of that.
The question then is this. Did the ATSB/Boeing/SSWG miss the implications of these simulation shortfalls and are the simulator results they report, of replication of the final descent in fact a result of these shortfalls?
Further, if in fact engine relight thrust was insufficient to cause the descents, the only remaining explanation for them is pilot input or trim imbalance. I believe the latter was discounted some time back and assume so here.
I know the above is a repetition but I think I might have missed getting this across earlier.
What I get at is that the ATSB’s deduction that there is an explanation of the final BFOs without pilot input could be flawed, with possible implications for search width and success probability, even though no-one would want to hear that.
As to engine relight still being a possible cause, noting the ALSM simulation, whereas I was of the view earlier that most residual fuel would be consumed in a relight or during run down when above idle, autopilot controlling the imbalance effects, @Andrew has said that residual fuel flow would not start until below idle, A/P at that point being disengaged.
My interpretation of the fuel available for a relight is the 30 lbs of the ATSB’s 3rd December 2015 report, page 8, leaving aside that in the APU fuel pipeline. Delivered at the APU DC fuel pump’s maximum 3150 lbs/hr rate that would last for 34 secs. I do not know what engines were selected for the ALSM simulation, the amount of residual fuel that simulator was programmed with, or the attitudes in which it would be unavailable. Assuming they were RR, relight at below idle revs at that high altitude was simulated.
I think if any appreciable thrust resulted that well could uncover the APU DC pump inlet, resulting in multiple engine “kicks” as that was covered then not, so little engine acceleration. In any case acceleration in 34 seconds would be short of developing much thrust based on warnings about acceleration rate and Andrew’s opinion; and it is quite possible that the relight(s) would not slow deceleration enough for the engine to shut down at 35% N3, before those 34 secs of consumption completed. Still, how much is enough to cause an appreciable pitch, yaw and roll? Is the ALSM simulation to be the guide? Andrew do you have comment on that please?
However I say again that that would not explain the Boeing results. If invalid it then becomes necessary to demonstrate that relight is an explanation, or else adjust search expectations based on there having been a pilot.
@sk999
When the data disagree with “theory” and your job is to find the plane, which way do you go?
Axiomatically you go with the data since those measurements, while possibly being corrupted, are what your equipment tells you. The same equipment (presumably) that you will use in the search.
@David
@Andrew
Re: “The main alternative is a rolling moment from RAT deployment. In reality though as per recent discussions the roll which resulted from that would have ceased when the APU came on line. At that point bank increase would cease and indeed, due to aircraft basic stability, start decreasing.”
David, the cessation of roll and the reduction of bank when the APU comes on line is not necessarily the case. Andrew can explain this far more succinctly than me but essentially once you start to bank the airplane, it will start to turn and, unpowered, it will start to pitch down. The turn will to a certain extent “feed” the bank, it will tend to exacerbate it. Once you’re past a critical bank angle, dihedral is not going to get you back to wings level.
In Victor’s June 2017 sim write up, he described the simulation of the airplane (a B777-200LR) at a similar weight and likely altitude/attitude/speed to MH370 at fuel exhaustion with a control wheel input of about 4.5 units to the left. The constant 4.5 unit input is enough to generate an average roll rate of around 3.6°/sec and a pitch-down of about 1°/sec. After about 30 seconds the simulation airplane had rolled through the vertical, was steeply (about 30°) nose down and had achieved a descent rate of something like 15,000 fpm.
We know from the AMM that the left flaperon float under RAT only requires “Two to three units of right control wheel rotation … to hold the wings level.“. In other words, it’s the equivalent of a control wheel input of about 2-3 units to the left. Simple proportionality would suggest that it should generate an average roll rate of around 1.6°-2.4°/sec and a pitch-down of about 0.4-0.7°/sec. You would therefore expect for it to take less than a minute for the airplane under RAT only to have rolled through the vertical, however, as the fine print on ads often says, “Actual results may differ.” APU start up is not going to bring the airplane back from that attitude.
Andrew, handing over.
@SK999
Well, it sure is better to make a decision based on the intent of the design and the emprical evidence of its operation.
Those ‘anomalous’ R-ch bursts might be telling us something of the error/variance implicit in the start of the AES transmit window. The T-ch bursts might help too, as the protocol describes exactly which slot the burst should be initiated.
(Flashbacks to work with DEC DR11-B interfaces…)
@Don
I am reminded of the joke about the young boy who asks his dad about the difference between theory and reality. The dad’s answer involves his mother and sister.
The punch line is “In theory we have a couple million bucks. In reality we have a couple of whores living in the house.”
I did have the opportunity to interact with Dr. Spilker (an icon in satellite communication) a lot over the course of my career. I was more of an observer than a contributor. Spilker’s position was always when the measurements and theory disagree, it is likely that the theory is wrong.
@David: I suspect the Boeing simulations do not show descent rates and timings of a 4,500 fpm descent 2 minutes after fuel exhaustion and a 15,000 fpm descent occurring 8 seconds later. Again, whatever the cause, with that progressive descent, the plane should be found close to the arc.
The purpose of the Boeing simulations was to bound the distance from the 7th arc of an impact. The recommendation was to consider a radius of 15 NM. The effects we are discussing (RAT-powered flaperon asymmetry, engine relight) would make the flight less stable, and reduced the radius. Do we believe that with a 0.65g descent, the plane would have flown another 15 NM? I don’t think so. If we believe the 0.65g descent from the BFO values, than those values are more relevant than the Boeing simulations in bounding the radius.
@Mick Gilbert, @David: The dihedral effect is not very strong in the B777, except under extreme bank angles and speeds. (With a fly-by-wire aircraft, the characteristics can be changed in software.) I base this on my observing the roll rates in the Level D simulations that Mike witnessed, as well as the many FSX PMDG777 simulations I performed. Even under secondary flight control mode, a small amount of rudder out-of-trim, for instance, easily overwhelms the dihedral effect. APU start up will reduce the roll rate due to the flaperon asymmetry, but will not level the wings.
@Andrew,
The maximum performance rates of climb are given in Figure 3 in a paper by Rolls Royce called “Derated Climb Performance in Large Civil Aircraft”. I posted it online at :
https://drive.google.com/file/d/1jMMlweMcaz4zqr_OjDOHPpCol22oIsuS/view?usp=sharing
@DrB
I am very skeptical of the description provided by Holland. It is unique to a competitor’s (Rockwell) implementation. I call BS on it.
@Paul Smithson,
You said: “@dr b thanks for your thoughtful examination of the implications of Dr Holland’s latest. I agree that something here must be awry. I’m puzzled that you might even expect an open and closed loop transient to look comparable. Doesn’t the closed loop by definition “bake everything in” to its correction? In fact, would a closed loop algorythm not assume a stationary aircraft?”
Good questions. If I were building this, I would check the Weight on Wheels (WOW) switch and simply set the calculated Doppler frequency compensation term to zero. That should get you quite close. Now, when I wanted to check for long-term drift, which is always done on the ground, you can either monitor the received frequency with the AES and use that to adjust the reference oscillator, or you could have the ground station figure out what correction is needed from the measured BFO and then simply send that correction value up to the AES.
The scheme described by Holland is overkill, unless I missed something, and most manufacturers don’t overdo.
The AES has “features” that complicates things. Apparently, when the reference oscillator increases in frequency, the transmitted frequency also increases, but the received frequency decreases (i.e., the received signal must be in the upper sideband of a heterodyne receiver), and the transmitted and received frequency shifts are not equal. This scalar difference in multiplying the reference oscillator frequency standard for transmission and reception must also be accounted for. Thus the loop is closed to control the received frequency, but the corrections applied to the reference frequency (from the OCXO) are in the wrong direction to fix the transmitted frequency transient error, because of the sign difference. Thus, in the process of removing the received frequency error, the transmitted frequency (BFO) error is not removed, it is doubled. The shape of the transient error curve is maintained. It will still be an exponential decay, but with twice the amplitude. I really don’t understand why this process would be used. It is not necessary since the occasional drift correction procedures keep the frequency errors within acceptable bounds. I don’t see any useful benefit. Why reduce the received frequency transient error during warm-up if the cost of doing that is doubling the transmitted transient error? It seems like a useless exercise to me. Maybe there is a subtlety I am missing.
@sk999,
Thanks for reporting your BTO result. Yes, it’s good that several of the BTO models agree within a high degree of precision. That’s what I wanted to check.
@Mick. Thanks for that. I agree there would be secondary effects of RAT/flaperon induced bank and the aircraft would turn and pitch down. I assumed that the bank would start decreasing once the RAT influence was halted but the effect of that most probably will be less than I have implied, as you and Victor say.
@Victor. I agree that the search radius is defined by those descents, so long as there is no pilot recovery. I also agree that is unlikely even if piloted.
You say, “I suspect the Boeing simulations do not show descent rates and timings of a 4,500 fpm descent 2 minutes after fuel exhaustion and a 15,000 fpm descent occurring 8 seconds later”. There is a statement at the ATSB’s 2nd November, 2016 report, page 8, 2nd dot point, “Simulations that experienced a descent rate consistent with the ranges and timing from the BFO analysis generally impacted the water within 15 NM of the arc.” This tends to convert your concern into one of probabilities that the aircraft necessarily will be within the 15 NM. The First Principles Review report of a month later makes clear that these were ‘recent” simulations, there having been the 2016 ten investigating “additional scenarios”. If say 3 were adjudged compliant and they all fell within 15 NM, what are the chances that on doing 10 more some would fall outside that even though compliant? Perhaps this is covered in the current 25 NM half width.
Possibly pedantic, I do not see inclusion of APU auto-start and relights as necessarily reducing terminus radii in simulations. The latter in particular might increase the radius if the left engine relit in a left turn.
Thanks for indulging my digression with this.
@Ge Rijn. “a brief restart of the left engine would at least relieve some of the windmilling drag pushing the plane further to wings level or to a right bank/roll again. Certainly if the left engine produced some thrust also depending how long the thrust continued”.
This still is in play I think. If as per the ALSM simulation a relight had a noticeable effect, though at likely engine acceleration rates from below idle rpm this to me would be unexpected, the outcome of the Boeing simulations (compliance with final BFOs), which did not include relights, would then be in question.
@DennisW wrote “Spilker’s position was always when the measurements and theory disagree, it is likely that the theory is wrong.”
Quite right, I agree with you when the effort is cutting new ground.
However, the case of the SDU function isn’t new ground or proving ‘theory’, it’s tested, certified, system operation.
@David @Don Thompson
Coming back on my ‘thought experiment’ on a sequence after second engine flame-out and the timing of the final BFO’s, I see also based on your comments a lot more is possible and therefore it will be very hard to find out what the sequence realy has been.
IMO for now the more important conclusion for now could be a sequence with pilot-input cannot be excluded. Which would not affect the impact-radius of 15Nm. if the 0.65g descent continued (as per @VictorI).
Unless there was a recovery from the 0.65g descent and a following glide.
@DrB: Please don’t take my comment as discouraging. The anomalous start-up transient has me concerned that some of our assumptions about the BFO are incorrect. The work you and others are doing to sort this out is important.
@SK999,
Perhaps another way to look at the R-ch LOA burst timing when it exhibits delay by n slots…
It’s likely (n x 7812.5µs) + t
The variable t being a consequence of the P-ch sync event running through the AES clock regime, but t is implicit in any AES initiated burst.
Consider T-ch bursts. A T-ch burst is simply a long R-ch burst: same preamble structure, just more SUs as its payload. When the GES assigns the AES a T-ch, it also defines the slot in which the burst should be initiated. There’s a good range of assigned slots exhibited in the Log for the T-ch bursts.
(I’ll look in further detail later, commenting now if others wish to consider)
@Don
However, the case of the SDU function isn’t new ground or proving ‘theory’, it’s tested, certified, system operation.
I agree. The numbers are so close to 7812.5us that the correction has to be dominated by that effect.
@Victor
There is nothing in the ATSB reports, the DSTG Book, or the Inmarsat JON article that even hints at what Holland is saying. Even the BFO calibration at KL would be corrupted by the adjustments suggested.
@DennisW: The closed-loop compensation only occurs at SATCOM power-up when the navigation data is not available, which essentially would be when the electrical system to the aircraft is first brought up. I don’t think this would affect the BFO calibration when logged in as Class 3 and using the HGA, which would necessitate the availability of navigation data.
@Victor
What about the logins at 18:xx and 00:19:xx? I can rationalize the case you are talking about. That is not what Holland is talking about.
@DrB, @ALSM: Having read Ian Holland’s paper a bit more closely, I think this is what he is saying regarding closed-loop compensation:
1) During frequency calibration, the difference between the expected and received P-channel frequencies (where the expected frequency may be corrected for the modeled satellite orbit, which in our case is no correction) is assumed to be due to long-term drift oscillator drift, and is zeroed. The calibration only occurs after reaching thermal stability.
2) After calibration and using closed-loop Doppler compensation, any difference between the expected and received frequencies is assumed to be due to AES-satellite motion, and the Doppler correction is applied on that basis.
3) After calibration and using closed-loop Doppler compensation, if the difference between expected and received P-channel frequencies is due to oscillator drift rather than Doppler shift, the correction will by applied with the wrong sign, and the BFO deviation due to drift is made worse by a factor of 2.
Essentially, the closed-loop Doppler compensation algorithm is meant to handle Doppler shift, and not oscillator drift, which is assumed to be small after calibration. As a result, errors due to oscillator drift are made worse.
@DennisW: The log-ons at 18:25 and 00:19 had navigation data available to the SDU and therefore used the open-loop Doppler compensation scheme.
Gentlemen,
just a quick question since you’re the experts. You may have heard that a top-secret U.S. gov’t(?) satellite (code name: ZUMA) was launched aboard a SpaceX rocket on Jan. 8. There is a lot of conjecture about what happened to this payload, but some unconfirmed reports claim that the payload may have gone into the southern Indian Ocean after having completed just 1 1/2 orbits. The Ocean Infinity team was in the area at the same time. Just a coincidence, right?
@Don Devane: Yes, just a coincidence, and not much of a coincidence considering the size of the Indian Ocean.
@victor
@DennisW: The log-ons at 18:25 and 00:19 had navigation data available to the SDU and therefore used the open-loop Doppler compensation scheme.
But that is not what Holland is saying.
@DennisW: What have you read to indicate that closed-loop Doppler compensation was used at 18:25 (log-on 7) and 00:19? He specifically says that log-ons 1-6 use closed-loop compensation and at 18:25 and 00:19, the compensation was open-loop.
@victor
Traveling right now. Later.
@All;
Seabed Constructor still appears to be operating in Test Mode.
Its location at 2011UTC was – 37°30’47″S 036°51’37″E and stationary.
Transit time to 35.6°S 92.8°E is 8 days at 14 knots.
@Victor
He specifically says that log-ons 1-6 use closed-loop compensation and at 18:25 and 00:19, the compensation was open-loop.
Yes he does say that about 1-6. What he actually said was (page 7):
For log-ons 1 to 6, we work on the premise that closed-loop
Doppler compensation was used. In that approach, the frequency
of broadcast P-channel transmissions received by 9MMRO
were used to determine the AES Doppler contribution. In this approach all the difference between the received and expected P-channel frequency is attributed by the SDU to be caused by AES motion induced Doppler.
Bolding is mine.
How can you have AES motion induced Doppler with the aircraft sitting at the gate? If the aircraft were moving for 1-6 using the correction scheme stated by Holland that would be a violation of the Rockwell patents.
If closed loop compensation were used by 9M-MRO at the gate prior to the accident flight why is there 150Hz of BFO bias? It should have been removed. The OCXO is not electricaly steered to remove bias. The bias is removed by the numerically controlled oscillator (NCO) driven by the OCXO. It is an instantaneous value adjustment.
Just adding to Don Devane’s comment – I would expect the US Gov’t to try to make use of OI’s resources quickly to recover ZUMA before some other government finds it. One report says it crashed into the SIO after failing to separate from the rocket’s second stage.
More on the anomalous BTOs.
During the MH371 flight, there were 20 anomalous BTOs that also had enough subsequent normal R1200 bursts that one can establish a reliable mean of the latter (11-34 measurements) and thus provide a awy to determine the error in the corrected anomalous value. Further analysis shows that 7820 is materially superior to 7812.5 for the unit of BTO offset. The number of units needed to correct the anomalous BTOs varies from 2 to 5. The rms of the corrected BTOs is 25 microsec. There is no obvious correlation of the error with number of units.
I also looked at the errors in the initial R600 BTO after applying the -4600 microsec correction. The range is large (-190 to +150 microsec). However, in 6 cases the difference between the corrected R600 and anomalous R1200 BTOs was 25 microsec or less. In all cases, the difference between the average of these two values and the mean of the subsequent BTOs is small – mean offset of -6 microsec and rms of 18 microsec. It’s a small sample but might be indicative.
At the 7th arc, the corrected R600 BTO is 18400, and the corrected anomalous R1200 BTO is 18380. The difference is 20 microsec. This suggests that the average of the two (18390) might perform better than one might expect as a predictor of the true BTO at that time and thus provide a better measure of the location of the 7th arc than expected.
On a separate note, I revisted the value of the BTO bias and find a best value (based on measurements at gate C1) of 459678, which is closer to what others here are using than my older value.
On yet another note I looked a bit more at instances in the MH371 BTOs where there were large offsets in a series of values from the predicted values (60 microsec, e.g.). It seems that these values correlate with times when the aircraft was doing a lot of maneuvering, along with step climbs. The ACARS position reports are only tabulated every 5 minutes, and it is easy to imagine that the aircraft could have been maneuvering in a way that would not be accurately represented by simple interpolation.
@sk999
459678 or -495678 ?
@DrB
Re: ‘The maximum performance rates of climb are given in Figure 3 in a paper by Rolls Royce called “Derated Climb Performance in Large Civil Aircraft”.’
Thanks for the link. As you can see, the graph is based on a 250/310/M0.84 climb schedule. The crossover between 310 KCAS and M0.84 occurs at just over 31,500 ft. There’s a bit of a ‘zoom’ climb effect to begin with as the CAS starts to decrease at constant Mach, but that is quickly overcome by the reducing thrust as the altitude increases. I suspect the graph isn’t particularly accurate at showing the altitude at which the change in ROC occurs.
@DennisW: Let me start by saying that I do think something is amiss. Although I can understand why Ian Holland believes the frequency drift is amplified by a factor of 2, I am not persuaded that Ian’s description is 100% accurate. The fact that the transient behavior of log-ons 1-6, after correction, is not close to matching log-on 7 (at 18:25) indicates to me that something is not right.
You asked, If closed loop compensation were used by 9M-MRO at the gate prior to the accident flight why is there 150Hz of BFO bias?
First, I doubt that during the AES calibration, the GES ever tells the AES what the BFO value is for an AES transmission, or the BFO would be zero, as you say. So, the calibration routine uses only the shift in the received P-channel frequency, and the 150 Hz must represent the residual shift outside of the AES oscillator drift, including C-band Doppler residual and satellite oscillator shifts. It would be helpful if we had a description of how this works, including whether the GES pre-compensates for the Doppler shift for the C-band uplink of the P-channel signal, and if it does pre-compensate, how that compensation is determined.
A lot of words, but in a nutshell, we need more information to understand if Ian Holland has things correct. At a minimum, we need a very precise understanding of the conditions under which the AES uses closed-loop Doppler compensation. It might not be as simple as whether or not there is navigation data available.
Victor: Re: ” It would be helpful if we had a description of how this works, including whether the GES pre-compensates for the Doppler shift for the C-band uplink of the P-channel signal, and if it does pre-compensate, how that compensation is determined.”
The Miteq EAFCs can be configured for an inbound downconverter or an outbound upconverter. Thus, at Perth, there is one used to pre compensate the uplink C band carriers, and one to compensate for the inbound downlink 4 GHz Doppler.
@David
Re: “Still, how much is enough to cause an appreciable pitch, yaw and roll? Is the ALSM simulation to be the guide? Andrew do you have comment on that please?”
Good question; I can’t put a figure on it. If the N1 did increase by the amount shown in ALSM’s video, then I’d say it would definitely have caused the aircraft to yaw and roll. The question is whether or not that thrust increase is realistic. Given the amount of residual fuel that was supposedly available, and the fact that some of that fuel would have gone to the APU, I doubt the engine would have relit and accelerated to a high RPM before it flamed out again. However, it obviously depends on how much fuel was available to the engine, of which we’re not certain.
Still, I don’t think it matters very much. Hopefully all will be revealed if the FDR is ever recovered and we’ll be able to put this argument to bed!
@ALSM: Then that pre-compensation is wrong, which will creep into the calibration of the AES.
Victor: Yes, exactly. I have pointed this out many time in the past. To the degree that the EAFC latitude input sign error causes the precompensation to be in error, then the outbound downlink (and P channel carrier frequency) will be off. When using the Rochwell closed loop compensation for the inbound uplink, the outboutbond error will be carried over.
@Victor,
thanks for the response, and yes, the Indian Ocean is indeed a vast expanse. Btw, thanks for this excellent site, very much enjoy the great work by you and the regulars here! (Just for the record: I have full confidence in the ATSB findings, and I think that the whole thing was pilot-induced).
Anyway, with respect to my earlier post, what got my (slight) attention, was that the general area of Seabed Constructor’s loitering the past week, intersects with the posited eastern-end of the re-entry footprint for the SpaceX upper stage on Jan. 8 (to which Zuma may, or may not, have been attached).
@Lauren H.,
you wouldn’t send some outside party to look for highly classified assets. What struck me though, is that this newly-founded Ocean Infinity company (2017), with no track record of any deep-sea work, only 5 months later had the most formidable assets ever assembled for a deep-sea search, sitting in the port of Durban. All this without assured financing.
To be clear, I think it’s a bunch of coincidences, but given how bizarre the MH370 odyssey has been, why not some more ‘strangeness’…
Cheers!
@Andrew. Thanks. Putting aside the ALSM video for the moment a thrust vs flight idle vs altitude graph would be handy.
I pursue this only because, conceivably, it could have a bearing on search width.
OI Update: Seabed Constructor was delayed due to some technical issues and weather. The last two AUVs will be recovered soon and then the ship will depart around 0400 UTC.
@ALSM
@David
Gents, with regards to the recovered inboard section of the right outboard flap, can I ask how you would properly describe the point at which (the structure along which) it has separated from the outboard section?
OI Update: New departure time for the 7th arc is 0600 UTC.
@Andrew. Below flight idle, both left engine and APU drawing on 30 lbs fuel max, 3150 lbs/min, 34 secs max, relight or not. At that point the APU will continue with line fuel, engine will be shut down; that is before APU has started and well before final BFO descents.
Hence there is no chance of relight in descent proper.
So;
• the prospects of a relight affecting search width other than in pre log-on dynamics are zero;
• the only influence a relight can have is on right yaw/roll leading to BFO descents and;
• the importance then is only to pre log-on AND because Boeing did not allow that possibility.
The extreme case is that the engine relights immediately at idle and accelerates for 34 secs, or if limited by fuel flow, ‘accelerates’ (or not) at 3020 lb/hr consumption (supposing no suction supplement and APU is drawing).
I think the main question therefore might be, what is the thrust at 3020 lb/hr at FL300? (ALSM’s possibly limited by that if supplied by the APU pump. A like right engine relight would suggest not.)
A second question would be what is the drag at the windmilling right engine?
Quantitative answers to these will be unavailable I would say.
Aircraft drag would be simple to work out from glide L/D but of little use I think.
A tricky simulation I would think, particularly the 3020 lb/hr supply from idle for 34 secs seeking sign of appreciable yaw/roll, leaving the APU out of it . Maybe a home computer would do?
last line, …leaving the APU and RAT out of it…”
@Mick,
The break line of the o/bd flap aligns with the carrier structure of the flap support mechanism. A machined alloy ‘rib’ forms the flap torsion box, also at this line.
@Don Thompson
Thank you, Don.
Mick: The flap broke near the same place the wing broke in the Boeing B777 static load destructive test avalable online.
@ALSM
Victor: Yes, exactly. I have pointed this out many time in the past. To the degree that the EAFC latitude input sign error causes the precompensation to be in error, then the outbound downlink (and P channel carrier frequency) will be off. When using the Rochwell closed loop compensation for the inbound uplink, the outboutbond error will be carried over.
So what is the bottom line? My assumption is that Table 4 of the ATSB documentation includes the error you refer to. My BFO values agree virtually perfectly with the JON paper and the ATSB examples. Am I missing something here?
Frankly, your posts frustrate the shit out of me. We could never work together that is a for sure.
Dennis: The bottom line is that BFO values observed at Perth, transmitted from SDUs using open loop Doppler compensation, will not be affected by the EAFC C band uplink errors caused by the (north) latitude sign input error. If an SDU uses the open loop method, the only EAFC related error is on the inbound link. That error has been effectively calibrated out using the pilot carrier observations, as reported by Inmarsat and ATSB. So that is not an issue for the open loop case. OTOH, all the Rockwell (and any other closed loop Doppler compensation SDUs) will be affected by an error in the C band uplink Doppler compensation AND the C band downlink compensation. Removing the uplink C band Doppler compensation error would be much more difficult since there is no pilot carrier to use in that case.
Fortunately, there is growing doubt Holland’s assumption that logons 1-6 used closed loop compensation. It does not fit the transient analysis. There was no reference given for this NEW assumption. No one else has ever suggested that this is how the MSC6000 works. It is nowhere described in the manual or in any ATSB or Inmarsat document that come to the surface in this whole investigation. In any event, he concludes that the logons at 1825 and 0019 occurred using open loop compensation, therefore not a concern for the logon values during the flight.
It is only a concern if in fact the MSC6000 does use closed loop compensation (contaminated by both C band uplink AND downlink EAFC errors), AND that type of compensation occured at any time during the MH370 flight. If that happened, then it’s a whole new ball game. Fortunately, there is no evidence that is the case. Still, Holland needs to come forward and justify his assumption for closed loop compensation on the ground. Where did this come from? When was it revealed to Holland and by whom? It is critical know the truth about this.
@ALSM
It is only a concern if in fact the MSC6000 does use closed loop compensation (contaminated by both C band uplink AND downlink EAFC errors), AND that type of compensation occured at any time during the MH370 flight. If that happened, then it’s a whole new ball game. Fortunately, there is no evidence that is the case. Still, Holland needs to come forward and justify his assumption for closed loop compensation on the ground. Where did this come from? When was it revealed to Holland and by whom? It is critical know the truth about this.
Exactly. I have no idea where Holland is coming from in his latest paper. It is truly bizarre, IMO. It raises a lot more questions than it answers, and that is not a good thing.
@ALSM
Thanks Mike, I’ve watched that wing loading test video but had never made that connection.
@ALSM. About, “The flap broke near the same place the wing broke in the Boeing B777 static load destructive test available online.”
Expanding on that, a sequence (1/2 page).
https://www.dropbox.com/s/bboo0i3zxxhr6n3/wing%20break%20flap%20break%20sequence.docx?dl=0
@David @ALSM
The load testing video shows maximum loading till breaking in an upward direction. IMO this will only occure during a pull-out form a high speed dive-descent like ChinaAir 006.
In this 747 incident the wings structure was bend permenentally upwards by ~7 inch. Probably the wings endured bending of several meters during the pull-out.
IMO this will not happen during only a high speed descent/dive till the surface of the ocean without a pull-out. The video shows the utter extreme case of loading which even ChinaAir 006 did not reached (although it must have been close).
IMO when considering this scenario you have to presumme very high vertical loads on the wings during a steep dive descent which would not occure IMO without a extreme pull-out.
Far less complicated and more logical is the seperation of the flaperon and outboard flap section on impact with the ocean surface.
Remember during the ditch of US1549 also its inboard flap section broke off. And not only on its right wing inboard flap section but also on its left wing section:
https://www.flickr.com/photos/lorcanotway/3206842758
@Mick Gilbert
@David
Sorry for the late reply to yesterday’s comments.
@Mick Gilbert commented: ”…essentially once you start to bank the airplane, it will start to turn and, unpowered, it will start to pitch down. The turn will to a certain extent “feed” the bank, it will tend to exacerbate it. Once you’re past a critical bank angle, dihedral is not going to get you back to wings level.”
Most aircraft have a small degree of spiral instability and when trimmed for straight-and-level flight will eventually develop a tightening spiral-dive when no control inputs are applied. When a wing drops due to a disturbance, the aircraft tends to sideslip in the direction of roll. The sideslip, together with dihedral and/or sweepback, causes the lift on the down-going wing to increase, generating a rolling moment that tends to correct the disturbance.
However, the sideslip also causes the aircraft to yaw into the airflow in the direction of the dropped wing due to the side-force on the vertical stabiliser. The yaw is greater in swept-wing aircraft because the sideslip causes the leading wing to experience more drag. The yaw causes the upper wing to travel faster than the lower wing and it therefore develops more lift, generating a rolling moment in the direction of yaw. If the rolling moment caused by the yaw is greater than the correcting moment, the aircraft will enter a spiral dive.
Re: ” We know from the AMM that the left flaperon float under RAT only requires ‘Two to three units of right control wheel rotation … to hold the wings level.’ In other words, it’s the equivalent of a control wheel input of about 2-3 units to the left. Simple proportionality would suggest that it should generate an average roll rate of around 1.6°-2.4°/sec and a pitch-down of about 0.4-0.7°/sec. You would therefore expect for it to take less than a minute for the airplane under RAT only to have rolled through the vertical, however, as the fine print on ads often says, “Actual results may differ.” APU start up is not going to bring the airplane back from that attitude.”
Thinking about this some more, the hydraulic accumulators would have maintained pressure in the L, C & R hydraulic systems for a short time after the second engine failure. In addition, the L, C & R PFCs, the L1, C & R ACEs and associated PCUs would have all remained powered by various backup power sources for at least one minute. Only the L2 ACE would have lost electrical power immediately.
I don’t know how the ACE failure monitor detects hydraulic failures in the PCUs, but if hydraulic pressure was still available in the L & R systems, then perhaps the left flaperon PCUs remained in normal mode until the L & R hydraulic system pressures depleted. If that were the case, the left flaperon might have been unpowered for only a short time before the APU started and restored AC power and hydraulic pressure.
That said, I am still trying to understand why the aircraft rolled to the right in all three of the simulations witnessed by @ALSM. The roll direction might be explained by the one unit of right rudder trim that was set in two of the scenarios. However, the rudder trim was zero in the scenario where both engines failed at the same time and the aircraft still rolled to the right, even before the left engine relit.
According to the ATSB, the aircraft rolled left in all the original Boeing simulations (MH370 – Definition of Underwater Search Areas, 3 December 2015, p.13). It’s my understanding that was also the case in the additional simulations conducted by Boeing in 2016, except some that were flown in an abnormal electrical configuration.
@Ge Rijn,
You’re reverting to dangerous territory, much of that ol’ opinion there.
Yes, the wing loading must be high to initiate failure. The ‘effort’ necessary to initiate the necessary amount of wing loading isn’t so dramatic. Try some research into operation of the PFCS elevator control law.
@Andrew
@David
@Don Thompson
G’day Andrew,
Thank you for the proper description/explanation of spiral instability.
Regarding how long the hydraulic systems would have maintained pressure subsequent to the second engine flaming out, it makes sense that the left flaperon wouldn’t float immediately. However, Don made the following observations back on June 5, 2017 at 10:09 am regarding how quickly pressure would be restored once the APU lit:
“The brief APU start will enable the AC Motor Pumps in all L, R & C systems, however, the volumetric capacity of these pumps is only 6 gal/min. The jackscrew motor controlling the horizontal stabiliser is likely to be in motion during this time, it consumes more than an 8 gal/min of the available flowrate in each system.
After the APU flames out, only the RAT and centre system remains pressurised and the flaperon fairs again.”
My attempt at summarising all of that would be
1. the left flaperon will not float immediately upon the deployment of the RAT subsequent to the second engine flaming out,
2. nor will the left flaperon return to the faired position immediately upon the APU coming on line, and
3. the float will return subsequent to the APU flaming out.
Don, would the jackscrew motor controlling the horizontal stabiliser be in motion while the airplane was being powered by the RAT only?
@Don Thompson
I added at least three times ‘IMO’. My opinions are based on my ofcourse limited knowledge on ALL aspects. Instead of going down the screwing ants path I ask you to explain what you mean with the PFCS control law of the elevator.
Suggesting this can cause the loadings needed to brake off a wing.
Please explain.
What I understand the right elevator will go in fixed neutral position after second engine flame-out. The left elevator will be in the same position. So I don’t see a spiral dive being initiated by the elevators.
But again please explain. You seem to know, so this will spare me a lot of (unnecessary) research.
Then I think @Andrew made a good point IMO!! Something I considered also.
Hydraulic (central) pressure would not fade instantly but gradually after second engine flame out. He suggests one minute.
Then the left flaperon moving upwards and other control surfaces affected by the central hydraulic system would only take effect on banking/roll and yaw ~one minute after second engine flame-out.
About the time the APU kicks-in and corrects these (minimal) effects again there is very little time and arguments to explain the high descent rates during the 00:19 log-on sequence IMO! Unless it was pilot-induced IMO!
I look forward on your specialist information/opinion. I’m eager to learn you know.
Seabed Constructor is on the way to the search area.
https://www.dropbox.com/s/vdqglwyenb8w1a5/SC%2013012018%20111616.png?dl=0
At this rate, it will arrive in the search area at 36°S on the 7th Arc around 21st January at 13:00 UTC.
https://www.dropbox.com/s/apt0w6wfzuwu21w/SC%20Track%20to%20Search%20Area%2013012018.png?dl=0
@Ge Rijn
Re: “Then I think @Andrew made a good point IMO!! Something I considered also.
Hydraulic (central) pressure would not fade instantly but gradually after second engine flame out. He suggests one minute.”
I’m pretty sure that Andrew is refering to electrical, not hydraulic, power to the PFCs, ACEs and PCUs when he states “In addition, the L, C & R PFCs, the L1, C & R ACEs and associated PCUs would have all remained powered by various backup power sources for at least one minute.”
You also seem to be forgetting or ignoring the fact that the APU would not restore hydraulic pressure immediately. Sure, there would be a delay in the left flaperon floating after the second engine flamed out but there would also be a delay in it being returned to the faired position after the APU starts up.
And the rate of roll, something in the order of 2°/sec, and the associated pitch down, about 1°/sec, induced by the floating left flaperon is hardly “minimal“.
@Don Devane
You stated: “What struck me though, is that this newly-founded Ocean Infinity company (2017), with no track record of any deep-sea work, only 5 months later had the most formidable assets ever assembled for a deep-sea search, sitting in the port of Durban. All this without assured financing.”
I have investigated the financial background of Ocean Infinity, its shareholders, investors and partner companies with the data that is in the public domain. The result of my investigation is that Ocean Infinity is privately funded by highly reputable investors with deep pockets and a significant successful track record of large scale projects.
@Mick Gilbert
Re: “Regarding how long the hydraulic systems would have maintained pressure subsequent to the second engine flaming out, it makes sense that the left flaperon wouldn’t float immediately. However, Don made the following observations back on June 5, 2017 at 10:09 am regarding how quickly pressure would be restored once the APU lit…”
Thanks, I had completely forgotten that earlier discussion. Must be old age!
A couple of further observations:
1. I’m not sure why the stabiliser would be active, with no autopilot and no manual inputs from the crew. Perhaps Don has more info.
2. Even if it were active, the stabiliser is powered by the right and centre hydraulic systems, but not the left. In the absence of other demands, the left system should be able to drive the left flaperon back to neutral once AC power is restored by the APU.
@Mick Gilbert
Hydraulic or electric is for @Andrew to answer what he means, but I asumme he means hydraulic if the centre hydraulic system electical pumps stop operating the hydraulic pressure won’t fade instantly. He estimates ~one minute.
This is about the time the APU would start. I see a delay would occure in restoring the left wing flaperon postion but IMO this could be minimal if any IMO.
http://www.flight-mechanic.com/large-aircraft-hydraulic-systems-boeing-777-hydraulic-system/
@Ge Rijn
@Mick Gilbert
Re: ‘I’m pretty sure that Andrew is refering to electrical, not hydraulic, power to the PFCs, ACEs and PCUs when he states “In addition, the L, C & R PFCs, the L1, C & R ACEs and associated PCUs would have all remained powered by various backup power sources for at least one minute.”’
Yes, I was referring to electrical power.
@Andrew
Thanks. So basically it’s an instant electrical shut down with a gradual decrease of hydraulic pressure. Did I understood this well?
@Ge Rijn
Er, no. The various components of the primary flight control system are electrically powered by three independent power supply assemblies (PSA). The PSAs are normally powered by permanent magnet generators housed within each backup generator and they also have several backup power sources to cater for various levels of failure.
In the event of a dual engine failure, the L PSA will be powered by the main battery.. The C PSA will initially be powered by the main battery and then by the RAT once it starts producing power. The R PSA will be powered by its dedicated flight controls DC battery for up one minute.
In summary, all PSAs should remain powered during the one minute it takes for the APU to auto-start.
@Andrew
So there will be no hydraulic pressure lost at all during this one minute between second engine flame-out and start of the APU. Do I understand you well?
@Ge Rijn: After the second flame out, there would be a decay in hydraulic pressure, but no loss in electrical power to the L, C & R PFCs, the L1, C & R ACEs and associated PCUs. When the RAT comes up, the flight control surfaces that receive pressure by the center hydraulic system become active. When the APU comes up, the left, center, and right hydraulic systems receive pressure and all flight control surfaces become active.
From @Richard Cole on Twitter: After a further 48 hours in the deep water test area, Constructor has now moved off to the #MH370 search area, an 8 day journey.
@VictorI @Andrew
IMO this is not what @Andrew stated. If I understood him well he states hydraulic pressure does not decay in the first minute after second engine flame-out.
I’d like to hear his answer on this too.
@VictorI
To repeat @ Andrew;
“In summary, all PSAs should remain powered during the one minute it takes for the APU to auto-start.”
@Ge Rijn: PSAs are electrical, not hydraulic. Please re-read what was said.
@VicorI
I understand. Just curious about the implications. IMO @Andrew stated something that could be significant about decay of hydraulic power in the first minute after second engine flame-out.
I would like to hear his explanation.
Seabed Constructor is on the way to the search area.
At this rate, it will arrive in the search area at 36°S on the 7th Arc around 21st January at 09:00 UTC.
https://www.dropbox.com/s/apt0w6wfzuwu21w/SC%20Track%20to%20Search%20Area%2013012018.png?dl=0
@Andrew
Further to previous comment concerning horizontal stabiliser: after further review & your contribution sometime back, I don’t see that the stabiliser could have been operative during an uncontrolled EoF scenario. Apologies, if I caused any confusion.
@Ge Rijn
You are correct that, while limited hydraulic & electrical power is provided only by the RAT, the right elevator is ‘locked’ and only the left elevator is operable. The elevator provides pitch control.
@back to Andrew, and others…
However, the 777 airframe inherently exhibits relaxed static stability[*] and the elevator control law synthetically augments pitch control thus delivering certifiably acceptable longitudinal stability. In PFCS secondary mode, pitch rate sensing available to the PFCs is degraded. Among the elevator control law functions: a) force feedback to the control column; b) stability augmentation. Considering an uncontrolled scenario, a) is irrelevant; conversely, in the ‘ditcher’ controlled scenario a) is relevant as force feedback is degraded & it’s feasible that pulling back too forcefully leads to overload. Now stability augmentation, b), I can only speculate what commands the PFCs might generate to the elevator in an effort to provide pitch damping if the aircraft is at extreme attitudes. Would it damp or exacerbate the scenario? Do the pitch rate sensors mounted on the forward face of the wing box produce usable output in extreme roll scenarios, do the PFCs constrain pitch damping outputs in extreme attitudes?
An EOF, high RoD scenario, is a most likely a pre-crash condition, I am not suggesting at any level that a design solution should be necessary to mitigate for it. If the wing became over stressed, the outcome may be two debris fields. Discussion welcome.
[*] as do all FBW aircraft, that’s the objective, efficiency.
Andrew: You are correct to note that in all 3 of the Level D Sim trials we recorded with a GoPro, the plane rolled to the right. However, Paul was using a little right rudder trim in all cases because, he said, in his experience, that was a common setting for most of the UAL 777 fleet. That is what put us on to the hypothesis that the rudder trim was probably a factor in determining which way the turn would start. We noted that after the lights came back on (APU power up), the TAC/trim indicator returned not to neutral, but to the setting prior to FE. Thus, we suspect, the cruise rudder trim position can dominate if the trim is significant.
That said, we found some evidence to suggest that in the case of MH370, the turn was to the left. In the Boeing sim’s that were published, 9 of the 10 went left, and one went to the right. So the one to the right may have been a test of the rudder trim hypothesis we suggested to ATSB back in Nov 2014.
I agree with your description of the tendency for the plane to roll one way or the other, depending on the specific initial conditions, and once started, the tendency is to tighten the turn radius. As the bank angle increases, the pitch down increases resulting in a steep, tight spiral. We observed a final turn rate under 1 minute (360 degrees) in 2 sim’s.
@ALSM said: However, Paul was using a little right rudder trim in all cases because, he said, in his experience, that was a common setting for most of the UAL 777 fleet.
I don’t think that’s correct for all three cases. For the case in which both engines flamed out at essentially the same time, there appears to be no rudder trim. Yet, after flame out, the plane begins a slow roll to the right, which ends at a bank of 20°, coinciding with the start of the APU. The attempted re-start of the left engine then re-initiates a more rapid roll to the right.
Victor: You are correct. I forgot about that case. In fact, now that you remind me, we were already suspecting that the rudder trim was a factor after the first few trials, and we conducted the trial you referred to (zero trim and both engines FE at once) to see if the hypothesis was correct. That one also went right, but it took much longer to develop the turn, suggesting that the theory is correct. I wish we had had more time to explore that theory.
@Ge Rijn
Re: ”If I understood him well he states hydraulic pressure does not decay in the first minute after second engine flame-out.”
No, that is not what I stated. The PCUs (ie the hydraulic actuators) that move the primary flight control surfaces are powered by hydraulic pressure, but are controlled by electrical signals from the ACEs. In the dual engine failure scenario, hydraulic pressure will not drop to zero immediately upon loss of the second engine, because the hydraulic accumulators will maintain some pressure in the systems for a short time – I don’t know exactly how long. All flight control components except the L2 ACE will have electrical power available from various backup sources during the one minute it takes for the APU to auto-start.
The upshot is that the left flaperon won’t immediately float upwards upon loss of the second engine. Both of its PCUs should have hydraulic pressure available for a short time and both PCUs will remain electrically controlled by their respective ACEs. The flaperon will only begin to float upwards when hydraulic pressure in the L & R systems falls, at which point both PCUs will go to bypass mode. I don’t know the exact point at which that will occur, but it might only happen slightly before the APU comes on-line, at which time the PCUs will go back to normal mode and the flaperon will go back to the neutral position.
I hope that’s now clear!
@ALSM
Re: ”Paul was using a little right rudder trim in all cases because, he said, in his experience, that was a common setting for most of the UAL 777 fleet.”
Yes, most B777s do seem to require a small amount of right rudder trim to zero the ailerons during the cruise. I’m not sure why that occurs; the common explanation is that the aircraft are all ‘bent’, but I suspect it has something to do with the control rigging. That said, I doubt the simulator actually needed any rudder trim while both engines were operating – I’m yet to come across a ‘bent’ simulator! By introducing right rudder trim, Paul may have artificially created conditions that forced the aircraft to roll right in the two simulations where rudder trim was used. However, that still doesn’t explain why the aircraft also rolled right in the simulation that had zero trim input.
“The known behaviour of the B777, after loss of turbine driven generator power, is that a left wing down roll (and consequent descent) is initiated: that such a descent became established is substantiated by the BFO values recorded during the 00:19 Log On. The last recorded BFO indicates an excessive rate of descent. This is not anyone’s mere opinion, it’s the result of research into the characteristics of the SDU’s behaviour by DSTG, supported by senior design staff at the SDU manufacturer, and the staff of the SATCOM network operator” Re Mike Exner.
The ditching
As already stated in the previous paragraph 1.1.10., the CVR continued to record until the impact
with the water which indicates that the main battery was still working. The FDR, which is
powered by the same main battery of the CVR stopped the recording approximately one minute
before the ditching. This could indicate that this equipment was less tolerant to the decrease of
current tension due to main battery about to be exhausted.
The attempts to restart the engines may have partially exhausted the main battery. However the
CVR continued to work10.
The latest values of the significant parameters recorded by the FDR (time: 14988 seconds) were
the following:
• Altitude: 728 ft.
• Speed: 125 kts (calibrated speed).
• Pitch attitude: 4,2°. This is the angle between the longitudinal axis of the aircraft and the horizon.
• Angle of attack: 10,2°.
• Roll angle: – 2.1° (slight inclination to the left).
• Heading: 111°.
• Vertical acceleration: 0.96. (ATR 72 2005)
I’m not going to start commenting here because I have no technical background to offer. Whilst a ditching is unlikely is it completely unfeasible?
@Ge Rijn
For clarity, the last sentence in my 1:06pm reply should read:
I don’t know the exact point at which that will occur, but it might only happen shortly before the APU comes on-line. Once the APU is on-line, the PCUs will go back to normal mode and the flaperon will go back to the neutral position.
@Andrew;
“I doubt the simulator actually needed any rudder trim while both engines were operating – I’m yet to come across a ‘bent’ simulator!”
Just an aside: With 9M-MRO we don’t know the altitude or TAS, so is there a possibility that “coffin corner” was being approached with two wind-milling engines, and the re-light and spool-up of No.1 could have created a classic Vmc roll
@Andrew
I would wonder if what you are describing is actually the case. I have a number of hydraulic systems – tractor buckets, various tractor three point lift attachments (weighing many hundreds of pounds), hydraulic lift gate on my box truck, a couple of log splitters, and several hydraulic jacks. In all cases when power is removed from the hyraulics (pump off), the hydraulic actuator stays exactly where it was. Hundred pound lift attachments do not sag to the ground even over very long time periods. The hydraulic fluid holds the actuator exactly where it was until the fluid pressure is increased or decreased. This is accomplished with the actuator valve. If I turn off my tractor, I cannot raise a tractor bucket or raise lift attachments because the hydraulic pump is off, but they readily lower under the force of gravity if (and only if) I open the valve that lets fluid flow from the lift side of the actuator.
My guess is that hydraulically controlled aircraft components stay exactly where they were when power is removed.
After engine failure . . . does it turn Right or Left?
In April 2015 I wrote a paper discussing the last 15 minutes of flight, based on information that was available at that time. In that paper I developed an argument which showed that it was most unlikely that the aircraft could have banked to the right in a turn that ultimately developed into a spiral dive, because if it had done so it could not have actually reached the 7th arc.
More likely, the aircraft banked to the left, and indeed may have intercepted the 7th arc travelling in a northerly direction as the banked turn developed into a spiral dive. The argument was based simply on the geometry of the presumed aircraft track, and the position of the 7th arc, together with assumptions on the rate of deceleration [demonstrated in sim testing to be to be 19 knots/min] following the first engine flameout, and the possible time of the second engine failure.
One of the conclusions reached was that the aircraft speed at the point of the first engine flameout had to be relatively high [perhaps 450 knots] in order to reach the 7th arc at all.
Now that the prime area of interest has moved to say 30degS I decided it might be of interest to revisit the geometry to see if the original observations were still relevant.
For a 30degS intercept on the 7th arc, aircraft tracks will all be close to an azimuth of 173.2 deg, and will cross the 6th arc at an angle of about 45 deg. If the track is continued on an azimuth of 173.2 deg, the track distance to the 7th arc [at sea level] is approximately 55nM. In fact, that distance is very little different from the track distance in my original paper, and hence there would be minimal change to the calculations on which that paper was based.
Note: In the interim ATSB revised their assumptions on the possible time for the SDU power up process [originally 3 mins 40 secs]. I’m not sure that there is a consensus on a revised time. This is of some importance in that it allows one to estimate the time of the second engine failure which of course ultimately has an effect on the deceleration assumptions.
However, I think the original premise still remains true. The speed at 00:11 must be relatively high in order to reach the 7th arc at all, and there is a much greater likelihood of a turn to the left following the failure of the second engine.
@DennisW
RE: “My guess is that hydraulically controlled aircraft components stay exactly where they were when power is removed.”
Nope. In aircraft it is not always desirable for a control surface to remain in the same position following a failure. In an extreme case, imagine what would happen if you applied full aileron and then the hydraulics to one (or both) ailerons failed. The result certainly wouldn’t be pretty if the aileron(s) remained in their previously commanded position.
The PCUs in the B777 have several different modes of operation, depending on the control surface to which they are attached. The flaperon PCUs operate in two modes: normal and bypass. According to the Aircraft Maintenance Manual (AMM): “The flaperon PCU is in the bypass mode when there is an electrical or hydraulic failure with the PCU or its ACE…If both PCUs on a flaperon are in the bypass mode, the flaperon can move freely. In flight, the aerodynamic lift then causes the flaperon to move 10 degrees up from the faired position”
@Andrew
Thanks. I am not to old to learn something. 🙂
@DennisW
Dennis, with regards to your tractor analogy by my albeit limited understanding of the airplane’s systems you are describing a hydraulic actuator in the equivalent of the airplane’s actuators’ “blocked” mode.
Blocked effectively creates a hydraulic lock. The actuators on the flaperons don’t have that mode, they instead go into “bypass” mode which allows them to move freely.
@Andrew
@DennisW
Yeah, what Andrew said.
@Mick
Yeah, I am batting 1000 lately on being wrong.
@DennisW
Join the club, it’s a learning experience. Well, it’s a learning experience for most of us, some others might still be insisting that the tractor was right.
@Don Thompson
RE: ”Now stability augmentation, b), I can only speculate what commands the PFCs might generate to the elevator in an effort to provide pitch damping if the aircraft is at extreme attitudes. Would it damp or exacerbate the scenario? Do the pitch rate sensors mounted on the forward face of the wing box produce usable output in extreme roll scenarios, do the PFCs constrain pitch damping outputs in extreme attitudes?”
According to my information, only the -200LR, -200F, -300 & -300ER variants have pitch rate sensors mounted on the wing box. In normal mode, those aircraft use a blend of pitch rate data from the ADIRU and the pitch rate sensors to provide a feedback signal for pitch damping. The -200 & -200ER variants, on the other hand, only use ADIRU pitch rate data for pitch damping in normal mode. In secondary and direct modes, all variants use pitch rate data provided by pitch rate sensors in the ACEs for pitch stability augmentation.
That said, I have not come across any information about B777 pitch damping in extreme attitudes.
@Mick, I do love my tractors.
https://photos.app.goo.gl/cMUGoC2lrEdWvBhi1
https://photos.app.goo.gl/v02fMMiJ1nlFHw1u1
@DennisW
I spent a few years on a cattle property as a kid so I can understand the attraction; I learned the basics of manual transmissions on “Henry”, a cantankerous old Ford 8N. A bit surprised to see you’ve got a Kubota though, a sign of the times I guess.
That second pic puts your sometimes less than serenely tolerant demeanour after splitting wood into razor-sharp focus.
@Mick
My SO cannot tolerate forced air heating because of the particulates. We are wood stove only in both our places, and completely on solar – on grid but negative energy usage. The only downside to the boonies is horrible internet connectivity. FWIW, the trees are growing faster than we are consuming them. Not viable for commercial purposes, but more than viable for people with acreage.
@Victor
Sorry for the off-topic posts, but my sense is we are all waiting for the OI adventure to begin. I know I am.
@Barry Carlson
RE: “Just an aside: With 9M-MRO we don’t know the altitude or TAS, so is there a possibility that “coffin corner” was being approached with two wind-milling engines, and the re-light and spool-up of No.1 could have created a classic Vmc roll”
I don’t think coffin corner & Vmca are relevant in the uncontrolled scenario. If one of the engines relit and developed significant thrust, the aircraft would have yawed and rolled regardless of the altitude or speed, assuming it was uncontrolled.
@Ge Rijn. About overstress you said, “IMO this will not happen during only a high speed descent/dive till the surface of the ocean without a pull-out.”
I had thought that the 777 might tend to pitch up like the Silk Air 737. Boeing indicated in the NTSB version of the accident report that would have occurred to the 737 but for there being a pilot and him winding on full manual nose down trim. Also, the 777 might not have got to the same speed, the 737 engines being operative and the nose forced down.
Now however I think now that the analogy is ill-founded given that Boeing would surely have made known to the ATSB/SSWG had there been such a 777 propensity, in other words such as I cannot read across from the 737 to the 777.
The other circumstance where it is quite possible the aircraft could break is in a tight high speed spiral and that seems to be to be a real candidate. To me it does not follow necessarily that a wing would be the first thing to go: it could follow a failure elsewhere.
As to US 1549 and its inboard flaps, first he had flaps deployed which seems unlikely MH370 would have, based on the evidence; and second much of that damage could have been during salvage. Lifting strops were near the wing roots as I remember.
On your 6:20 AM post, “About the time the APU kicks-in and corrects these (minimal) effects again there is very little time and arguments to explain the high descent rates during the 00:19 log-on sequence IMO! Unless it was pilot-induced IMO!
Without flaperon asymmetry there remains still the possibility of a relight induced yaw and roll, depending on what impulse, ie thrust by time, it could aggregate from APU DC pump fuel supply for a maximum of 34 seconds. The 34 would of course be less if in fits and starts due to pump suction failing in fuel surges. It would be less too if air became entrained in the fuel, sucked through its tank suction supply; that is if the engine supply manifold pressure dropped below zero due to demand outstripping supply.
Besides I think the RAT asymmetry time could still be near enough 47 secs still, the APU 60 secs start less the RAT’s 15 to deploy, 2 to wind up, based on the hydraulic pressure not lasting. Alternatively it might be nil if windmilling keeps the hydraulics going. See below.
@Andrew. Thank you for your clear description of the principal influences on spiral stability.
A couple of points on other matters:
• the hydraulics need 4794 lb of fuel in L & R tanks to cool the L & R hydraulic system oil, 7303 in the right to cool the centre system. While the cooling needed on A/P without a course or altitude change or turbulence will be lessened, there will be some time when the oil is cooled insufficiently, that eventually becoming not at all.
Since low or zero fuel could be expected at other than fuel exhaustion I would not expect the consequences to be dire. Even so in the Training Manual it says that there will be a warning to the cockpit not to keep using a system if its oil temperature rises above 100 deg C since the hydraulic fluid, “can become too hot”. I have found nothing on what the eventual consequences might be if use continues. I think of the centre hydraulic system in particular, without a pilot and even with.
Amongst other things this has some ramifications for this being a perfectly planned flight by an experienced pilot as some have suggested.
• According to the TM (for the -200ER) there are accumulators for the elevator and rudder only.
The hydraulic reservoirs are pressurised by bleed air so I would expect them to offer no reserve.
Based on that the RAT induced imbalance would start at its spin up.
Even so I think a supply from windmilling might be sufficient to inhibit that, indefinitely.
Evidence for that is at p24 of the TM 29-10-00, which states, “If the engine continues to turn (windmill) with the valve in the OFF position, the EDP may be damaged.” EDP = engine driven pump. The valve shuts off supply from the reservoir.
The point made is obvious enough but its clear implication is the pump will keep pumping during windmilling. This is hardly surprising being directly driven by the engine and of the piston, ie positive displacement, type. Note that the EDPs can each up to 48 gpm @ 2850 psi, to their left and right systems. In comparison, the two AC powered hydraulic pumps, responsive to APU start, deliver just 2 X 6 gpm to the centre system. These are the primary pumps for that system. The RAT delivers 10 gpm.
I would expect the EDPs to deliver proportionately to the engine spool speed % that drives them, and that that would be a significant quantity therefore. The air driven pumps supplying the centre system on demand also can deliver 53 gpm. The APU might power them though perhaps not at high altitude, where it will not supply air for engine starting. However with the RAT operating that has no importance.
Thus it may be that the there is sufficient hydraulic power to keep all systems hydraulically functional, with just L2 electrical control missing until APU start; and that would not lead to RAT imbalance, no control system being fully de-powered thereby except some spoilers. Thus there will be no RAT induced roll at any stage, if (!) I have this right.
@Andrew. TM page reference for accumulators is 29-10-00, pages 56 & 57
On 14th January at 07:15 UTC, Seabed Constructor was at 37.1624°S 43.6103°E, 604 km into the journey from the last stop and with 4,268 km to go. The average course is 88.7454°T and the average speed 13.5 knots.
Seabed Constructor is now expected to arrive at 36°S on the 7th Arc on 21st January at around 10:00 UTC.
@Andrew. “If one of the engines relit and developed significant thrust, the aircraft would have yawed and rolled regardless of the altitude or speed, assuming it was uncontrolled.”
Thanks Andrew. I have always been in the controlled corner, be it right or wrong.
@David
RE: ’Since low or zero fuel could be expected at other than fuel exhaustion I would not expect the consequences to be dire. Even so in the Training Manual it says that there will be a warning to the cockpit not to keep using a system if its oil temperature rises above 100 deg C since the hydraulic fluid, “can become too hot”. I have found nothing on what the eventual consequences might be if use continues.’
Hydraulic fluid overheat reduces the service life of the hydraulic components (eg seals & hoses) and of the fluid itself. The fluid will also lose viscosity, which reduces its ability to transmit power and to lubricate components in the hydraulic system.
The heat exchangers cool the case drain fluid from each hydraulic pump and deliver it back to the reservoirs. Given the low system demands in the cruise and the low ambient temperature, I would not expect overheating to present too much of a problem, even though the heat exchangers would be uncovered with the reduced fuel state. It’s not uncommon to land with less than 7303 lb of fuel in each tank and I have never known it to cause an overheat issue.
RE: ”According to the TM (for the -200ER) there are accumulators for the elevator and rudder only. “
Yes, the accumulators are located in the tail and are intended to absorb pressure changes caused by the operation of the tail flight control PCUs. Nevertheless, there are no check valves in the system that would stop the accumulators maintaining some system pressure in the event of a pump failure. I don’t know how long that would last.
RE: ”The hydraulic reservoirs are pressurised by bleed air so I would expect them to offer no reserve.”
The reservoirs are pressurised to maintain pressure in the supply side and prevent cavitation of the hydraulic pumps. They cannot maintain pressure in the pressure side of the hydraulic systems, even if they were to remain pressurised.
RE: ”Based on that the RAT induced imbalance would start at its spin up.”
I don’t believe so, but it wouldn’t be the first time I’ve been wrong!
RE: ”Thus it may be that the there is sufficient hydraulic power to keep all systems hydraulically functional, with just L2 electrical control missing until APU start; and that would not lead to RAT imbalance, no control system being fully de-powered thereby except some spoilers.”
Again, I don’t believe so. The pumps would keep turning and deliver some fluid as the engines windmill, but at a much-reduced pressure. I doubt the pressure would be sufficient to prevent the ACEs eventually recognising a hydraulic failure, at which point they would de-energise the flaperon PCU bypass solenoid valves and the PCUs would revert to bypass mode.
@Barry Carlson
RE: “I have always been in the controlled corner, be it right or wrong.”
In that case, I guess it’s a possibility, if there were a large enough thrust increase for a sufficient time. However, I’d have to question why a pilot would allow such a state to develop, assuming a pilot was in control. Even if such a state did develop, it should have been relatively easy to reduce thrust and lower the attitude to increase speed.
@Andrew. Thank you. The EDP is a swash plate pump. It will deliver at a pretty constant pressure, altering its delivery by changing piston displacement while at that pressure. If the demand is low or near enough zero then it will deliver at the required pressure so long as that spool of the engine maintains a reasonable windmilling speed. It should not be slowed much by the pump torque with very low demand and I assume the demand would in fact be close to nil, unpiloted.
Yes I see what you say about the hydraulic lines to the wings and tail being connected even though the accumulators are designed for the tail and positioned there. 50 cu in though is 0.22 of a US gallon. With the left system having two and the right and centre one each I think they could be characterised more as pulse dampers than as reserves.
What we have is uncertainty as to whether there was any flaperon or relight induced roll before the log-on or after, how the United Airlines simulator was programmed when uncertainties in fuel availability etc deterred any programming of the Boeing for this, how representative the UA results are therefore for this and why it is the ATSB and Boeing describe the Boeing results as representative, even if they did somehow match BFO’s and their timing.
All we are left with are the BFOs without, to me, any independent confirmation as to why they should be taken as valid, I at least then being left with the conjecture that IF there were representative simulations they would validate the BFOs. In other words the basis the search width looks to be more an act of faith than demonstrable.
@Andrew,
“The -200 & -200ER variants, on the other hand, only use ADIRU pitch rate data for pitch damping in normal mode. In secondary and direct modes, all variants use pitch rate data provided by pitch rate sensors in the ACEs for pitch stability augmentation.”
Thank you for that clarification concerning -200/200ER ACE pitch rate sensing. That is, the pitch rate sensor associated with each ACE is contained within the LRU itself, and not externally mounted at the front spar of the wing box. Of course, the ACE units are installed in the forward fuselage: C, L1, L2 in the MEC and R in the E5 rack.
I recollect a note that ADIRU pitch rate sensing output is modified in accordance with the ADIRU’s position, in the MEC, well forward of CG. Can’t locate that now, if it rings with you, I’d appreciate a reply.
@Andrew,
Located that reference. It’s a quote from the ATSB, describing the ADIRU operation, in the Board’s investigation report of 9M-MRG’s inflight upset:
“The B777 does not have accelerometers located at the aircraft centre-of-gravity(c.g). The ADIRU, located in the electronics bay, contains accelerometers that calculate the c.g accelerations within a range of ±8g for all accelerations…”
Unfortunately, it doesn’t describe if the ACE pitch rate sensing function accounts for the sensor’s displacement from the CG. That the later models relocated the ACE sensor close to the CG is perhaps indicative of a desire to sense the acceleration in that area?
@David You stated: “All we are left with are the BFOs without, to me, any independent confirmation as to why they should be taken as valid, I at least then being left with the conjecture that IF there were representative simulations they would validate the BFOs. In other words the basis the search width looks to be more an act of faith than demonstrable.”
We have the detailed flight data from the previous flight MH371, where the BFO fits.
Simulators are designed and built for Pilot training, not for crash analysis.
The basis of the search width is a number of analyses, not an “act of faith”.
It looks like you are throwing the baby out with the bath water.
@Don Thompson
Thanks for the reference. Unfortunately I don’t know if the ACE pitch rate sensor output takes account of the sensor’s location, so I can only speculate. There is precious little information on the subject.
The ACE pitch rate sensors are located within each ACE in the MEC of all variants of the aircraft. They are only used for pitch rate sensing in secondary and direct modes. The newer variants have additional pitch rate sensors located on the wing box. The pitch rate data from those sensors is only used in normal mode, where it is blended with the ADIRU data.
I believe the requirement for wing box-mounted PRS first came to light during development of the -300. The longer fuselage of that variant flexes considerably more than the original -200/-200ER variants. Consequently, the ADIRU senses additional motion not related to changes in pitch. The wing box-mounted PRS help improve the accuracy of the pitch rate sensing. I assume they decided to fit the additional PRS to all new variants from that point on.
@Andrew, @Don Thompson: For a rigid body, I don’t think a pitch rate sensor needs to take into account the position relative to the CG. The fuselage is essentially rigid. On the other hand, if there is angular acceleration, then the local lateral acceleration for a rigid body will vary depending on the position relative to the CG, so it’s possible that the lateral acceleration outputs of the ADIRU are adjusted according to the position. The flexibility of the wings dictate that to damp wing deflections, sensors on the wing box are required.
@Victor
The following is a quote from Boeing training material for the -300/-300ER:
Pitch rate data supplies a feedback signal to the pitch command for pitch damping. The PFCs get pitch rate data from the ADIRU and the pitch rate sensors.
The ADIRU is in the main equipment center. The ADIRU senses airplane pitch, but also senses main equipment center vertical motion due to airplane structural deflection. Airplane structural vibration modes are excited during flight.
The pitch rate sensors are on the wing front spar. They sense pitch, but not the additional movement of the main equipment center.
The pitch rate data from the ADIRU and the pitch rate sensors are blended to supply the feedback signal for pitch damping.
@Andrew: Thanks. For a -300/-300ER, the pitch sensed by the ADIRU is adjusted by the pitch sensed at the wing box in attempt to observe fuselage deflection. That’s a bit different than adjusting pitch for distance from the CG, which for a rigid body, I don’t think is necessary.
But it appears that for the -200ER, the pitch output assumes the body is rigid, and no adjustment is made for structural deflection.
@David
Thanks for your reply. As I see it now (and said in a later comment) it seems to me there is too much room for speculation about what exactly happened during those few minutes after second engine flame out.
ATSB showed 9 out of 10 simulations made a left turn, @ALSM’s simulations took right turns. Both state the final BFO’s fit the timing and high descent rates. To me it seems the high descent rates were pilot-induced and the plane was in level flight before the log-on sequence started.
But it seems there is no way to prove this.
And anyway it won’t matter for the search width of +/-25Nm unless the pilot recovered á la ChinaAir006.
IMO as long as this is not cleared up definitely OI has to take the latter possibility into account. But they will find out soon enough if the plane is not found within +/-25Nm from the 7th arc.
On the US1459 damage to the outboard flaps; also this is something I/we can not prove happened during the ditch. There has been no detailed damage report for this was not necessary in this accident.
The straps were at the wing roots not at the flaps. To me the damage is obviously caused during the ditch but we’ll never reach consensus on this.
OI’s search will give definitely some answers on these questions.
Also if they don’t find the plane were it is expected now.
@Andrew, Victor
Thanks for following through on the pitch rate sensing subject. Now, considering the ADIRU has 6 gyros from which to generate pitch, yaw, & roll rates: is there anything that might warrant attentuion concerning a sensor implemented to measure only one axis (presumably, with a ‘typical’ airliner flight envelope?
@Ge Rijn.
Here is another term of dissuasion that I have seen used on social media:
Get.In.The.Sea
It’s consistent with previous advice.
Obviously, your long held conclusion is incontrovertible, no further comment required. I’m sure there’s an Amelia Earhart, Area 51, or similar, discussion where you’d fit right in.
@Brian Anderson
“One of the conclusions reached was that the aircraft speed at the point of the first engine flameout had to be relatively high [perhaps 450 knots] in order to reach the 7th arc at all.”
I would ask what assumptions are inherent in the 450 knot assumption. I am thinking the “people in charge” may have started descending and slowing down sometime earlier let’s say starting at 23:00. So is this 450 knot conclusion universal, or is it consistent with high altitude passive flight to FE?
@Don Thompson
We’ll see. Let’s leave it there. It’s not my confirmation-bias but yours you are projecting on me IMO in your God-like know it all perceptions.
OI will find out in three months time who was right or wrong in their assumptions on this issue.
If I turn out to have been wrong I have no problem at all. All I care for is the plane to be found and OI does not miss any possible opportunity/possibility.
If it gets found within +/- 25Nm of the 7th arc I would be glad as anyone.
But you would have something to explain with stupid remarks to me.
..if it does not get found within +/- 25Nm from the 7th arc. We’ll see.
Anyway neighter of us have the answer yet.
ATSB, in 3rd Dec 2015 publication, stated APU start time as 1 minute. This is corroborated by video clips available online showing 777-200 APU ground starts.
ATSB extrapolated L engine flame out as 2 minutes prior to SDU boot: 1 minute for the APU to spin up & bring generator online, & 1 minute for the SDU to boot before issuing Log On Request.
Garrett* published a paper presenting their APU, developed for the A320, that shows elapased time from APU start command to generator online as much as 2x longer for high altitude start.
This suggests 9M-MRO experienced a period of more than 60sec on RAT power alone before APU power became available.
Is it likely that developments overcame the physics related to gas turbine engine operation in the few years between the A320 programme and the B777?
[*] Garrett was subsumed into Allied-Signal, the developer of the 331-500B APU used on the 777, itself subsumed by Honeywell.
@TBill,
The assumptions to arrive at somewhere in the region of 450 knots are very straightforward. It is possible to determine from the fuel burn that the first engine flamed out shortly before the 6th arc. The deceleration from that point [at least until the second engine fails] is 19 knots/min. The straight line distance [continuing on the same azimuth] to the 7th arc is 55nM.
A little bit of maths shows that you need to start at about 450 knots in order to reach the 7th arc [continuing on that azimuth].
@Andrew
Do you have any knowledge of the endurance of the FCDC batteries? Above, you mentioned 1 minute & the AMM mentions a minute but it’s not clear if that implies FCDC battery endurance.
@Brian,
Thank you for reviewing your previous 6th-7th arc path hypothesis. That saved me some effort!
@Victor
RE: ”That’s a bit different than adjusting pitch for distance from the CG, which for a rigid body, I don’t think is necessary.”
Thanks, I get your meaning now. In the -300 and newer variants, the ADIRU pitch rate sensor (PRS) data is corrected by the wing box PRS for fuselage deflection. In the -200/-200ER, it seems the ADIRU PRS data is not corrected, and the same goes for the the ACE PRS data in all variants.
@Don Thompson
RE: ”Is it likely that developments overcame the physics related to gas turbine engine operation in the few years between the A320 programme and the B777?”
Where I work, we regularly test our APUs for start times after cold soak at high altitude as part of a health monitoring program for EDTO (previously called ‘ETOPS’). I’ve started the B777 APU at high altitude on many occasions and don’t recall it ever taking more than about a minute. The requirement was that it had to start ‘within three attempts’. In my experience, it started on the first attempt, every time.
RE: ”Do you have any knowledge of the endurance of the FCDC batteries? Above, you mentioned 1 minute & the AMM mentions a minute but it’s not clear if that implies FCDC battery endurance.”
As you noted, the AMM states the FCDC batteries supply ’the flight control components with power for one minute during PSA switching.’. The component maintenance manual (page 3) for the FCDC battery states:
The battery is capable of supporting short duration fill-in and path check loads of approximately 15 A dc as well as a parasitic demand of 1.5 mA dc (when the airplane is without electrical power) and an emergency condition loading of up to 30 A dc for up to one minute duration.“
@Richard. I am not suggesting there is no baby, the BFOs. I do think though that the Boeing simulations were characterised as blessing if not confirming them.
As to where the baby is in the bathwater, the 15 NM plus 10 half search width came not from the BFOs but those simulations. My point was that the interpretation of the simulations did not include one feature leading to the BFOs, an APU auto-start, nor a relight possibility for up to 34 secs prior to the BFOs, both of which possibly could have led to a wider trajectory and of different descent timings; and there may be yet more weaknesses in those simulations in replicating what would have happened.
That is why I think the interpretation offered looks to be an unsound basis for the 15 NM estimate.
Besides, collectively we have fallen short of finding how the simulations came to be interpreted with the confidence they have been, including timings, particularly when based on a sample of ten scattered scenarios.
It seems to me that the simulations generally are an untrustworthy yardstick. We really do not know how big the bath is, with apologies for my misuse of your characterisation.
That is not to say the search should not be conducted just as planned. But with eyes wide open. No confirmation bias please.
@David: First, I agree that the Boeing simulations do not explain the timing of a 0.65g downward acceleration 2 minutes after fuel exhaustion. So, the confidence of the 15 NM radius is related to the confidence we place on the final BFO values, independent of the Boeing simulations, which have known shortcomings, especially relative to system interactions. If we have confidence in our interpretation of the BFO values, the Boeing simulations are almost irrelevant, as the 0.65g downward acceleration at 15,000 fpm would produce a crash close the arc, barring an extraordinary recovery and glide.
Let’s assign some probabilities. If you assume that there is a 90% chance the BFO values are good, and an 90% chance of no recovery and glide, that means there is an 81% chance the plane crashed near (let’s say 10 NM from) the 7th arc. Go out to 25 NM, and you pick up the cases where the BFOs are wrong, but the radius of the uncontrolled flight is still less than 25 NM. So maybe we’re now at 86%. To get to 100%, you need to probably cover out to 80 NM from the arc to include the long, controlled glide. The marginal return is much lower.
@David,
There are some other indicators which might also constrain the search width. I tried to demonstrate that just reaching the 7th arc [the location of which is determined by BTO alone . . . ] itself requires a reasonably high groundspeed at the 6th arc [and hence a relatively high altitude too . . . ], and depending on exactly when the second engine failed, possibly an acceleration after that point, which means a nose down attitude.
An alternative to the continuation of a straight path is a Left turn after the second engine quits and the possibility of a spiral dive developing. Reaching a rate of descent of 15,000 ft/min is no problem in this situation, and it also means that sea level is reached rather quickly, and within a few miles [inside or outside] of the 7th arc.
What if the FCDC batterie(s) were either flat, damaged, or their leads were diconnected in some way ?
If so, on second engine flameout, there would be no remaining electrical power source for AIMS until the RAT or the APU came on line, correct ?
In which case, I assume, AIMS would shut down. Thus, any control deflection(s) that were balancing the single engine thrust asysmetry to maintain heading up to that point, would thus be lost. If there was a little left rudder trim to counter the left engine’s thrust, and if it remained, an immediate left yaw would develop, would it not ?
Brian’s “An alternative to the continuation of a straight path is a Left turn after the second engine quits and the possibility of a spiral dive developing.” could well be correct ?
@Brian Anderson @Victor
If it hard to get to Arc7, that indicates we should center the search at a lower altitude Arc7 definition FL150 or so max.
Victor I think you have had several arguments about why the search should bias a little to the outside of Arc7, but my intuition says inside.
Also maybe they should do close to Arc7 +-10 nm before the outer areas.
@TBill,
“ lower altitude arc 7 definition”
I’m talking about the 7th arc at sea level. Can’t get much lower than that.
@Ventus.
An FCDC battery supplies only the PFC, ACE and PCUs attached to its Power Supply Assembly (and no battery for L2 ACE & connected PCU).
AIMS is supplied by the main battery.
@ventus45
RE: “If so, on second engine flameout, there would be no remaining electrical power source for AIMS until the RAT or the APU came on line, correct ?”
The short answer: INCORRECT!
The long answer: The FCDC batteries are used as a short-term power source for the power supply assemblies (PSA) that power various components of the primary flight control system, such as the primary flight computers (PFC), actuator control electronics (ACE) and power control units (PCU). They do not power the AIMS.
The PSAs are powered primarily by the permanent magnet generators in the backup generators. They also have various sources of backup power. The FCDC batteries are only intended to provide continuity while the associated PSA is switching between power sources in the event of a failure. As a last resort, an FCDC battery can also power its associated PSA for about a minute.
In the dual engine failure scenario, only the R PSA is likely to fall back on its FCDC battery. The L & C PSAs would initially be powered by the aircraft’s main battery via the hot battery bus. All PSAs would be powered by the APU once it came on line.
@Don Thompson
Beat me to it!
The Northernmost end of the 25,000 km^2 search zone reaches into the boundary region between the Australian-Antarctic basin (where Fugro and Go Phoenix searched) and Broken Ridge (which is actually the plateau North of the boundary). The boundary region itself is sufficiently complex that even the names of features differ among different sources. Which is it – the Diamantina escarpment? Fracture Zone? Trench? Or is it the Ob’ Trench?
The following article seems to have a good review of the ATSB bathymetric data. Figure 2 probably has the best presentation of the topography.
https://eos.org/project-updates/geological-insights-from-malaysia-airlines-flight-mh370-search
This figure covers about 1.2 degrees of arc along the 7th arc. The left edge of this figure is at about -32.0 deg latitude. Make no mistake -this is ugly terrain. The vertical drop from the rim of Broken Ridge to the bottom of the Diamantina trench is over 5000 meters – three times the vertical drop of the Grand Canyon. It is also the region of peak probability for data-driven routes. Will OI be able to map this region of seafloor?
@SK999
Projecting Constructor’s recent tracks in the vicinity of S37E36 onto NOAA’s and ESRI’s arcGIS tools showed it was working around the Mozambique Escarpment at depths approaching 6000m. All 8 AUVs were deployed in this area.
Geoscience Aus has made all the bathymetry data openly available.
@SK999
Let me try again… the AUVs have operated at those depths. Whether the tech image adequately in that ‘terrain’ is another matter. The coverage won’t be 1200km per day.
Given GA has made the bathymetry available, the conditions won’t be a surprise.
A presentation given by a member of the Swire-OI team at a Perth, WA, conference was linked here, previously. That presentation shows AUVs working steep escarpments during ops back in Aug 2016 on the mid-Atlantic Ridge. The date & time on the image can be correlated to Constructor’s position at that time, and hence the bathymetry over which it was working.
Don Thompson –
Thanks – that is encouraging information.
Brian Anderson – the distance between the 6th and 7th arcs is not known precisely since the BTO’s all have errors. Thus, the minimum speed needed to cover that distance has some error as well. However, as one goes further North along the 7th arc, the typical speed needed to reach such a location drops, to the point that, even with the BTO errors, one might be able to exclude such latitudes as well. Will have to see if that places a practical limit on how far North to search.
@sk999
There is no “practical limit” on how far North to search. Are you living in a cave?
DennisW, you ask, “Are you living in a cave?”
I wish. How good are your wood stoves? Do they keep the house warm to 32F? 0F? -13F? -23F? Can they plow the driveway and dig you out from 2 feet of snow?
It’s summer in the SIO. Sound’s like a nice place to be this time of year.
@all An article of interest perhaps for some of you.
http://www.cbc.ca/news/canada/newfoundland-labrador/kraken-sonar-technology-stjohns-newfoundland-mh370-missing-indianocean-1.4481709
@sk999
No, the driveway needs the tractor. My b’day today, and I had to call 911 for Ami. She messed up her back and needed a “blanket carry” to a gurney. I could not load her into a vehicle. Not a happy day for sure. We are back home after half a day in ER, but I am still very concerned relative to mental status and breathing, although SPO2 is at 95+.
https://photos.app.goo.gl/ZRXurknqDJM3jYK63
@Andrew,
RE: “The pumps would keep turning and deliver some fluid as the engines windmill, but at a much-reduced pressure. I doubt the pressure would be sufficient to prevent the ACEs eventually recognising a hydraulic failure, at which point they would de-energise the flaperon PCU bypass solenoid valves and the PCUs would revert to bypass mode.”
At a much-reduced pressure; should be at a much reduced flow rate. Hydraulic demand (as in flow) at the point of fuel exhaustion would not be particularly high IMO.
The flaperon PCU is in the bypass mode when there is an electrical or hydraulic failure with the PCU or its ACE.
To do this:
The ACE has a failure monitor to do a check of the correct operation of the PCU. The monitor uses the ACE command signal and the feedback signals from the EHSV LVDT and the actuator LVDT. When the failure monitor finds a fault with the PCU or the ACE, it de-energizes the bypass solenoid of the PCU.
A hydraulic failure of the PCU is a discrepancy between command signal and either or both LVDT feedback signals; not a failure of the hydraulic system.
@sk999
BTW, she got a presciption for some mega-valiums which I intend to divert to me and my reactions to posts here. It will help me a lot.
@DennisW
Well, many happy returns for today Dennis and sorry to hear about Ami. Best wishes all round.
@Victor. You said, “If you assume that there is a 90% chance the BFO values are good, and an 90% chance of no recovery and glide, that means there is an 81% chance the plane crashed near….,” then, “….(let’s say 10 NM from) the 7th arc”. I think we would agree that the closer to the arc the higher the probability of a find. That said I do not know how you come up with 10 miles other than intuitively, albeit well informed.
I have a stab below at an outline of an analysis.
Here I use the Holland minimum and maximum descent rates at 7th arc log-on of 14,800 fpm and 24,100 fpm, associated with southerly tracks. Assumed is a log-on altitude of 25,000 ft, ie 4.11 NM.
Please visualise a flat cone with a 10 NM base at SL, log-on at its 4.11 NM apex. The average descent angle along that will be 22.4˚. Taking the maximum 24,100 fpm descent rate, 237 knots, if the aircraft were to travel directly down the cone in a straight line, the shortest route to the circle at the surface, the average descent TAS will be 623 knots. Obviously if the aircraft spiralled on the conical surface to that radius instead of heading direct it would cover a much greater distance in getting to the bottom, its descent TAS then being proportionately higher again. Likewise if it separated from the cone’s outer surface.
With an average 22.4˚ descent doubtless the aircraft speed would accelerate. In a spiral with the cone widening, to maintain a constant rate of descent its airspeed would need to increase anyway.
If the aircraft continued to accelerate vertically the airspeed would need to increase even more. Such acceleration would reduce the log-on airspeed while increasing that at sea level.
Leaving complications such as this and spiralling aside for the moment for simplicity, assume a straight descent, no spiral. If the aircraft starts at 25,000 ft with a 24,100 fpm ROD and 22.4˚descent and maintains that 623 knots TAS average throughout, he will start at 448.75 knots CAS, M=1.035. At sea level he will be at 623 knots CAS, M=0.9418.
These speeds obviously are unrealistic and no acceleration adjustment, trading speed at the top for that at SL, would fix that. Remember too that the speeds would have been higher in a spiral, so therefore one can deduce I think, despite all the simplicity, that the aircraft would not have reached ten miles from the 7th arc while averaging a 24,100 fpm ROD from 25,000 ft.
I include all lesser altitudes in that, where the descent angle would lessen, because to realise the 24,100 fpm ROD, airspeeds would need to increase further.
What this means is that there is a potential here to determine what are realisable radii with different RODs, arriving at hypothetical exclusions, that is, a ROD ‘A’ will limit crash terminus radius to ‘X’ for a straight line descent.
For a spiral, supposing the straight descent length were multiplied by ‘Y’, a modified radius would result.
Now following that line of thinking, let’s examine the minimum ROD above of 14,800 fpm. The descent angle will remain the same, that descent rate being 146 knots. That would lead to a descent TAS of 384 knots and CAS of 264 knots, M=0.638 at 25,000 ft; 384 knots CAS M=0.58 at sea level. Sounds OK? Not really if there is a spiral which doubles the air distance say, in which case the speeds would become unrealistic.
Indeed it may be that BFO descents will not match such derived speeds in a likely scenario at all unless the radius assumed is small. There then needs to be a tight spiral to consume the potential energy via induced drag, unrealistic speeds resulting otherwise.
The end result of an analysis developing the above therefore could be the exclusion of more distant 7th arc-to-terminus horizontal distances; along similar lines to drift analysis, which excludes some terminus areas. At the extreme, search width could be reduced, demonstrably and rationally.
The above does suppose the final BFOs are valid….
I might muck about with this though I doubt I could match others with better skills than I have retained with my 10-horsepower, 9 dead brain, and more modern than I ever got to. I hope Gysbreght for one might take a look.
@Brian Anderson. I remember the point you made but would appreciate a repost of the reference.
On 15th January at 07:16 UTC, Seabed Constructor was at 37.0426°S 50.4489°E, 1210 km into the journey from the last stop and with 3,660 km to go. The average course is 88.7436°T and the average speed 13.6 knots.
Seabed Constructor is expected to arrive at 36°S on the 7th Arc on 21st January at around 09:00 UTC.
@David: You are right, there are still a number of unknowns in determining the search width. We do not know the exact location of the 7th Arc, the altitude, lateral and vertical speed of MH370 when passing the 7th Arc, the actual wind and temperature, the exact interpretation of the BFO, the precise timing of events in the last minutes of the flight and how close the simulators are to the reality.
In searching +/- 25 NM either side of the 7th Arc at 20,000 feet, there is a chance that MH370 will be missed. I agree with Victor’s assessment of the probability, that maybe we’re now at 86%.
What probability would you estimate?
@DennisW
Hi Dennis. Sorry to hear about your troubles. Hope Ami gets better soon.
@DennisW
I wish you all the best for you and Ami, especially a speedy recovery for Ami’s back.
@DennisW: A belated Happy Birthday, and all wishes for a speedy recovery of Ami.
@David: Assuming the BFOs are valid, I don’t think you are going to do much better than the banked descent simulation for estimating the path radius, which predicted a value of 5 NM. The transonic drag might be underestimated, and the speed over-predicted, but I think the result is reasonable.
@David @VictorI
You explain better what I try to express for some time now and also @Richard adds up to this in his latest comment.
Probability might be as high as 86% within +/-25Nm of the 7th arc and then even more probably within +/-10Nm.
I suppose OI will search starting at the 7th arc inside out (after they searched the CSIRO/Godfrey hotspot zones).
OI will find out first if this assumptions are correct. I hope so.
If not all changes and areas outside the +/-25Nm zone become an option.
I don’t think a glide of ~80Nm is possible though if we accept the final BFO descent rates. Air Transat Flight 236 (Airbus 330) holds the gliding record from ~33.000ft after second engine flame out with ~65Nm.
If a high speed descent was involved with MH370 end-of-flight after second engine flame-out and a recovery was made with a glide after, it’s impossible IMO the plane glided that far. I assume ~40Nm from the 7th arc would be a maximum if a recovery was made at about ~20.000ft.
Then again also a long glide would not have been an objective if the objective was to ditch the plane. If it has been the objective to ditch the plane it would rather be done in the shortest possible glide.
Like a para-glider would do.
@DennisW
Best wishes.
one thing – “working together”
https://www.youtube.com/watch?v=0oyWZjdXxlw
@DennisW, all well to you and family, take a break from all this; I was quite happy when Jeff banned me, in fact…
@All,
My latest paper on “MH370 Maneuvers Matching the 18:40 CBFOs” is available HERE .
This paper discusses the possible causes of the variations seen in the 18:40 and 23:14 BFOs. It then presents eleven solutions to the 18:40 CBFOs, two of which are found to also have the correct endurance. Both of these involve a climb to higher altitude. In one case the climb is completed before 18:40, and in the other case it is still ongoing then.
A new figure is included which displays the airspeed needed after the FMT, as a function of flight level, to match the known endurance (assuming there was not an extended loiter). I have also computed the southern latitude limits on the 7th Arc as a function of the post-FMT flight level (with MEFE at 00:17).
@All,
Sorry. There seems to be a problem with the link in my post above. Try this one instead.
My latest paper on “MH370 Maneuvers Matching the 18:40 CBFOs” is available HERE .
@Bobby,
In your latest paper (15-Jan-2018), I see in para 3.1 that you have taken into consideration our suggestion (private conversation) for a safe descent to FL270 and with a subsequent turn left underneath route N571 which is well described and detailed in our report at http://www.mh370-captio.net).
If I may, I would challenge your argumentation to “not consider this option further” for the following reasons:
1) the air was extremely stable and calme (basically 0.4 knots of wind)
2) the B777 behaves beautifully and its automatiion is very reliable.
3) a stable rate of descent (RoD)in such condititions is not a all an issue
4) in your computation of BFOs uncertainty, you mention a RoD uncertainty of around 100fpm (including uncertainties you did not include:-)
Thus it is perfectly in the B777 capabilities during the descent that we estimate around 2:30min.
Then the plausible piloted trajectory could have been followed as explained in our report.
Thanks.
@Dr B: In your paper, you said: It may be possible to achieve this extreme ROD stability using the vertical speed control mode, which applies pitch corrections during the descent, but this is not the normal descent method used, and there is no obvious reason why a pilot should want to depart from normal procedure in order to achieve a very stable descent rate in this situation.
I don’t know what occurred at 18:40, but for the record, I don’t agree with your logic for dismissing a V/S descent.
Some quick points:
1) I believe what is normal procedure for a descent for a typical commercial flight is irrelevant. Any scenario here is extraordinary. Planes don’t normally fly to fuel exhaustion over the SIO.
2) If a pilot was at FL350 and descended to FL200 using V/S=-2500 fpm, for example, there are six minutes at relatively flat BFO values produced as the V/S is held very constant and the descent angle levels out around 3 deg.
3) There is no danger of overspeed for this vertical speed and the minimum flight level of FL200. In fact, the maximum speed doesn’t exceed around 310 KIAS. There is no additional work for the pilot unless the altitude is set lower.
4) The reason for selecting a V/S descent would not be “to achieve a very stable descent rate”. Rather, it is a way to expedite a descent (without using speedbrakes). A similar result could have been achieved using FLCH and setting the speed to 310 KIAS, but it is less likely (but not impossible) this was done based on the small variation in the BFO.
@Jean-Luc Marchand: Welcome to the discussion here.
Just to get you up to speed a bit…in a previous blog post, there was a lot of discussion of descent modes and the affect on the measured BFO. I proposed at that time that a V/S descent might have occurred at 18:40 to explain the stability of the BFO values.
Our B777 pilot contributor @Andrew helpfully reminded us that a V/S descent is not typical compared to selecting FLCH or ECON modes. An important point to consider is the potential for the plane to overspeed during a V/S descent, and the extra work on the pilot to monitor speed.
However, as I said in my recent comment, for V/S=-2500 fpm and for altitudes above FL200, there is no danger of an overspeed. Of course, even if the speed approached Vmo = 330 KIAS, the envelope protection would pitch the nose up and reduce the vertical speed.
@David,
Are you looking for this one:
http://www.duncansteel.com/archives/1461
@DrB
Many thanks for your excellent paper.
What position do you have for MH370 at the start of the first call at 18:39:55 UTC please?
I am trying to fit your Case #2D and Case #3 to the satellite data between 19:41 UTC and 00:19 UTC, which I appreciate you have not included in this paper and was planned as one of your next steps.
@Richard. Thanks. Re, “In searching +/- 25 NM either side of the 7th Arc at 20,000 feet, there is a chance that MH370 will be missed. I agree with Victor’s assessment of the probability, that maybe we’re now at 86%.
What probability would you estimate?”
As per my earlier I cannot get the aircraft to 10 NM from 25,000 and below when it is in a spiral. To me it will be within that unless piloted or it chances not to enter a spiral.
That to me is a guess. Mine is that is more like 5% than 14%.
@Brian Anderson. That is it thanks. Will take a refresher.
@Victor. Re the simulator’s 5NM, that may be so though that leaves us up in the air, so to speak, for how well it replicates. I would prefer something with less uncertainties if that is possible, so will iterate towards a maximium radius for max and min final descent rates and let you know of any outcome of interest. The dynamics of drag in a tight spiral might be a problem though.
@David: I was careful in the simulation to introduce a roll rate consistent with the descent rates and the downward acceleration indicated by the BFO values. This differs from the Boeing simulations. Without a roll rate introduced (regardless of the source of the lateral asymmetry), it is very difficult to explain the downward acceleration without a nose-down pilot input. It’s not just the descent rate that is relevant. If the drag is underestimated and the speed overestimated in the simulation, as is likely, a more accurate result would reduce the radius.
From the horse’s mouth. Malaysian Minister’s statement. Para 7 adds a little and para 9 describes extensive monitoring arrangements.
http://www.mh370.gov.my/phocadownload/News/Talking%20points%20MH370%20Search%20Operation%20OI.pdf
@David: See Update on Jan 10, 2018 in the post above.
@DrB
Some interesting data analysis there, and your interpretation could be correct. But the assumption is passive or dead pilot flight until fuel exhaustion. I personally hold open pilot action after Arc5, if that’s true the analysis breaks down.
Also I have a feeling there is a close encounter with UAE343 if altitude is maintained in a turn south at IGOGU. Which in turn could explain why a descent and cross over maneuver or hold made sense to MH370.
@All: Here are the maps for the new search showing the primary, secondary, and tertiary areas for the first 25,000 sq km and the area beyond that.
@DrB
Dr Bobby’s, you’re a jewel. Re possyour 18:40 BFO paper: Case 3 is definitely worthy of further study.
I posted my results here a few days ago. Cannot find it right now (possibly it was deleted by the thought police)
At 1840, I have the aircraft beginning a gradual climb from FL350 to FL390, taking 17 minutes (FL400 final cruise works just as well, here) which is very close to 303fpm ROC. Gradual climb was performed as a range determining exercise, possibly. My bearing is 186.69, rather than 186.
Mach 0.82 at FL390 matches the BTO all the way to the 7th arc, and possesses the required range and endurance.
However, I don’t include the SLOP at 1824. I just have final radar contact timestamp closer to 1823.60 and a slowdown from M0.84 to M0.81 by 1825.
FMT starts at 1836.50 with slowdown from M0.81 to M0.72, then slow speedup and climb to M0.82 and FL390 (FL400) begins at 38.50 and is completed by 1857.
Aircraft follows GC path toward waypoint S41.00 E88.00.
Now the bad news. 7th arc is crossed at S37.84 E88.62 which has already been searched. However, pilot glides up to 97nM after MEFE and splashes down at about S39.20 E88.36 for as 16:1 ratio glide.
The final BTO at 0019 have been misinterpreted as a 0.65g acceleration descent captured by an 8 second snapshot. This will turn out to be the biggest mistake. The revised drift results have only compounded the problem.
The recovered debris in no way supports a high speed uncontrolled impact. On the contrary, it supports a carefully set up impact designed to make the plane break up and sink rapidly, with the minimum of floating debris left behind.
I admire your measured approach, but time is limited. OI are now on their way to search an area of ocean that does not contain the aircraft wreckage. How crazy is that? How could it have come to this?
I have written to OI, ATSB and the Malaysian MOT with these findings, but will anything be done? I am still very doubtful.
We will have to see how things pan out.
Rob said: I posted my results here a few days ago. Cannot find it right now (possibly it was deleted by the thought police)
More likely you didn’t bother to search very carefully. Your “few days ago” was actually 10 days ago. Here it is, which I found by doing a simple search on the page with my browser.
@Victor
Re: Search area
We need some words to describe it…correct me if I am wrong, but looks like after 25000-km2 they skip over Broken Ridge and go up to 30S region and then maybe McMurdo path region?
@TBill: If they are jumping around, that’s news to me. My understanding is that after the first 25,000 sq km, they will be search progressively north to around 30S and possible further north, as has been reported in many news outlets.
@Victor. Your, “See Update on Jan 10, 2018 in the post above.” Yes, missed it.
Re: Gaps in the sonar search coverage
Figure 73 of the October 2017 ‘The Operational Search for MH370’ gives a poor mark for sonar coverage to the section of the ‘First Principles review’ area that was covered by Fugro – 7.8% has lower confidence coverage, several times worse than other parts. I guess this because that section was finally outside the high-priority DSTG designated area, so Fugro didn’t go back to fill the gaps.
https://www.dropbox.com/s/swe85cgt0794sg6/fig_78.jpg?dl=0
This is right around the area that David Griffin has flagged as a priority. Has this gap been mentioned as an issue for this phase of the search?
On 16th January at 07:25 UTC, Seabed Constructor was at 36.9216°S 57.1972°E, 1810 km into the journey from the last stop and with 3,060 km to go. The average course is 88.7338°T and the average speed 13.5 knots.
Seabed Constructor is expected to arrive at 36°S on the 7th Arc on 21st January at around 09:45 UTC.
@DrB
I really enjoyed your latest paper. Very easy to follow as usual!
Quick questions :
Since most of your preferred paths at 18:40 seem to indicate a holding speed, do you believe that the plane was actually flying in a holding pattern at that time?
You’ve demonstrated the possibility of paths on a 296° course. Does that mean that we have to reassess the “plane went south because of 18:40 BFO” theory (ie would the northern paths be back on the table)?
@Falken,
Thank you for the link to the 777 documentary, I don’t recall ever watching that even though it was co-produced, & presumably, broadcast by Channel4 in the UK.
@Jean-Luc,
You said: “. . . I see in para 3.1 that you have taken into consideration our suggestion (private conversation) for a safe descent to FL270 and with a subsequent turn left underneath route N571 . . . .”
Not exactly. Our resident pilot expert, Andrew, has already commented that the suggestion you made in your posted paper, that this “right lateral offset + descent + left turn” maneuver was a standard procedure, was unknown to him. I was not intending to imply that that three actions were taken in a coordinated fashion. Rather, that first a right lateral offset (SLOP) was done to get off the airway, following the airlines manual. Then, at a later time, a descent might have been done to a reduced flight level as a precursor to a Holding pattern, which might have been at ANOKO for instance, which would require a left turn. One thing is clear, if a descent occurred, it was not simultaneous with the SLOP, but must have begun later, and therefore the decision to descend may be unrelated to the decision to make the SLOP.
@VictorI,
You said: “The reason for selecting a V/S descent would not be “to achieve a very stable descent rate”. Rather, it is a way to expedite a descent (without using speedbrakes). A similar result could have been achieved using FLCH and setting the speed to 310 KIAS, but it is less likely (but not impossible) this was done based on the small variation in the BFO.”
I should also emphasize that the entire variation of the BFOs at 18:40 is explainable with zero contribution from ROD, since the 23:14 BFOs show very similar dispersion (after the linear slope then is removed). This means that the contribution of the ROD variation at 18:40 is actually at most a fraction of the observed +/- 2 Hz dispersion. Thus, the upper limit on the ROD fluctuation is actually smaller than +/- 88 fpm (+/- 2 Hz), making it even less likely to be achieved in a normal descent.
@Richard,
You said: “What position do you have for MH370 at the start of the first call at 18:39:55 UTC please?”
For Case #3, I get 7.56N, 94.42E at 18:39:55, although the track then is 188 degrees, the FMT to 180 degrees not being completed until 18:40:08.
In this particular case (#3) of a 180-degree true track, I previously found the best fit to the 19:41-00:11 data passes very close to BEDAX after about a ½ hour delay relative to the most direct course. That means that the 180 degree true course in Case #3 at 18:40 cannot be the final turn. The trick here is to align the post-18:40 route to the pre-18:40 route. That may require two turns and a hold (after 18:40) with about 2 circuits. Seems awfully complicated to me. One key question is, how can you get as far west as BEDAX for the final turn? There don’t seem to be any waypoints north of BEDAX that align N-S with it closely enough (within the width of a holding pattern).
@TBill,
You said: “Some interesting data analysis there, and your interpretation could be correct. But the assumption is passive or dead pilot flight until fuel exhaustion. I personally hold open pilot action after Arc5, if that’s true the analysis breaks down.”
This paper assumes piloted flight through 18:41. There is no assumption in it of piloting after 18:41, so my conclusions in this paper do not depend on what might not might not have happened to the pilot after 18:41.
@sinux,
I’m glad you were able to follow my explanations of the sometimes complex BFO solutions.
You said: “Since most of your preferred paths at 18:40 seem to indicate a holding speed, do you believe that the plane was actually flying in a holding pattern at that time?”
Good question. It is possible a Hold began between 18:29 and 18:40, and the BFOs at 18:40 occurred during a roughly southbound leg of the racetrack pattern. Or perhaps a Hold was commanded, but the holding fix had not yet been reached by 18:40. I think it is very likely that a Hold was commanded if Holding speed were actually implemented (although it is possible to just set the current Best Holding speed without actually entering a Holding pattern into the active route). But it does not follow that racetracks must also have been flown, because the hold could have been aborted beforehand, either manually (say by entering TRK HLD on the MCP) or as a result of an unremoved offset at the holding fix. The true track and great circle routes within a few degrees of 180 seem to require a significant delay between 18:40 and the final turn south, so in this case a Hold is the obvious choice for creating the delay.
In addition, Holding speed tends to fit the 19:41-00:11 BTO data better than constant Mach at most altitudes, because it declines with aircraft weight as well as with decreasing air temperature. This is very important when fitting curved routes like CMT or CTH, and it also provides a better fit at 180 degrees true track.
You also said: “You’ve demonstrated the possibility of paths on a 296° course. Does that mean that we have to reassess the “plane went south because of 18:40 BFO” theory (ie would the northern paths be back on the table)?”
In principle the 18:40 BFOs are consistent with a NW course and a rapid descent. This only point of disagreement is that I think it is unlikely in practice the aircraft could fly with significantly less than 88 fpm of variation in the high ROD required during one minute. VictorI thinks it is possible if one uses an atypical descent mode (V/S), which he now postulates as being done because a more rapid descent than normal was desired. I wonder why that would be the case for a Hold which, according to VictorI, lasted an hour?
@Andrew,
Just to confirm my understanding, if a B777 is on a lateral offset path, and a new waypoint is entered which requires a turn of more than 90 degrees, the offset is removed automatically, correct?
If, instead of a new waypoint, one enters a holding fix, then at some point before the holding fix is reached, an error message will appear prompting the removal of the offset, correct? And if no action is taken, then a route error will occur when the “fix + offset” position is reached, correct?
@DrB
“This paper assumes piloted flight through 18:41. There is no assumption in it of piloting after 18:41, so my conclusions in this paper do not depend on what might not might not have happened to the pilot after 18:41.”
Yes right? At least the estimated bias from 23:14 sat call level flight assumption (I tend to favor a descent and slow down at 23:14). Plus any fuel savings I get from the descent at 23:14 enters into the picture, I presume.
@SK999, others
Were any of the details, below, concerning 9M-MRO’s 2014-03-07 MH371 ZBAA-WMKK service ever concluded authoratively.
Departure gate at ZBAA;
Departure runway at ZBAA;
Arrival runway at WMKK;
Arrival gate at WMKK;
Any 9M-MRO parking location after pax disembarkation & shutdown, prior to arrival at the gate for MH370 service.
Thank you.
@Andrew
Over on JW, someone posted an IR sat photo of MH370 region at 19:40. There are no readily apparent contrails pending further review by others (I have no expertise reading sat photos).
But presumably contrail visibility management might have been an MH370 strategy, depending on whether or not there was an intent to hide the track to SIO.
At night time, can a pilot see the contrails (and adjust altitude to avoid leaving trails)? Can the pilot predict when contrails form due to humidity readings on board?
After sunrise I presume there is the option to fly in the clouds that day to hide if contrails were forming.
@DrB said: “This means that the contribution of the ROD variation at 18:40 is actually at most a fraction of the observed +/- 2 Hz dispersion.”
That’s an interesting point. I do wonder if a V/S speed descent would have higher vertical speed variation than if the plane holds constant altitude. My guess is that for the same level of turbulence, the vertical speed variations would be similar.
Don Thompson,
Nothing authoritative. For departure, I used gate E10, since that is currently in the range used by MAS. Departure runway – never tried to figure it out. I suspect the takeoff runway can be guessed from the 1st position report. Arrival runway – again, did not guess. Arrival gate – not known, but currently MH361 flights arrive at the main terminal (presumably for immigration and customs processing). The plane would likely be towed somewhere for parking until needed for the return flight.
@VictorI,
You said: “I do wonder if a V/S speed descent would have higher vertical speed variation than if the plane holds constant altitude. My guess is that for the same level of turbulence, the vertical speed variations would be similar.”
Yes, that would be a very interesting comparison.
@Andrew,
Do you have any information on the typical ROD variations in V/S mode and in level flight?
Update – ZBAA departure runway looks to be 36R; WMKK arrival runway guess is 32L.
@DrB:
TBill says:
January 16, 2018 at 1:24 pm
@DrB
“This paper assumes piloted flight through 18:41. There is no assumption in it of piloting after 18:41, so my conclusions in this paper do not depend on what might not might not have happened to the pilot after 18:41.” &
“In principle the 18:40 BFOs are consistent with a NW course and a rapid descent”
So is it possible that the rapid descent & Mh370 becoming an unpiloted flight are connected? Is this a crucial period of events?
@DrB
RE: “Just to confirm my understanding, if a B777 is on a lateral offset path, and a new waypoint is entered which requires a turn of more than 90 degrees, the offset is removed automatically, correct?”
Yes, I believe so. Perhaps Victor or someone with access to an FSX/PMDG package can experiment to confirm.
RE: “If, instead of a new waypoint, one enters a holding fix, then at some point before the holding fix is reached, an error message will appear prompting the removal of the offset, correct? And if no action is taken, then a route error will occur when the “fix + offset” position is reached, correct?”
By ‘enters a holding fix’ I assume you mean enters a hold at an existing waypoint. That being the case, the offset will not be propagated beyond the waypoint where the hold was entered. An alerting message ‘END OF OFFSET’ will appear in the CDU scratchpad two minutes before the abeam point is reached, and an ‘END OF ROUTE’ message will appear when the aircraft passes the abeam point.
RE: “Do you have any information on the typical ROD variations in V/S mode and in level flight?”
Nothing authoritative, but from experience the AFDS is good at maintaining the ROD in VS mode; I’d say within ±100 ft/min. Deviations from the selected VS only occur for very short periods. In level flight, ALT mode is about the same, possibly a bit more in turbulence. Again, deviations only occur for very short periods and the selected altitude is recovered quickly.
Just a note on terminology. Strictly speaking, a 15 nm offset is a ‘contingency procedure’, designed for cases where a descent or diversion is required in oceanic airspace and no ATC clearance is immediately available. It can be flown left or right of the airway centreline, subject to other aircraft, adjacent airways and terrain. A ‘SLOP’, or strategic lateral offset procedure, is a 1 or 2 nm offset to the right of the airway centreline, and is usually flown to avoid wake turbulence from aircraft on the same airway.
@TBill
RE: “At night time, can a pilot see the contrails (and adjust altitude to avoid leaving trails)? Can the pilot predict when contrails form due to humidity readings on board?”
No, pilots can’t see their own aircraft’s contrails. It might be possible to see the contrails of another aircraft at close range in clear skies if there were sufficient moonlight. Pilots can’t predict the altitudes at which contrails will or won’t occur. There is no humidity reading!
@sk999
@Don Thompson
RE: “Update – ZBAA departure runway looks to be 36R; WMKK arrival runway guess is 32L.”
I have nothing authoritative for the flight in question, but in my experience that runway pairing sounds reasonable. Rwy 36R is the typical departure runway for international flights out of ZBAA and Rwy 32L is the typical arrival runway for WMKK, tailwind permitting. I have rarely used anything else.
On 17th January at 07:41 UTC, Seabed Constructor was at 36.7996°S 64.0061°E, 2,416 km into the journey from the last stop and with 2,453 km to go. The average course is 88.7297°T and the average speed 13.5 knots.
Seabed Constructor is expected to arrive at 36°S on the 7th Arc on 21st January at around 09:45 UTC.
@Don Thompson
Details, below, concerning 9M-MRO’s 2014-03-07 MH371 ZBAA-WMKK service as per my analysis from ACARS and Satellite Data.
Departure gate at ZBAA: Gate 22 (40.077404N 116.609116E).
Departure runway at ZBAA: 36R (40.056287N 116.601804E).
Arrival runway at WMKK: 32R (2.7615N 101.7127E) – note: not 32L as is typical.
Arrival gate at WMKK: C34 (2.74667N 101.71500E).
9M-MRO parking location: Tarmac opposite MAS Building (2.7258N 101.7145E).
Departure Gate at WMKK: C1 (2.746622N 101.712669E)
@Andrew,
Thank you for the explanation of the difference between a SLOP and a contingency procedure. That is helpful to know.
Thanks also for providing information on typical ROD variations in V/S mode and in level flight. What about FLCH descents? In that case, is 2,500 fpm typical? Are the ROD variations larger than V/S mode?
@DrB: On an idle descent, the descent angle will be between approximately 2.4 and 3.2 deg. You can do the math for what the vertical speeds are under various conditions. 3 deg is a good place to start. The vertical speed will therefore vary with true airspeed. As for variations of vertical speed, almost by definition it will be higher for FLCH than V/S, as the pitch is controlling (indicated) airspeed or Mach, not vertical speed.
@DrB
The ROD is much more variable in FLCH mode because the AFDS is attempting to maintain a constant IAS (or Mach) rather than a constant ROD. The ROD is subject to short term changes due to wind changes and turbulence; ±500 ft/min is not uncommon. The ROD also tends to decrease with altitude as the TAS decreases during descent. A ROD of around 2,500 ft/min is typical at high altitude.
@Michael John,
You said: “So it is possible that the rapid descent & MH370 becoming an unpiloted flight are connected? Is this a crucial period of events?”
If, in fact, there were no pilot inputs to the AFDS between 19:41 and fuel exhaustion, which seems probable based on the satellite data, then I would say that this may be a result of the most recent onboard events prior to 19:41. That would include the intent of the maneuvers which created the 18:40 BFOs, including a possible descent then. Whatever onboard event caused the maneuvers between 18:29 and 19:41 (which indicate a change in the flight “strategy”) may also have caused the lack of subsequent pilot actions to redirect the aircraft to make a landing. So, yes, this is a “crucial period of events” because it led to the loss of the aircraft.
So far, we have two principal theories (unintentional and intentional): (a) pilot impairment (overcome by smoke or lack of oxygen) and (b) pilot commitment to mass murder and suicide. Both theories have shortcomings in explaining the aircraft path.
The pilot impairment theory requires a cascading sequence of onboard events that increase in severity. The first event, possibly creating a widespread electrical outage, occurs circa 17:21 and causes a diversion back to a suitable alternate airfield (Penang). However, no landing is attempted, and we don’t have an explanation for not landing at Penang in the impairment theory, nor a good reason why the aircraft continued out the Strait of Malacca. The second event might have begun when the power was restored circa 18:22, and it quickly escalated to the level of pilot impairment circa 18:40.
The murder/suicide theory also has puzzling holes in it, the size of which depend on the beholder. In this theory, the diversion at 17:22 is preplanned. The pass over Penang is made to suggest the aircraft has an onboard technical problem and is returning to an alternate airfield. The subsequent trip out the Strait of Malacca is to reach the vicinity of Car Nicobar Air Force Base (VOCX), and the extended loiter near it is for conducting negotiations (for which there is no corroborating evidence). When those negotiations fail, the aircraft is purposely flown into the SIO using a path near some waypoints recovered from the pilot’s home computer.
It is possible, but not guaranteed, that the FDR, if recovered with readable data, may shed enough light on the actual path to discriminate between intentional and unintentional causes. Such a finding would be extremely important to the victims’ families and to the insurance companies (i.e., who pays for the hull loss depends on whether the cause is intentional or unintentional). If equipment items from the MEC and the flight deck are also recovered, then evidence for or against smoke and fire damage might settle the technical failure theory.
@Andrew,
@VictorI,
Thanks for the additional info on FLCH descents. In this case, clearly the BFO would vary a lot more than we see in the 18:40 phone call BFOs, so if a descent was ongoing then, it must have been in V/S mode. Since V/S is not the typical mode used, I’m still puzzled by why it would have been used at 18:40. I’m not (yet) convinced by the argument that a much more rapid descent was desired, or that V/S really is much faster than FLCH.
Bobby: I think your description of the two principal theories is a good summary. It is frustrating to not know which scenario played out. If we did know, it could possibly help narrow the search area. But we don’t know, so we need to find the boxes.
@DrB
“So far, we have two principal theories (unintentional and intentional): (a) pilot impairment (overcome by smoke or lack of oxygen) and (b) pilot commitment to mass murder and suicide. Both theories have shortcomings in explaining the aircraft path.”
I would add that each of these cases have sub Options (1) passive or incapacitated pilot for most of the flight or (2) active pilot for most of the flight.
I would call Option (b) intentional diversion, and that could be hijacking or pijacking. We do not know if the motive was suicide, extortion, asylum or what, but I definitely would consider suicide one possibility.
@DrB
Thank you for your response.
As ASLM said finding the boxes is crucial to solving the answers.
Is it fair to assume that whatever was the initial reason for Mh370 disappearing there was also a secondary event that took place & this is why there is no single viable explanation?
Whilst scant & possibly unreadable does the ISAT Data coupled with the debris found to date offer any insight into what might have happened?
& finally the French analysis of the Flaperon. Has anyone here read the full report & is there anything in there that is beneficial to the overall picture of events?
@ALSM
Glad you also now state the problem clear like this. We just don’t know (yet). If an all controlled flight, more could have happened regarding the search width.
IMO the debris finds and various forward drift analyzis constrain the possible impact zone between ~29S and ~36S.
And those constraints, constrain the Inmarsat-data/flight-paths.
IMO it’s all about the possible search-width now.
You and @DrB made the dilemma more clear again.
IMO OI can not afford to stick to the uncontrolled flight scenarios only.
Glad to see there seems to form some consensus on this point that I’ve been attacked for so often.
@Michael John
You stated: “Whilst scant & possibly unreadable does the ISAT Data coupled with the debris found to date offer any insight into what might have happened?”.
Since you are so busy with over 100 comments on another blog in the last 6 weeks, you may have missed the insight offered by the ISAT data and Debris Finds:
ISAT Data:
https://www.dropbox.com/s/atn7cnx2i27hrgb/Reconstructed%20Flight%20Paths%20-%20Richard%20Godfrey.pdf?dl=0
Debris Finds:
https://www.dropbox.com/s/clks6522vquagwb/Drift%20Model%20Results%20using%20an%20Independent%20Drift%20Model%20-%20Richard%20Godfrey.pdf?dl=0
@TBill
I definitely would consider suicide one possibility.
While not completely dismissable, suicide ranks low in likelihood IMO. The flight path (that we are relatively sure of) is simply not compatible with suicide as a primary motive for the diversion.
@DrB
Thanks for your analysis dated 2018/01/15. I see that you are gradually moving towards the intervention of human in commands :-))
Nevertheless, you paper raises an important, logical issue of reasoning.
You propose to calibrate the 18:40 BFOs using the ~23:15 BFOs. But the 23:15 BFOs correction is based on an interpolation between 22:41 and 00:11 BFOs considering a quasi constant speed and FL and no manoeuver between these points.
It means that the interpretation of 18:40 data is based on a huge assumption made a priori on the end of the flight: you justify the 18:40 calibration by taking for granted the way the trajectory will end in a quasi stable trajectory. This just proves that your computations are valid for this type of trajectory only.
If one considers a trajectory with a Contingency Procedure (CP is a standard procedure cf. operational manual : lateral offset followed by a descent) as in http://www.mh370-captio.net, then the calibration you made can’t be applicable in that case.
In fact, considering a descent, as in the scenarii 1A to 1C, means that a necessary human intervention took place as the aircraft cannot descend by herself. Thus a human presence should be assumed for some of the rest of the flight for these scenarii. Thus Table 2 cannot be used for these scenarii as it is unlikely that the trajectory would ressemble those « considered » as autonomous in your views i.e at FL270 for example. Thus your endurance estimation is far from reality with people in command. If there was a manoeuver there followed by a kind of almost constant flight afterwards, one could probably accept your conclusions. But a priori rejecting a manoeuver at that point is not justified as such. You should consider that the extra endurance (which is not matching the 6th Arc) possibly leads to a further descent to a lower level consumming more. This is logical.
Your analysis shows that the best scenarii you found are autonomous because you made the « a priori » conclusion that they must be autonomous until the end. Doing so, your conclusions on the scenarii with possible multiple manoeuvers are thus not applicable. All options with a descent manoeuver cannot be eliminated as you did not explore other types of “non constant” trajectories.
Thus I would invite you to withdraw all conclusions on these as they are not substantiated by scientific evidences. Or at least better delineate the domain of validity of your analysis.
In addition, I would suggest to add an analysis including a descent with an evolving speed which would better reflect reality.
Thanks,
Best regards,
JLuc
PS: concerning your question of the saw shape of the measured BFO, I don’t know whether you took into account the Honeywell Doppler adjustment algorithm based on the aircraft speed. The adjustment step is +/- 1 Hz which contributes to relax a little the BFOs reliability 🙂 I have made a quick sensitivity analysis: this relaxes also the required precision of Rate of Descent up to 130fpm which is even easier for the B777. Thus, I think your conclusion 6 b) needs some revisiting.
@DennisW
@Bill
@Michael John
We are of the opinion that suicide was not the aim. See our detailed analysis on a plausible piloted trajectory (www.mh370-captio.net) fully compatible with the available data.
The aircraft was probably aiming to land safely but just missed its target because of fuel shorthage.
The sophisticated way to escape from surveillance does not fit well with a suicide option.
@andrew
@VictorI
@DennisW
@Bill
@Michael John
No need to use the V/S mode. Entering FL270 and RoD -2500fpm on the MCP is sufficient and fits very well the data. The flight mode swtiched automatically to ECON DES 240kts as she would have considered this as the TOD (Top of Descent). Thus the KIAS was probably close to 240kts towards the end of the descent. It probably stayed that way until reaching 5000ft in several steps.
The steep descent during ~2:30 min occured very strategically at the exit of the Radar coverage, at the exit of Kuala Lumpur FIR and at the entrance of Chennai FIR. Doing so allowed her to quit Kuala Lumpur FIR at FL328, heading 296° at ground speed ~495kn and appear few seconds later in Chennai FIR at FL270, heading 210° at ground speed~350kn (KIAS 240kn).
This means a complete different flight for who would have monitored her.
Descending to FL270 was to pass safely underneath Routes N571 and later P 574 whose Minimum Flight Levels are FL280.
Full operational analysis available at http://www.mh370-captio.net 🙂
@Jean-Luc Marchand
IMO your arguments could be right considering pilot intentions/paths after 18:22 but not on an ending near Christmas Island.
The debris finds and drift analyzis based on them just exclude this possibility. Debris should have been found on Java/Sumatra and other northern coast like Sri Lanka and India.
The locations debris was found and not found exclude the possibility of a crash near C.I.
IMO you have to match your scenario with this debris finds and drift-scenarios.
Any scenario which is not able to produce a reasonable match between debris/drift-data and Inmarsat-data is out of the question IMO.
But ofcourse it’s up to you to make your case.
I’d like to hear your explanation on no debris finds in the regions I mentioned.
@DennisW @Jean-Luc
“While not completely dismissable, suicide ranks low in likelihood IMO. The flight path (that we are relatively sure of) is simply not compatible with suicide as a primary motive for the diversion.”
I do not want to detract from the current focus on defining search area(s), but to me ruling out suicide (1) contradicts the pilot surveys conducted after Germanwings which showed depression was a significant issue in the profession, and (2) It also contradicts the discussion of the suicide problem in the book Goodnight MH370 by Ewan Wilson, which is my adopted overall scenario for MH370. But I do not rule out other possibilities.
@Jean-Luc
I appreciate your new paper, because it is basically the first major paper assuming active pilot. This option has been largely unexplored by the more rigorous path projection experts to date.
Your path through 1941 is very similar and improves on my own proposed path (draft report in limited circulation) except I then head 180S CTH at ISBIX, with pilot actions after Arc5.
‘Suicide’ is a wide concept. Did the 911 ‘pilots’ comitt suicide? Ofcourse they did. But in their own perception they did not. They did not end their lives out of dispear but out of commitment to a cause they believed was bigger then themselves and would reward them with a better live after death.
Many suicides are not committed out of impulsive dispear or completely delusional, irrational minds but well planned and calculated.
I’ve seen many examples like this throughout my 25 years working in psychiatry.
We were often amazed and overcome by the complexity they foold us all for weeks or months everything looked fine before they committed there act.
And we were the ones trained in observing those things!
No need to use the V/S mode. Entering FL270 and RoD -2500fpm on the MCP is sufficient and fits very well the data.
Please explain how to enter V/S = -2500 in the MCP window when not in V/S mode.
Your analysis shows that the best scenarii you found are autonomous because you made the « a priori » conclusion that they must be autonomous until the end… Thus I would invite you to withdraw all conclusions on these as they are not substantiated by scientific evidences.
The scenario with automated flight after 19:41 is an assumption that may or may not be true. However, it is based on a single assumption. Your scenario ending near Christmas Island is based on MANY unsubstantiated assumptions which are required for active flight up until impact after 00:19. Yet, this doesn’t faze you in the least.
Andreas Lubitz, the co-pilot responsible for the crash of the German Wings A320 in the Alps, was treated by a large number of doctors for suicidal tendencies, but managed to keep his problem hidden from his employers, colleagues and friends. Lubitz was declared unfit for work by a doctor, but never gave the sick note to his employer.
@JLuc
First, I want to say that I really thought your paper was well done. Even after a couple of careful reads I could find nothing in the “error” category. Additionally, I thought the formatting was the best I have ever seen, and wonder what composition tool you used to create the paper.
I also had a flight path I created fairly early on, and it was not well received. I took a bit of heat from both IG people and JW (who put it in the “nonsense” category). Your flight path logic is exactly the same as mine.
JLuc
The aircraft was probably aiming to land safely but just missed its target because of fuel shorthage. The sophisticated way to escape from surveillance does not fit well with a suicide option.
I could not overcome two issues:
1> In extended discussions with Don Thompson (very reliable source of information), it became clear that the fuel remaining instrumentation on a 777 is very accurate. A pilot should not have been “surprised” by running out of fuel unless the instrumentation was somehow corrupted.
2> Like you, I really felt the PIC did not wish to harm anyone, but wanted to land safely. I cannot reconcile that with the failure to broadcast the aircraft position when fuel ran out. This would have given the PAX and crew at least a chance of rescue. But no.
A final reminder relative to what composition tool you used.
@JLuc
https://docs.google.com/document/d/1SG79753eTNK1UsUgq4oaepKRlnR5FK8V4tnbohUWjU4/edit?usp=sharing
@DrB said: I’m not (yet) convinced by the argument that a much more rapid descent was desired, or that V/S really is much faster than FLCH.
We don’t know why V/S may have been used, if it was. As to whether a V/S descent is faster than an FLCH, it all depends on the selected values of V/S vertical speed and FLCH airspeed. The maximum descent rate will be achieved with a FLCH descent at Vmo = 330 KIAS. (Using speedbrakes would increase the descent rate substantially more.) If the plane was flying at 225 KIAS at FL340, for instance, using FLCH would result in a slower descent than a V/S descent at 2500 fpm.
@DrB
Further to Andrew’s clarification regarding strategic lateral offsets and contingency procedures here is how MAS describe the Oceanic/Remote Airspace Contingency Procedures.
@DennisW
Taking into account the intent, as I understand it, of Shah to land and release the passengers regardless of the outcome of the negotiations, his use of waypoints, the constraints of the Inmarsat data requiring a final position near the 7th arc does give a number of possibilities.
One flight path relies on Shah being in control and passive passengers right through the night and into daylight with a ditching at between 9°S to 10°S near the 7th arc.
Richard,
On further review I now agree that landing at KLIA was likely on 32R. Had to refresh my memory on the runway heading alignment. Also, according to one website, MAS likes to stamp your ticket with Terminal M (for main), even if you arrive at the satellite terminal.
@Mick Gilbert,
Thanks for the flow chart on Oceanic Contingency Procedures. Two items piqued my interest. The first item was that the initial action required is a 90 degree turn (R or L). That is followed by a second action to acquire a track 15 NM away. So far I have been modeling the lateral offset as a 60 degree right turn, a delay, and then a 60 degree left turn. Do you think the 90 degrees would be strictly adhered to in practice, or would the pilot initially just command a 15 NM lateral offset and let the AFDS do the turns?
The second item of interest was the change in altitude by 500 feet(when below FL410) as a third step in the flow chart. When do you think that FLCH would be done – during or after the lateral offset?
@Michael John,
You said: “Is it fair to assume that whatever was the initial reason for Mh370 disappearing there was also a secondary event that took place & this is why there is no single viable explanation?”
Yes, I think the dichotomy between the initial turnback route from 17:21-18:03 and the 18:03-19:41 route out the Strait and then a turn south into oblivion is why we don’t yet have a rational explanation for the entire flight that is corroborated in its major points. The initial diversion back to Penang is consistent with both theories, but then neither one can fully explain the second phase. Perhaps the pilots were already hypoxic at 18:00, and nothing done after that time makes any sense to us now. The “negotiation” phase of the intentional pilot hijacking theory was invented to explain an extended loiter, but we have no corroborating evidence, or even innuendo, to back it up.
You also said: “Whilst scant & possibly unreadable does the ISAT Data coupled with the debris found to date offer any insight into what might have happened?”
The Inmarsat Satellite (ISAT) data are indeed scant, but they are definitely not “possibly unreadable”. I think you are confusing the Inmarsat data recorded on the ground during the flight, with the data resident on the Flight Data Recorder located in the aircraft. The FDR does not store “satellite” data, just the onboard flight parameters and ancillary aircraft data. It can tell us exactly how the aircraft flew to its fate, and this will offer clues on the cause and motives of the pilots. However, if the route is bizarre, the motives may be unfathomable to a rational and sane person, and a clear consensus may not be reached even then. For instance, a hypoxic pilot can do odd, illogical things.
You said: “& finally the French analysis of the Flaperon. Has anyone here read the full report & is there anything in there that is beneficial to the overall picture of events?”
To my knowledge, that report (or even significant excerpts from it) has not appeared in a public domain. It’s too bad it has not been leaked. In my opinion, the secrecy surrounding it is counterproductive to achieving a proper inquiry into the potential criminality involved in the loss of the aircraft.
@Jean-Luc,
You said: “Your analysis shows that the best scenarii you found are autonomous because you made the « a priori » conclusion that they must be autonomous until the end. Doing so, your conclusions on the scenarii with possible multiple manoeuvers are thus not applicable. All options with a descent manoeuver cannot be eliminated as you did not explore other types of “non constant” trajectories.
Thus I would invite you to withdraw all conclusions on these as they are not substantiated by scientific evidences. Or at least better delineate the domain of validity of your analysis.”
In my recent paper presenting my analysis of the 18:40 BFOs, I made no assumptions regarding the post-18:40 satellite data, nor indeed whether or not the aircraft was piloted later in the flight. The conclusions I reached regarding the 18:40 maneuvers were not based on any such assumptions. You can do anything you like that fits the satellite data after 18:40, piloted or not, but that has no impact on what was happening at 18:40, as described in my paper.
@Mick Gilbert,
On the subject of a 90 degree initial turn for the contingency procedure, if you do that and then, as soon as it is completed, immediately turn back 90 degrees to the left, the minimum lateral offset is close to 20 NM. So you can’t do a 15 NM lateral offset with two 90 degree turns because the turn rate is too slow. You can do two 60 degree turns and get 15 NM offset.
@DrB
@Andrew
Bobby, Andrew would be able to provide far better advice on those topics. However, my understanding of the initial 90° turn is that, on the basis that you are seeking to do something “unexpected” (be it in response to an emergency or for an operational reason) you want get the airplane off the airway as expeditiously as possible. Accordingly, if you were following the procedure, I would expect that the initial turn would be 90°.
It’s possibly worth noting that the contingency procedures – the 90° turn and 15 nm offset – only come into effect when two way communications with ATC cannot be established; to a certain extent it is a lost or no comms procedure.
@ Dr. B:
I have been looking at your fuel model v5.3. Thanks for that, it’s great. I saw in a reply somewhere on here you estimated 1-2% accuracy. Would that be to say total flight time estimate is estimated to be accurate within 1-2%? I am specifically looking at the Mach 0.84 and LRC cases.
Is there any logic behind a 15nM contingency procedure
offset, starting at about NILAM.
Already the evidence seems to suggest that the aircraft tracked to VAMPI and then continued on N571 for 100NM, before reaching NILAM.
Why the sudden diversion, just for another 100nM?
Were waypoints flown from IGARI to say ANDOK, and thence to VAMPI, or was the track back over the Malaysian Peninsular flown manually, based on pilot familiarity with the area?
Or, is the contingency procedure evidence of some irrationality creeping in?
Perhaps followed by a manually induced hold using the MCP . . . which was incomplete for some unknown reason.
Did anyone ever check on the possibility of crossing traffic on P627 between the Malaysian FIR boundaries at POVUS / VIROT (JAKARTA FIR), and RUSET (BANGKOK FIR), for the time interval concerned ?
The thought occured, that he may have been concerned about either meeting, or being seen by, crossing traffic, on P627, if he continued on N571 to NILAM.
The SDU was repowered and rebooted (possibly without his conscious knowledge or expectation) without Flight ID, as we all know.
Could the purpose of the repower have been to get TCAS back on line to check for conflicting traffic ?
When an ‘old” transponder is repowered, (without itself haveing actually been switched of) I presume that if the “switch” was left in the standby position when it was depowered, that it would power back up in the same state.
The question is, do modern transponders still “receive” other aircraft transmissions in the standby state, even though not transmitting itself ?
@DrB
@Mick Gilbert
The MAS procedure seems to be based on the contingency procedures stipulated in the Malaysian AIP. The Malaysian procedure is a bit different to the guidance in ICAO Doc 4444 PANS-ATM. The ICAO document stipulates a turn of ‘at least 45 degrees to the right or to the left’. It also states the following:
If the intention is to acquire a same direction offset track, the pilot should consider limiting the turn to a 45 degree heading change, in order not to overshoot the offset contingency track.
Malaysia AIP – Contingency Procedures
ICAO PANS-ATM – Contingency Procedures
I agree with @DrB; the aircraft would overshoot the 15nm offset track if the crew attempted a 90° intercept.
@GlobusMax: Welcome to the discussion.
For those interested in tracking the progress of Ocean Infinity’s search, Richard Cole will be posting updates on his site
@ventus45
re: P627- There are FR24 screen captures of UAE343 near VAMPI and to the best of my recollection, nothing is on P627. Traffic on P627 is always minimal when I look at FR24. Why wouldn’t UAE343 be the main concern? Also we do not know where the military aircraft were (eg Singapore radar plane), but possibly a pilot would know their habits.
re: TCAS- Jean-Luc’s paper postulates TCAS may have been one reason for turning on SDU, and somewhere around MEMAK is the edge of the Indonesian civil radar coverage, if operating.
Keith Ledgerwood originally raised the question- can TCAS work without Transponder squawking? but @RetiredF4 told me repeatedly that not possible, to the extent I am embarrassed to keep asking. If I recall, I might have even emailed Keith with no luck on a response.
@TBill
@ventus45
As I understand it, TCAS is a two-way street; it let’s you know about other traffic but it also let’s the other traffic know about you. If a stealthy departure was the goal I can’t see how having TCAS operating would be part of the plan.
@TBill,
Obviously TCAS can not work in the “active handshake sense” with other aircraft if it is in standby, but my question specifically concerns the state of the electronics itself. If the transponder repoweres in standby, presumably it will be communicating with AIMS etc for it’s own position information, “to be ready” to transmit it’s position to other aircraft (and the ground stations) immediately the switch is moved from satndby to mode S.
If that is correct, then it is reasonable to assume that it’s receiver is active and receiving other aircraft transmissions, and processing them internal to itself. If that is the case, if it detected a conflict, will it display it to the pilot ? In other words, could a pilot use the transponder in “stealth mode”, to “receive only”, just like our hobby ground based SSR receivers do ? It raises again, the question of precisely what were the nature of those phone discussions between Z and that MAS avionics engineer ?
@ventus45
Yes that was basically Keith Ledgerwood’s question, for the case of shadowing the SIA flight. Ledgerwood also suggested the other possibility is having a small ADS-B receiver on board.
@TBill
Indeed, now that you remind me of Keith’s question (which I had forgoten about) it essentailly boils down to the same thing. The idea of Z possibly having a small ADS-B receiver warrants critical investigation.
So, going back to Z’s simulator.
It would be interesting to know whether anyone who had “flown it” (since Z was proud of showing it off and letting other people use it) has been interviewed.
Presumably other MAS pilots “flew it” and would be able to give definitive answers to some or all of the following questions.
The first questions I would ask them, as an investigator would be:-
(1) Did Z have an ADS-B receiver and or a radio scanner attached to the simulator setup, to give “the realism” of flying amongst “other traffic” around Peneang ?
(2) Did they see, hear of (from others), or themselves experience any direct evidence of such a capability ?
(3) If so, did Z have it directly feeding the main flight simulator computer, or did he have a laptop (perhaps dedicated to the purpose) networked to the main flight simulator computer ?
(4) If (3) correct, was Z known to travel with that laptop (and possibly the ADS-B dongle with it) ?
(5) How big was it, an older bulky Dell lappy, or a very modern slimline one.
Subsequent question for the security people at WMKK on the night.
(1) Did Z “check in” with a laptop bag, and if so, was it scanned ?
(2) If so, did the scan show anything in the laptop bag that could have been an ADS-B dongle and / or antenna ?
(3) If not, was his “flight bag” scanned ?
(4) If so, did the scan show anything in the flight bag that could have been an ADS-B dongle and / or antenna ?
@Victor
The scenario with automated flight after 19:41 is an assumption that may or may not be true. However, it is based on a single assumption. Your scenario ending near Christmas Island is based on MANY unsubstantiated assumptions which are required for active flight up until impact after 00:19. Yet, this doesn’t faze you in the least.
Yet your single assumption resulted in $150M USD being tossed in the toilet. That does not seem to faze you. We argued about this extensively in the timeframe when the underwater search was initiated. You just don’t seem to be able to learn anything.
On 18th January at 08:10 UTC, Seabed Constructor was at 36.6833°S 70.6497°E, 3,008 km into the journey from the last stop and with 1,859 km to go. The average course is 88.7334°T and the average speed 13.4 knots.
Seabed Constructor is expected to arrive at 36°S on the 7th Arc on 21st January at around 11:00 UTC.
@Richard
We look forward with anticipation to Seabed Constructor’s arrival in the search zone. For me, it will be doubly special as it’s (regrettably) also my 70th birthday. I asked OI if they would push the boat out on my behalf. Their reply; Will an AUV do?
@Ge Rijn
In Section 7 of our report (and also in Annex2), we explain that we have used the CSIRO model for the flaperon and we have used the available the real meteo data taking into account the Hurricanes (among which Gillian which pushed the debris southwards)
Of course, we would like them or GEOMAR to perform computations with the same hypotheses for a cross-check.
@Victor
@DrB
1) Concerning the flying mode and the descent type, it is more that a question of vocabulary.
You discuss what descent TYPE is fastest. The V/S type at -2500fpm is the fastest and the more comfortable one as it is the maximum RoD avoiding overspeed as it allows the a/c to adjust its speed to the reference IAS computed by the flying mode. With the FLCH descent type, the a/c will descent at RoD “above” -3000fpm and it gives priority to the FL and will eventually make the a/c overspeeding. Her automation will correct the RoD to 0 until the speed is back with the fly mode envelop and then same behaviour again until the next overspeed (or the targeted FL).
Using the descent type V/S makes the FMS to switch from LRC Mode to ECON DES 240kn.
@victor
Concerning “The scenario with automated flight after 19:41 is an assumption that may or may not be true. However, it is based on a single assumption.” Yes, and I fully appreciate this and I have not difficulty. I recognise that if the flight was “automated” the posted computations make sense. But IMO the computations are valid for the automated trajectory only and do not demonstrate that a “piloted” trajectory is to be rejected. It is not because it is true for “automated” that they apply for “piloted” :-). (that’s what I wanted to say)
“Your scenario ending near Christmas Island is based on MANY unsubstantiated assumptions which are required for active flight up until impact after 00:19. Yet, this doesn’t faze you in the least.”
Oh! yes the fact that we had only Inmarsat data and try to understand what human and the a/c could have done was an issue before publishing our report. But please, I invite you to follow the 24 key events we have identified, take a piece of paper and a pencil and for each of them note any real blocking objection you may find. Then let’s discuss them together. We did not find any, that’s why we think this piloted trajectory is “plausible” 🙂
As human are considered to be in the loop, the required precision is not of the same scale than the automated. We consider that matching the Inmarsat data with a continuous flying mode and an adequate fuel consumption are the important things.
@Ventus45, others
“If the [transponder] function selector is not in the TA ONLY or TA/RA
position, the ND:
– Shows a cyan message, TCAS OFF. TCAS OFF shows in all ND modes
– Removes all TCAS symbols.”
The above applies to a pre-TCAS II 7.1 configuration. If any reader can confirm the TCAS II 7.1 behaviour, it might help resolve the TCAS question. With the transponder selector in TA ONLY or TA/RA position, it would be expected that a functional transponder was under surveillance from Indonesia and/or Thai civil ATM assets.
@Jean-Luc Marchand said: The V/S type at -2500fpm is the fastest and the more comfortable one as it is the maximum RoD avoiding overspeed as it allows the a/c to adjust its speed to the reference IAS computed by the flying mode.
No, that’s not right. The fastest descent is obtained in FLCH mode with speed set to Vmo = 330 KIAS. With speedbrakes, it’s even faster. That’s why pilots use this mode for a rapid descent under a depressurization scenario.
With the FLCH descent type, the a/c will descent at RoD “above” -3000fpm and it gives priority to the FL and will eventually make the a/c overspeeding.
No, that’s not right. Under a FLCH mode descent, pitch controls speed, and the plane will not overspeed. It is for a V/S descent that the airspeed may increase during the descent, and could reach Vmo at which time envelope protection would limit speed by pitching up and reducing the descent rate.
Using the descent type V/S makes the FMS to switch from LRC Mode to ECON DES 240kn.
No, that’s not right. Manual intervention is required to switch from a V/S descent to a VNAV mode.
But please, I invite you to follow the 24 key events we have identified, take a piece of paper and a pencil and for each of them note any real blocking objection you may find.
You are missing my point. My objection was YOUR objection to automated flight after 19:41. Your scenario may have occurred. But it requires MANY unverifiable assumptions, and requires the plane to be inexplicably flown in a way to run out of fuel and crash. There was plenty of fuel to reach Christmas Island if the plane was properly flown. Basically, you start with the assumption that the plane was trying to reach Christmas Island, and then make other assumptions to support this.
@DennisW
1) Yes the remaining fuel indication is accurate. But the estimation of fuel at destination is not so accurate especially at low altitude at which the a/c is not supposed to fly for such a long time.
2) for the SOS broadcast : jocker, I have no anwser 🙂
3) the composition tools : Word and GoogleEarth
@victor
Thanks for your reply. I would like to be more explicit: I don’t object to automated 🙂 I underlined that the piloted trajectory cannot be ruled out because computations shows that Automated is possible as concluded by Bobby.
Let’s go back to the technical exchange:
1) the word “comfortable” is important to me, at this location I don’t see why an emergency descent would take place because if there was emergency then the automated could make sense afterwards but not at FL270 but at least down to FL80. And then using the speedbrakes make sense. I tested this and its works fine. If no emergency, the natural operational way is to enter a FL and a RoD. Flying with FLCH with a Vmo=330KIAS means that you are playing with the limits, why compared to speedbrakes? On the other hand a V/S type gives the speed control to the a/c succeeding to keep it within the envelop (never above 284KIAS in this configuration).
Thanks for your paragraph on overspeeding. What you say, is excactly what I meant. Sorry for my poor mastering of English. For me “overspeeding” is when the alarm on speed is ringing and when the a/c reduces the pitch because the speed is approaching the upper limit: of course the a/c will not overspeed :-). I tested this a number of times and the overall descent was slower than the v/s type.
Your comment on manual entry to switch back to VNAV is interesting: in our view the People in Command did not press the VNAV and the a/c stayed with the descent type VS (at 0 RoD) within ECON DES 240kn mode because they did several descents in succession. Thus the FMS controlled the speed as the reference KIAS stayed at 240kn until after Arc4. Then they pressed VNAV after entering the new LRC altitude of 5000ft for optimal speed adoption. The FMS mode switched to LRC.
Xmas Island as the target: It was not an “a priori” target. Initially we thought it was the Australian Continent. But the safe respect of the Airspace Route structure, the hiding treck wrt the radars coupled with the reading of the Inmarsat data imposed on us step by step like in a compass run to a curved trajectory. The fuel estimations (using Bobby Ulich’s model) and flight simulations showed us that it was impossible to reach Australia (by ~200NM either ways).
The analysis report is necessarily presenting the study with the conclusion first 🙂
“inexplicably flown in a way to run out of fuel and crash” : Please could you be more precise and may I ask you to underline where the aspects you feel are not explained in our analysis ?
Thanks.
@Jean Luc Marchand: I am not proposing an emergency descent with speedbrakes. I am simply trying to illustrate that maximum descent rate occurs for FLCH with speed set to 330 KIAS, not for V/S descent unless the airspeed increases to Vmo during the descent. Whether or not a V/S descent is faster than an FLCH descent depends on the parameters chosen for each.
The common way to perform a long descent is NOT to use V/S. It’s either a ECON descent or FLCH descent because of the potential to overspeed. This has been discussed previously on the blog.
You describe a mode in which ECON DES is used combined with V/S. For an ECON DES mode, VNAV must be selected. You can’t be in both VNAV mode and V/S mode. However, it is possible to do a V/S descent and use the ECON DES speed as displayed in the MCP speed window. Basically, the MCP speed is “carried over” from the previous mode. However, the descent mode is V/S, not a VNAV mode.
I don’t think you can generate your path without a prior bias of crashing near Christmas Island, independent of whether you first considered other end points. You might have reasons for why Christmas Island was the goal, and perhaps you are right, but that bias is evident in your assumptions.
To run out of fuel for such a short distance would require the plane to be flown at an inefficient altitude and speed for an extended period of time. To me, this is inexplicable. If the goal was to land at Christmas Island, there was more than enough fuel to achieve this.
I think we are boring people here to death about topics that have been previously discussed. Let’s try to find a way to quickly end it.
@Victor
You are not boring us. You can discuss, and then sum it up for us in 25 words or less (just kidding about the 25 words). Re: overspeed, FL270 is not so much of a descent as FL200, so that puts a limit on it. I personally use V/S most of the time in PSS777, but of course I can’t say what a trained pilot would do. But sounds like Jean-Luc is favoring that too.
@Jean-Luc
“But IMO the computations are valid for the automated trajectory only and do not demonstrate that a “piloted” trajectory is to be rejected. It is not because it is true for “automated” that they apply for “piloted” :-)”
Yes I agree totally…I also keep trying to make this point, but you said it better than me.
Intentional paths (after 1941) have not gotten equal consideration. Thanks to Victor and Richard for being the champions of the intentional paths up until 1941.
However, if MH370 is found east of about 34 South, it was probably intentional path. I am thinking 34-36S is approx. boundary between intentional and unintentional, which is good to have a search area that has both arguments going for it. The fact that Arc6 comes up too fast in most flight modes is one give-away to the active pilot scenario.
@Jean Luc Marchand
I’ve read your study completely now and first want to complement on the clear logical lay-out and thorough presentation of facts, data and interpretations (as you and your colleagues interpreted them).
Basically I can follow you till the 2d Arc. Many of the interpretations you present happening before that point have been already proposed and researched by VictorI, DrB, Richard Godfrey, which you consequently referre to (which I appreciate in your paper), and many others here and else.
I myself have been consistently advocating a well planned and executed flight with a ditch-attempt at the end with same kind of assumptions you present but only till the second arc. After that you loose me regarding a C.I. destination.
First I don’t understand why you introduce the ‘People in command’. I see no explanation at all on this in your paper.
Your assumption is one of those ‘People in command’ was in the EE-bay before take-off switching off essential power at exactly the right moment (~IGARI) then other ‘People in command’ took over the plane.
What is your factual basis for this assumption?
And even if you have one, it’s an unnecessary complicated assumption for there was a pilot on board who had all the skills and experience needed to perform every act we see in the facts and data available.
And everything could have been done from the cockpit. Simply isolate power at the control panel above the captains seat with a few finger pushes. And restore it if needed the same way.
Then your Gillian-hurricane argument. This was a clockwise rotating hurricane arriving at C.I. ~two weeks after MH370 disappeared.
If it ditched or crashed near C.I. the resulting debris already floated north-east for two weeks well within range of beaching on Java or Sumatra. When Gillian arrived this debris would have been pushed to the north even faster initially, not south IMO.
CSIRO (and others like Pattiarachy) already provided conclusive answers that the latitudes you propose (~11S/~21S) are virtualy impossible to match with the found debris locations and there timeframes (especially the Flaperon and Roy-piece).
GEOMAR only did reverse drift studies based on the flaperon landing-site at Reunion. They initailly came up with almost half of the IO as a possible crash-site. Which was not helpfull at all to decide on a managable crash-region.
Later they refined their study and limited the area to between ~23S and ~35S. Still way far from C.I. and your proposed drift-analyzis.
Since we’ve heard nothing from GEOMAR again.
Let me end with saying that I realy appreciate your study and effort.
It won’t influence the search of OI though.
If the plane not gets found this time maybe your scenario will be considered more seriously after.
@TBill said However, if MH370 is found east of about 34 South, it was probably intentional path. I am thinking 34-36S is approx. boundary between intentional and unintentional, which is good to have a search area that has both arguments going for it.
I’d put the northern limit for automated flight after 19:41 to be closer to around 26S latitude, if you allow for what the DSTG believes is typical drift of the BFO bias. That’s already further north than OI will likely search, and further north than the drift models suggest. I don’t dismiss the possibility of pilot input during the cruise after 19:41. I just don’t see it as a practical way to to refine the search area.
@GlobusMax,
You said: “I have been looking at your fuel model v5.3. Thanks for that, it’s great. I saw in a reply somewhere on here you estimated 1-2% accuracy. Would that be to say total flight time estimate is estimated to be accurate within 1-2%? I am specifically looking at the Mach 0.84 and LRC cases.”
My estimated error is +/- 8 minutes (2 sigma) in endurance, which is +/- 2% (2 sigma) in average fuel flow. I would expect M0.84 and LRC to be quite accurate (probably better than my general error estimate) since I have Boeing tables for those cases.
@Brian Anderson,
You said: “Is there any logic behind a 15nM contingency procedure offset, starting at about NILAM. . . . Or, is the contingency procedure evidence of some irrationality creeping in? Perhaps followed by a manually induced hold using the MCP . . . which was incomplete for some unknown reason.”
Those are good questions, and I have wondered about that myself. It seems that, at least some (if not most) of the route after 17:21 until VAMPI was reached circa 18:15, may not have been flown in LNAV. By the time VAMPI was reached, it is clear than N571 was being followed in LNAV. About 7 minutes past VAMPI, the power is restored to the SDU (and probably the Left AC bus) and a lateral offset Contingency Procedure (“LOCP”) was executed. The order of execution of these two actions (power up and contingency procedure) is puzzling. I’m thinking perhaps the pilot thought it was more important to first get the power back up, but, once that was done, then the contingency procedures were followed.
I don’t know how you can manually induce a Hold using the MCP. You can set the speed in KIAS using the MCP if you want to slow down, but you need the FMC to fly a racetrack. I can envision that a Hold in progress could be exited by simply using TRK SEL or maybe even TRK HLD on the MCP. That could be done even by a cabin crew member. All you need is two (TRK SEL then TRK HLD) or one (TRK HLD only) button pushes.
@Andrew,
Thanks for your useful comments and information. Based on your advice, I have revised all my recent papers to use the term lateral offset Contingency Procedure (“LOCP”) instead of Strategic Lateral Offset Procedure (SLOP).
My publications, including the latest revisions, are all available HERE .
The MAS procedures you linked also say this: “1.8.4.7.2.3 Before commencing a diversion across the flow of adjacent traffic, the aircraft should, while maintaining the 15 NM offset, expedite climb above or descent below levels where the majority of aircraft operate (e.g., to a level above FL 400 or below FL 290) and then maintain a level which differs by 500FT (150 M) from those normally used. However, if the pilot is unable or unwilling to carry out a major climb or descent, the aircraft should be flown at a level 500 FT above or below levels normally used until a new ATC clearance is obtained.”
That is very interesting. If I understand the procedures correctly, there could be two changes in Flight Level in the situation we are discussing here. First, if the aircraft is initially at roughly FL340, then there would be a 500 foot change up or down in coordination with (simultaneously?) making the 15 NM LOCP off N571 (with an initial turn of at least 45 degrees). Then, at a later time, if a left turn were desired that would cross N571, a second altitude change either above FL410 or below FL290 is required that is also offset by 500 feet, so the new FL in this case would be either FL405 or FL285.
I don’t know which is better (up or down), but at the current weight FL405 is certainly attainable and near optimum for cruise. So, let’s say the PIC decided to climb to FL405 instead of descending to FL285. FL405 is quite close to the altitude I previously found (FL407) for matching the endurance (using any one of MRC/Holding/250 KCAS speed modes). Maybe this is a coincidence, but maybe not. The contingency procedure for crossing an airway appears to provide a reason for making a climb/descent circa 18:40.
If the PIC were following the contingency procedure for the lateral offset, then there should also be an initial climb/descent of 500 feet at about the same time as the lateral offset (which may or may not affect the sparse BFOs measured then because the duration would be less than a minute). In addition, if a left (southward) turn were desired at a somewhat later time, another climb should be done to FL405.
Let me repeat my previous question (to Mick and now to you): “The second item of interest was the change in altitude by 500 feet (when below FL410) as a third step in the flow chart. When do you think that FLCH would be done – during or after the lateral offset?” A second question is, would a climb to FL405 be completed before the left turn was made across N571, or might they occur at the same time?
@VictorI
Just my simple thoughts on this I wonder about in the discussion about an automated flight after 19:41 or one with pilot inputs after.
It seems to me that any flight after 19:41 till 00:11 at least will be a programmed automated flight towards a pre-set waypoint/destination.
Like all (long distance) flights are A/P flights till arriving close to destination.
If a consious pilot was at control before and after 19:41 he would also have opted for the most reliable and simple heading/track configuration with A/P settings for the stretch between 19:41 and 00:11. Asif no pilot was at the controlles.
Maybe some small changes were made in altitude settings along the way or other minor adjustments. But IMO no sudden great changes in heading/track, altitude or speed.
In this regard a fully automated flight after 19:41 till 00:11 is almost the same as a pilot-controlled flight in the same time IMO. Just like in regular flights the pilot sits back and the plane flyes his route.
The difference comes only after 00:11 when first engine flames out. Then the pilot-controlled scenario makes all the difference against un unpiloted scenario IMO. The pilot has to take charge now and direct the plane to its destination.
@Ge Rijn: If small changes are manually input, then the difference between fully automated flight and one with pilot input is…well…small. Again, assuming the flight path was automated after 19:41 already defines a possible search area that is larger than will likely be searched. Removing constraints will increase the search area.
@Victor
Removing constraints will increase the search area.
Yes.
My issue is that people keep asking the wrong question. I have no quarrel with the answers.
In the case of the original search, the ATSB basically said they were going to start an underwater search and asked the analysts where is the best place to look. The analysts picked the priority area used, and I am in full agreement with that choice given the information available at that time. The ATSB question should have been – do we have enough information to warrant starting an underwater search. The answer at the time was no, IMO.
Forward ~3 years. OI says they want to start another underwater search, and asks where is the best place to look. The new priority area is once again a very sensible answer (although I prefer ~30S, but that is irrelevant). Once again, if the question was – do we have enough information to start an underwater search, the answer would be no, IMO.
Well, we have all seen this movie before. Hopefully it will have a different ending, but I would not pay for a seat in the theater.
@VictorI
I think I just wanted to point out that a fully automated flight after 19:41 does not say anything about the possibility a consious pilot was behind the controls or not. Consious pilots are behind the controls luckily all the time during fully automated flights.
If so, after 00:11 till (and after) 00:19 the flight could have been taken over by an active pilot.
If so this could increase the search area but not that much in search width IMO if the final BFO’s indeed indicate a high speed descent.
To the north at ~26S the current drift-studies exclude this latitudes IMO.
But impossible? I’m not in a position to state this. Maybe it is possible.
We’ll know in a few months time what is impossible anyway.
@Andrew,
Could you please post the entire documents for the ICAO ATM and the Malaysia versions? I see that changes have been made after 2014, and it is important to use those documents in effect in March 2014, and I was unable to locate them online.
Another interesting requirement in the contingency procedures is to turn all exterior lights on.
A more careful reading of the ICAO Contingency Procedures shows me that the altitude should be changed by 500 feet once the separation during the 15 NM offset has reached 10 NM. The Malaysia document implies the 500 foot change should occur after the 15 NM offset is established. So these two slightly conflicting documents answer one of my previous questions.
The other question had to do with when the altitude should be changed when diverting across an airway. The Malaysia manual says that the climb or descent should be done while maintaining the 15 NM offset, prior to commencing a turn across the “flow of traffic”. The ICAO section on this particular maneuver is not in the file you posted. That’s why I wanted to get the whole document.
@DrB @Victor
“Before commencing a diversion across the flow of adjacent traffic, the aircraft should, while maintaining the 15 NM offset, expedite climb above or descent below levels where the majority of aircraft operate (e.g., to a level above FL 400 or below FL 290)”
Note that the PSS777 simulator cases go to higher altitude at DOTEN about FL400.
@DennisW said: My issue is that people keep asking the wrong question. I have no quarrel with the answers.
For the government-funded search, I think the right question the policy makers should have asked the operators running the search is “What probability do you assign to finding the plane in the recommended search area, and how much will it cost?” Then, policy makers can do the cost-benefit analysis to determine whether or not to proceed.
We now can say with a high degree of certainty that the probability of success was overstated for the first area.
I believe that Ocean Infinity has a good understanding of the probability of success for the new search area because they have received information from a variety of sources, including the information on this blog. They have done their own cost-benefit analysis and decided to move forward. We should help them in any we can.
@TBill: If the intention was to fly to the SIO, it makes all the sense in the world to climb to a higher, more efficient altitude once the decision to fly south was made. If the pilot was dealing with an emergency, or the plane was to enter a holding pattern, it makes no sense to climb to a higher altitude.
@DrB @Victor Thank you much.
@DrB It looks like you (correctly, imo) discount the potential “high speed maneuvers” near IGARI. Are those included or discounted in your error estimate? Also, may I potentially publish one of your graphics from this on my blog?
@ DrB,
“I don’t know how you can manually induce a Hold using the MCP”
Im not suggesting an automated Hold. Rather, simply by dialling a new heading on the Heading/Track window on the MCP the aircraft will commence a turn. Ex 777 pilots I have spoken to have often suggested this method of initiating a turn, particularly if they might have been very busy attending to some other issue at the time. All that is necessary to continue the turn into an orbit, maybe even more than one, is to continue to advance the heading in advance of the aircraft.
Dennisw,
You sound like that Malaysian minister who claimed that he needed exact location to start a new search. But that wouldn’t be a search it would be just fishing up the wreckage.
@GlobusMax,
You said: “@DrB It looks like you (correctly, imo) discount the potential “high speed maneuvers” near IGARI. Are those included or discounted in your error estimate? Also, may I potentially publish one of your graphics from this on my blog?”
My fuel model assumes the airspeed was increased from ECON with Cost Index = 52 to M0.84 at close to the same time as the diversion from the flight plan began circa 17:22. I don’t think speed estimates derived from measured radar track positions close together in time are reliable, if those are what you are referring to. You, or anyone else, is free to reproduce all or part of any of my papers. I just ask you provide a reference crediting my work and also provide the context to avoid misinterpretation.
@VictorI,
You said: ” If the pilot was dealing with an emergency, or the plane was to enter a holding pattern, it makes no sense to climb to a higher altitude.”
It also makes no sense to fly into the SIO until fuel exhaustion.
Dennisw,
Now I am asking you. Don’t we have to start from somewhere especially with the limited data we have ? In my opinion the search for MH370 should be target elimination. Only this way we will come ever close to the wreckage.
Fugro did it’s job and they already eliminated areas for further searches.
Now let’s see what OI will do. Best of luck to Oliver Plunkett and his team.
@sk999: I looked at the BTO log-on statistics that we have through the log-on at 16:00 to determine what offsets might be applied to the log-on request and log-on acknowledge. There were 29 cases in which there was an R-channel burst just after the initial (R600) log-on request and (R1200) log-on acknowledge. Of those 29 cases, the number of packets in the burst was 3 for 20 bursts, 2 for 6 bursts, and 1 for 3 bursts. I averaged the values in each burst, and used that as a reference for the log-in request and log-on acknowledge (without correcting for position skew between data sets). When there were two or more sequential log-on requests, I only included only the first as the subsequent log-on requests were not independent. In 4 of the 29 cases, the correction for the log-on request was near zero, so only 25 cases were included for log-on request statistics.
I found that for the log-on request, the mean offset is 4578 μs with a standard deviation of 94 μs. The maximum offset was 4800 (+222 from the mean) and the minimum was 4380 (-198 from the mean). That’s surprisingly bad.
For the log-on acknowledge, I considered a correction of the form (a + N*W), where a is a constant, N is an integer, and W represents the delay per slot. I found the standard deviation of the correction error (again using the R1200 burst as the reference) to be minimized for W = 7812.0 μs. That’s very close to the 7812.5 μs error suggested by the hardware design. By forcing W=7812.5 μs, the mean error to the correction is 23 μs, and the standard deviation is 30 μs. That’s close to the 29 μs recommended to use for “normal” R1200 values. Therefore, a reasonable correction appears to be (23 + N*7812.5).
Using these correction values produces BTO values at 00:19 equal to:
00:19:29: 23000 – 4578 = 18422
00:19:37: 49660 – 23 – 4*7812.5 = 18387
Combining the two, using the inverse of the variance as weighting, produces:
(18422/94^2 + 18387/30^2)/(1/94^2+1/30^2) = 18,390
I’d base the 7th arc on this value.
@DrB said: It also makes no sense to fly into the SIO until fuel exhaustion.
Unless that was the intention. Not something I would advise doing.
Doesn’t make sense to me. Dump the aircraft in a remote part of the Indian Ocean so it can’t be found…. Yet take the risk of flying the aircraft back across land & up a busy air & shipping corridor (Malacca Strait). Hoping everyone is either half asleep or incompetent.
I am NOT going into conspiracy mode but
Doesn’t make sense to me. Dump the aircraft in a remote part of the Indian Ocean so it can’t be found…. Yet take the risk of flying the aircraft back across land & up a busy air & shipping corridor (Malacca Strait). Hoping everyone is either half asleep or incompetent.
I am NOT going into conspiracy mode but I still think Malaysia knows more than it’s admitting too.
@Brian Anderson
You asked if the offset possibly seemed irrational.
I don’t think it was at all irrational, on the contrary, I think it was an integral part of the overall plan. A plan that was carefully worked out in advance and carried out to the letter.
The 15nm offset was made shortly after flying out of primary radar range, as was the LH AC bus reconnect. The timing is significant; once the danger of real time primary radar tracking (tracking after the event was not going to be a threat) the offset would go undetected, and so would the upcoming FMT scheduled to begin as soon as he was 25nm from abeam of IGOGU and more importantly, because the FMT would align onto manual waypoint S41.00, E88.00, the flight path after 1822 would be impossible to reconstruct afterwards simply because it wouldn’t be possible to tie it in with any of the regular waypoints.
The SDU was de energised until he got out of radar range purely to make it look as if the plane had fallen out of the sky when the transponder was switched off. The idea was to deter MAS and/or ATC from requesting a real time search by the military, possibly the Indonesian military, until he was well out of range.
A 15mn right offset now fits in with my flight path beteen 1829 and 1941 and with the 1840 BFO just about perfectly: after 1829 he begins to slow from M0.84 to M0.82 and to climb gradually from FL350 to FL400. FMT starts at 1835.50, and is complete at 1838.15 (a 2.65min turn through 108deg, at 0.68deg/sec, radius 11.40nm, ground speed 486kts neglecting wind). He reaches cruising altitude FL400 shortly after 1840. Constant M 0.82. Average airspeed between 1829 and 1941 is M0.821.
Total distance covered between 1829 and 0017.50 is 2789nm (measured at altitude) which ties in very well with Dr Bobby’s fuel model for M0.82.
@Victor
“If the intention was to fly to the SIO, it makes all the sense in the world to climb to a higher, more efficient altitude once the decision to fly south was made. If the pilot was dealing with an emergency, or the plane was to enter a holding pattern, it makes no sense to climb to a higher altitude.”
OK but in the context of the current discussion, FL400+500 could be consistent with the flight rules for crossing over the N571 etc. If MH370 took the sim path, McMurdo path, I would expect descent to FL280 IGOGU to DOTEN (as previously discussed for N877 path) with U-turn possibly coming back at FL400+ to take the “high road” over.
It does not currently seem like that sim path was followed by MH370, possibly because the Beijing route presented different options.
@VictorI
@DrB
“To run out of fuel for such a short distance would require the plane to be flown at an inefficient altitude and speed for an extended period of time. To me, this is inexplicable.”
Crossing safely underneath Routes N571 and P574 with Minimum Fligt Level (MFL) at 280 and later Routes P627 and P756 at MFL 260 and 160 respectively appear to me a good rationale. This is even more important if the TCAS was switched off.
These low levels explain the overconsumption and this was confirmed by Bobby’s fuel consumption model.
This raises the question of what is at stake?
Going nowhere efficiently or going somewhere inefficiently ?
All options should be considered, shoudn’t they ?
Victor,
Looks reasonable. 18390 is the value I am using as well.
Here are the .cvs and .kmz 7th arc files using the 001929_R600 and 001937_R1200 observations combined as Victor described above:
https://goo.gl/cHPuUU
There are several small breakthroughs here. An improved estimate of the final arc is the ultimate result. But the details are also interesting.
One of the breakthroughs was the discovery that the R1200 offset is not a fixed value for all R1200 Logon Acknowledge transmissions (i.e. N*720 usec). ATSB disclosed 2 years ago that the SDU has a clock running at at 128 Hz, which has a period of 7812.5 usec. That value was suspiciously close the average of many observed LOA bias values, 7820 usec. The 3Dec2015 ATSB Update at page 20 states:
“Burst timing offset (BTO)
The communications corresponding to the burst timing offsets (BTO) used to determine the location arcs were transmitted on 2 communications channels; the R1200 and the R600 channels.
The R1200 log-on acknowledgement BTO measurements that occurred at 18:25:34 and 00:19:37 were initially ignored due to an anomaly in the values. Instead, the R600 BTOs at 18:25:27 and at 00:19:29 were used to create the 1st and 7th arcs.
Because the R1200 BTO measurements were assessed as having a lower standard deviation, identifying the cause of the anomaly, in order to possibly correct it, would improve the accuracy of the model.
An analysis of similar transmissions from the previous flight was conducted which identified a variable offset coefficient with a factor of approximately 7820 μs (i.e. N*7820). The manufacturers of the SDU and the ground station equipment were unable to determine a specific reason for the offset, however they did note that the reference clock used for the time slots was 128 Hz, which equates to 7812.5 μs.”
A while back, I proposed that the form of the LOA correction might be different. I suggested it might be A+(N*7812.5) rather than N*7820. IOW, the true offset is indeed a function of the number of clock cycle slips but with a much smaller random number added. Victor has used statistical analysis to show that this is indeed the form of the correction needed, and the model now fits both the hardware as we know it, and the calibration observations. That provides the basis for a more confident use of the R1200 LOA values, especially important at 00:19:37. This is important because the R1200 00:19:37 transmission has a lower noise level. But until now, the R1200 LOA BTO error has not been clearly understood, thus discounted in usefulness. Now it can be used at 18:25:34 and 00:19:37 with confidence.
Another small breakthrough was the exploitation of the many near contemporaneous LOR/LOA events that occurred during MH371 on 7 March 2014. By using all those logons to calibrate the LOR R600 and LOA R1200 corrections, we can have more confidence in the accuracy of these crucial calibrated values. When combined with the well researched and calibrated base BTO Bias (495679 usec), we can nail the 7th more accurately than ever before.
@ALSM wrote “A while back, I proposed that the form of the LOA correction might be different. I suggested it might be A+(N*7812.5) rather than N*7820.”
Perhaps we both came to that conclusion, certainly (N*7812.5), around the same time. I do wonder if ‘A’ is implicit in the R-ch 1200 BTO when the starting slot is 0 (as it should be).
For those who are interested, here are the numbers:
Position at 1829 is N7.17 E95.702 (taken from Dr Bobby’s paper) Speed is M0.84, FL350, M1 taken as 593kts, then begins slowdown to M0.82 and gradual climb to FL400. Slowdown completed and FMT begun at 1835.50 at position N7.565 E94.902, when 25nm away from point abeam with IGOGU. KTAS486 neglecting wind. Position at 1840 is N7.136 E94.598 bearing 186.69.
Climb to FL400 completed by 1844 approx.
Distance from 1829 to FMT start point is 53.27nm
Distance covered during FMT is 21.50nm
Distance from FMT completion to 1941 arc is 508.41nm
Total from 1829 to 1941 is 583.18nm
Airspeed average M0.821 neglecting wind, for M1 average 591.5kts
Aircraft proceeds toward waypoint S41.00 E88.00 at constant M0.82, when varying air temp and winds taken into account.
MEFE position S37.61 E88.66, 47.50nm downrange of 6th arc. Sun approx 6deg above horizon.
Distance covered from 1829 to 0017.50 is 2789nm (as measured at FL400)
If pilot glides 16:1, aircraft will impact at approx S39.20 E88.36. Pilot would probably want to achieve as much distance as possible following MEFE.
A circle of radius 16nm centred on S39.20 E88.36 is 95 percent certain to contain wreckage.
25,000sq km search zone has zero chance of containing wreckage.
Rob: re: “Climb to FL400 completed by 1844 approx.”
I appears you are ignoring the 1840 BFO data which indicates a 2500 ft/min descent underway at that time (or a turn), not an ascent. Please explain.
Also, “…average 591.5kts…”? Typo?
@ Victor,
Taking the 00:19:29 and :37 independently, one might hypothesise that the track was again crossing the 7th arc from outside to inside at 00:19:37. Suggests the final impact may have occurred inside the 7th arc.
@Don Thompson
@All
RE: “The above applies to a pre-TCAS II 7.1 configuration. If any reader can confirm the TCAS II 7.1 behaviour, it might help resolve the TCAS question. “
The TCAS II 7.1 behaviour is exactly the same as you described. TCAS traffic will NOT be displayed on the ND unless the transponder mode selector is selected to TA ONLY or TA/RA.
On 19th January at 07:33 UTC, Seabed Constructor was at 36.5687°S 77.2133°E, 3,594 km into the journey from the last stop and with 1,272 km to go. The average course is 88.7367°T and the average speed 13.4 knots.
Seabed Constructor is expected to arrive at 36.3°S near the 7th Arc on 21st January at around 09:00 UTC, which is 16:00 Local Time.
@J-L
@all
Am I the only person here who wonders why a conscious pilot — whether a murder-suicidal Shah or some desperado emerging from the electronics bay and managing a dwindling fuel supply — would worry much about mid-air collisions? In the former case, especially, we have someone who’s thrown his life, and 200+ other lives in his charge — to fate. A mid-air collision might be possible, but really the least of his worries..
To knock on about this invites the image of an escaping criminal in the midst of a car chase, who stops to let an old lady across the street..
@ikr
“To knock on about this invites the image of an escaping criminal in the midst of a car chase, who stops to let an old lady across the street..”
If you’ve taken all precautions available to avoid detection, and your intention is to disappear, why would you allow yourself to be identified by stopping to let the old lady cross the street?
The analogy can be translated into a mid-air is collateral damage (to be avoided) and you haven’t disappeared. In fact the matter would have faded well into the history books a few months later, whereas we have what we have now – 3 years and 9 months later, still an unresolved mystery.
The first is unplanned, while the second is the result of a well executed plan.
@DrB
RE: “Could you please post the entire documents for the ICAO ATM and the Malaysia versions? I see that changes have been made after 2014, and it is important to use those documents in effect in March 2014, and I was unable to locate them online.”
The complete Section 15.2 SPECIAL PROCEDURES FOR IN-FLIGHT CONTINGENCIES IN OCEANIC AIRSPACE can be found here. These pages were extracted from the 15th Edition of ICAO Doc 4444, including all amendments up to 18 November 2010. The contingency procedures are the same as those found in the 16th Edition of ICAO Doc 4444, published in 2016, so it’s safe to say these are the ICAO procedures that were current in March 2014.
I only have the latest amendment to the Malaysian AIP, dated 9 November 2017; however, the pages that document the contingency procedures are dated 10 November 2016. The entire section has revision marks, which might indicate that it was new information that wasn’t included in previous amendments of the document, rather than a revision to the procedures.
RE: “The other question had to do with when the altitude should be changed when diverting across an airway. The Malaysia manual says that the climb or descent should be done while maintaining the 15 NM offset, prior to commencing a turn across the “flow of traffic”. The ICAO section on this particular maneuver is not in the file you posted. That’s why I wanted to get the whole document.”
The ICAO procedures don’t feature a paragraph that deals specifically with a diversion across the flow of adjacent traffic. Instead, you need to read between the lines. Para 15.2.1.1 mentions the cases where the procedures might need to be applied (eg. ‘en-route diversion across the prevailing traffic flow’), while para 15.2.2.3 states the aircraft should turn left or right to establish a 15nm offset, then climb or descend 500 ft (below FL410) once 10nm from the assigned track centreline.
@Brian Anderson: The difference between the two corrected values is 35 μs. That’s only 37% of the standard deviation of the log-on request. The difference in the values falls well within the uncertainty of the measurements. I would not use the difference to draw any conclusions about the direction of the path.
@ikr
Am I the only person here who wonders why a conscious pilot — whether a murder-suicidal Shah or some desperado emerging from the electronics bay and managing a dwindling fuel supply — would worry much about mid-air collisions? In the former case, especially, we have someone who’s thrown his life, and 200+ other lives in his charge — to fate. A mid-air collision might be possible, but really the least of his worries.
You are not the only person.
You make a very good point, and it argues strongly against a murder-suicide scenario. As JLuc points out, and as I have pointed out ad nauseam, murder-suicide was not the preferred outcome of the diversion. At the time when steps were taken to avoid a mid-air collision “negotiations” were still ongoing. A decision to turn South into the SIO had not yet been made. A loiter West of Penang was the intention at that time. Shah still was hoping for a “green light” to land the aircraft safely (probably at Banda Aceh). What made the Christmas Island flight path so appealing to me is that is represented a continuum of possible landing sites, and an extension of the negotiation time frame. Alternatively, a “hard stop” to negotiation time could have been the plan, after which the “team” running the diversion scenario pulled the trigger on the “or else” part of the plan, and pointed the plane into the SIO. For sure, if you are negotiating something based on an act as serious as the diversion of a commercial airliner, you cannot back away from the threatened outcome if you are not successful. It would undermine any subsequent efforts by showing that you are not serious.
When the SIO coordinates were found on Shah’s simulator coupled with the 777 fuel remaining accuracy and no attempt to communicate position when fuel ran out, I was forced to reconsider the Christmas Island flight path. It became clear (to me) that it was “game over” sometime after 18:40. I spent a great deal of time trying to reconcile flight paths between 18:25 and 19:40 with the Inmarsat data. I was never able to find anything workable, and simply drew a white circle with question marks on this part of my path. Victor and Bobby have come up with reasonable alernatives that not only work, but support the notion that murder-suicide was not intended until later in the flight.
I am glad you made your point, ikr. Thanks.
@Andrews
@DrB
A precision: The PANS-ATM says “select a final altitude which differs from those normally used by 150 m (500 ft) if at or below FL 410, or by 300 m (1 000 ft) if above FL 410”
It does say descend or climb by 500ft only (resp 1000ft if above FL 410).
It says: at whatever altitude you can/may fly select a flight level in between other potential traffic. There is no limit in the climb or descent altitude to which the aircraft may chose to fly eventually, which is understandable for an aircraft whose capabilities are depending on what triggered the contingency procedure.
Thus this does not constrain the amplitude of the FL change.
Correction: It does not say descend or climb by 500ft only (resp 1000ft if above FL 410).
@Jean-Luc
Agreed – poorly worded on my part in an attempt at brevity. The requirements are clearly stated in both the ICAO document and the Malaysian AIP.
Simple question. Mh370 has been missing for nearly 4 years now. There have been lengthy debates about debris, drift & data to name a few.
Are we actually any closer now to understanding the why, how & where than we were 2 years ago & if we are then what has changed?
@Jean-Luc Marchand: I can’t reconcile the safe pilot who expertly manoeuvers to avoid the slim chance of a collision with the unsafe pilot who carelessly runs out of fuel by flying inefficiently.
@Michael John said: Are we actually any closer now to understanding the why, how & where than we were 2 years ago & if we are then what has changed?
If you believe the plane crashed or landed in Kazakhstan, South China Sea, Andaman Sea, Bay of Bengal, or the Maldives, then the last two years have been a giant step backwards. It must be very frustrating for those individuals.
For most of the contributors here that believe the plane crashed in the SIO, we have eliminated large areas along the 7th arc, and we have used additional information and insights gained from path models, end-of-flight analyses, drift models, aircraft imagery, and satellite imagery to establish where best to search next. We also now have new technology that uses a team of autonomous vehicles to more rapidly and cost-effectively scan the seabed, and a company with that technology that is willing to assume the economic risk of the new search.
I’d say that’s a lot of progress. Of course, since you continue to make comments that you believe the plane did not crash in the SIO, you would not agree.
MJ: Re: “Are we actually any closer now to understanding the why, how & where than we were 2 years ago & if we are then what has changed?”
I don’t think we are any closer on the “why” or “how” parts, but the “where” part is definitely better understood. The main reasons are:
1. 2014-2017 null search area (120 km^2)
2. debris drift analysis (further NE on the 7th arc)
3. debris forensics (small pieces including interior, etc.)
4. broader consensus on the rapid descent, supported by data and debris
@Michael John
Now you come over here to spread your ‘doubts’ and ‘questions’.
What is your objective with this? People making second thoughts about issues that are debated in detail and lenght the past two years?
Most here do this all the time naturally. They don’t need a reminder.
And yes, we’ve come closer in those two years IMO. If you think not, than first study all that has been investigated and learned in those two years by very capable and dedicated scientists and contributors here and else.
Anyway OI has made up his mind and they are going to search. No ‘doubts’ or ‘questions’ you ask are going to put them off-guard.
And probably they know more than all of us anyway.
Could well be they’ve got the complete French flaperon analyzis f.i.
@ALSM
I don’t think we are any closer on the “why” or “how” parts
The “why” question has never ever been a part of the IG vernacular. That you would go there now is a mystery to me. From the very beginning back in the “Duncan days” any consideration of “why” was rejected and vigorously discouraged. I am OK with that, but I am not OK with you commenting on it now when is clear you have never had any interest in that consideration.
@MJ
I have never been more comfortable with a likely terminus than I am at this moment. The work of Bobby and Victor has done a lot to support the notion of a late FMT which leads to terminal locations further North. The work and persistence of ALSM relative to the the final ISAT BFO’s has convinced me of the proximity to the 7th arc. The work of Richard on drift analysis has been the biggest contributor to a 30S conclusion, and it supports my own weak analytics in that regard. The next few months should get it done although I am still reluctant to bet on it. Many other people (I did not want to name names in the first place) have contributed a great deal of supporting information relative to the aircrft itself, and likely flight dynamics. We are miles ahead of where we were a year or two ago.
@ikr
As far as mid-air collision, I assume one possibility is that MH370 may have been trying for stealth to disguise the FMT south. Once the aircraft was heading south, the air traffic on FR24 is minimal to non-existent west of Sumatra at that time at night, so (re: Jean-Luc’s path) I would not descend to FL160 and prefer heading 180S at ISBIX.
I feel that the BFO at Arc2 19:41 suggests the aircraft was ascending about 300 ft/min (as shown in FFYap’s IG path spreadsheet). The calculated BFO at ARC2 is 104 vs. 111 measured, so if you are going to match 111, you have to be coming out of a maneuver, either a turn or an ascent or going slow.
I am like-minded with Victor on the “why”.
@Airlandseaman
Re “Climb to FL400 completed by 1844 approx.” It appears you are ignoring the 1840 BFO data which indicates a 2500 ft/min descent underway at that time (or a turn), not an ascent. Please explain.
Also, “…average 591.5kts…”? Typo?”
Mike, I am not ignoring the 1840 BFO data, on the contrary, I am using it to confirm the gradual climb to FL400. I refer you to Dr Bobby’s MH370 Manoeuvers Matching the 18:40 BFO, Case #3, on page 8, relevant section quoted here:
Quote “3.3 Ongoing Climb (Case #3)
Solutions “above” (i.e., inside) the parabola in Figure 6 are possible only with an ongoing climb, and solutions
“below” (i.e., outside) the parabola are possible only with an ongoing descent. In Case #3 the green-filled circle
(inside the parabola) has a turn to 180°, a slow-down to Holding speed, and a climb from FL340 to FL407 at a best- fit ROC = 303 fpm when climbing through FL400 at 18:40. The expected ROC for a B777 like 9M-MRO will be
performance limited to < 1,700 fpm above FL350, and it will be reduced to about 300 fpm at FL400 due to limited
available thrust [10]. Thus, the best-fit value of 303 fpm is a reasonable number for ROC near FL400 in a B777-
200ER. In addition, the stability of the ROC is expected to be better than the maximum 88 fpm fluctuation limit
based on the CBFOs, so this low and steady climb rate could have occurred at 18:40." Unquote.
I take this to indicate that the 1840 BFOs are consistent with a climb rate of around 300fpm. Let's be honest, there was no descent between 1829 and 1941, there is no good evidence of a loiter between 1829 and 1941, only evidence of people under the influence of "confirmation bias" who subconsciously seek an alternative explanation to pilot pre-planned murder/suicide. Victor, please forgive me but the desire to find a convoluted explanation for a simple situation (Occam's Razor, notwithstanding)is something that will have to be faced up to sooner or later, preferably sooner if there is to be any chance of finding the aircraft. I know you are honourable men and we all share a common purpose. Hopefully we can remain friends, despite our differences of interpretation.
"average 591.5kts" not a typo but an approximation, adequate for the purpose.
@DennisW
“The next few months should get it done although I am still reluctant to bet on it.”
I wouldn’t bet on it. Stay with the stock market, instead.
@Rob: A climb from FL340 to FL407 would likely be performed using FLCH or VNAV, where the thrust is set to CLB and pitch controls speed. The variation in BFO values are more consistent with either level flight or a V/S descent.
@Michael John
First of all, I would say on 15-March-2014 PM Nazir Razak said it looked like intentional diversion, and a week later he said it was in the SIO. I feel the passage of time has strengthened, not reversed, both assessments.
The main new discovery is that the assumed FMT (turn south) at 18:40 sat call was a questionable assumption. Part of the reason that assumption was made was to narrow down to a rational “hot spot” to search, whereas other assumptions (eg; starting from anywhere on Arc2) do not yield a clear hot spot on Arc7.
Now there is a justification to widen the search area, and relax the 1840 FMT assumption. Unfortunately what I just said is not new info, once the early part of the orig search came up empty-handed that meant mathematically that the Bayesian study assumptions were looking more questionable.
DennisW wrote: “The “why” question has never ever been a part of the IG vernacular.” Come on Dennis. That’s BS and you know it is BS.
I was answering MJ’s question honestly. Everyone including everyone in the IG would like to know “why”. The question has always been an important and valid question. Some insist on constantly guessing. I ponder the question everyday, but since I don’t know the answer, I don’t presume any particular scenario in the analysis. If we did know that answer, of course we would use it. But we don’t. Lots of conflicting information and theories, but do answers yet. Hopefully the boxes will be found and they will have some answers.
@TBill
I am like-minded with Victor on the “why”.
???????
@DennisW
You stated: “The “why” question has never ever been a part of the IG vernacular.”
This is not the case. Whereas, Duncan did not want to entertain the “why” question, this has been discussed at length in the IG. We have privately considered all major scenarios and their variants.
We have decided for various reasons not to make our conclusions public.
In my view, there is a large amount of circumstantial evidence, but there is insufficient legal evidence, at this point in time, to support any particular conclusion.
@DennisW
I would say we cannot rule out a suicide motive…
@Rob. At FL400 591.5 knots = mach 1.03. Surely some mistake…
@Rob
Essentially it’s not the Inmarsat-data anymore which define the possible crash-area but the debris-failure and drift-analyzis. Those led us to a region north of ~36S, not the Inmarsat or any other data.
The 18.40 BFO’s where used to prove the plane made a turn to the south towards the SIO instead of a route towards Kasaghstan.
This has all turned, for it could as well has been an ongoing descent as a turn at 18:40 or a combination of both. The SAT-call exactly happening during a turn I consider very unlikely. Happening during an ungoing descent started before 18:39 I regard much more likely.
Nevertheless the found debris and drift-analyzis prove IMO the plane ended up in the SIO north of 36S.
39S or 40S is out of the question now IMO. Even with early FMT’s shortly after 18:25 those latitudes are not reachable anymore according @DrB’s calculations. And all forward drift studies by now exclude this latitudes also.
@VictorI’s proposed ultimate latitude of ~26S could be within possibility range though. Debris arriving on Reunion and Madagaskar would be much too early from this latitude in general but caught up in an eddy it could be possible I guess.
In one of Richard Godfrey examples a buoy got stuck in an eddy for ~140 days. Uncertainty with those eddies make it very hard to predict specific starting latitudes I agree.
But then the statistic historical GDP-data and Adrift based on them comes to help IMO.
The region OI is about to search is based on all the best that has been provided to them here and else.
It’s up to them now. Searching only in the lenght close to the arc (till ~26S?) or also searching wider (in a later stage).
Few days now and they will start searching. I look forward to it.
Much respect for this commitment. We can easily criticize their endavour but they do it realtime.
@Richard
We have decided for various reasons not to make our conclusions public.
Thank you, Richard. That was my point. As you might recall I was invited to leave Duncan’s blog for addressing the question of “why”, and I am quite OK with that. The theme of the blog was simply not consistent with the “why” question.
@TBill
I would say we cannot rule out a suicide motive…
🙂 Not much of a “why” statement, but OK.
I just feel that once in a while it is good to pause the conversation & get an understanding of people’s thoughts to lessen the chance of being misunderstood at a later date.
Yes I’m not convinced that Mh370 is in the SIO Although my conviction on this is waning more & more each day. That said I have never intended to imply that the whole 7th ARC principle is wrong. My understanding is that the 7th ARC principle is based on the SDU performing without irregularities, I’m not talking of tampering or duplication, I’m more interested in how much we can trust the data to be as accurate as should be. Afterall we have no idea why the operation of the SDU was interrupted in the 1st place.
So. If anyone can guarantee that the data is without a doubt unquestionable then I will happily be as convinced as the rest you. Just because I am unconvinced it shouldn’t make me a martyr. I follow the work done here with genuine interest & my questions are not meant to be negative but a request for clarity.
Bottom line is no matter what some people think of me I am not a Troll.
@Michael John
My objection with your general comments is, that you tend to put everything that has been debated and in essential issues concluded/consensus after years of thought and investigation, again you bring all this in as ‘doubts’ and ‘questions’.
To me this comes across as distracting and trolling content.
In the years commenting on Jeff’s blog and now on VictorI’s site I’ve developped a sensitive view on comments like yours.
I might be wrong ofcourse. But for now I follow my intuition.
@Michael John
I am a little different than you, working with a lot of experimental lab data in my career, so I want to first assume the data is correct. I am comfortable with the 6 or 8 data points as that is usually what we had to work with. When I study the data, it looks like good data to me. If I thought there was a bunch of scatter, I would not be comfortable working with it.
My opinion is the sat data says that MH370 went 180S to somewhere between 32-35 South. @sk999 has recently referred to 32.4 South as the best fit calc.
Somehow 32-35 South is not where the search started…but I see that as the base case. If it is not there, then we have to say maybe the data is slightly off (Victor is saying McMurdo path requires some error bars on the BFO data) or totally off. But also we are hoping MH370 is close to ARC7 so there is some chance the data is perfect but the aircraft scooted off a ways.
@Paul Smithson,
@airlandseaman,
The post from Rob actually said: “M1 taken as 593kts”. I think you missed to “1” after “M”. I believe he meant the Local Sound Speed = M1 = M593 kts.
@Rob,
You said: “Let’s be honest, there was no descent between 1829 and 1941, there is no good evidence of a loiter between 1829 and 1941, only evidence of people under the influence of “confirmation bias” who subconsciously seek an alternative explanation to pilot pre-planned murder/suicide.”
The evidence may not be “good”, but there is indirect evidence of a loiter. That is because a loiter is required to make the straight paths work from 19:41 to 00:11. Even at 180 degrees true, you need about a ½ hour loiter between 18:40 and 19:41. So I would say that there are straight path solutions to the satellite data (how many depends on how large one is willing to let the BFO errors be), and in the region not yet thoroughly searched, you need ½-1 hour of loitering. There is also the possibility of a descent at 18:40, based on the BFOs then, and to some this can be interpreted as the beginning of a Hold that created the necessary delay in southbound movement.
@VictorI,
You said: “A climb from FL340 to FL407 would likely be performed using FLCH or VNAV, where the thrust is set to CLB and pitch controls speed. The variation in BFO values are more consistent with either level flight or a V/S descent.”
I disagree with this conclusion only for the case when the climb rate is very low. We know the ROD fluctuations during the 18:40 phone call, if a descent was underway then, are smaller than +/- 88 fpm. Let’s call it 50 fpm. Now when the ROD is 2,500 fpm, a 50 fpm fluctuation is only 2% of the average descent rate. I would think that would generally be exceeded, maybe by a large factor, in a typical FLCH or VNAV descent. That’s why you suggested the V/S mode as an alternative. On this we agree.
However, the Rate of Climb is quite low at high altitude near FL400, about 300-500 fpm, because of limited thrust. In this case the 50 fpm fluctuation limit from the 18:40 BFOs is now 10-16 % of the average ROC. Using the 303 best-fit ROC from my last paper, the ROC only needs to be stable within +/- 16 % of 300 fpm. That is quite different for the high-altitude climb case (16%) than for the high ROD case (2%), and I suspect the FLCH mode can maintain that level of precision (16%) in the ROC. I would think that FLCH would have the capability to execute a ROC of 300 fpm without exceeding 50 fpm error. Maybe Andrew or someone else has knowledge of the ROC variations at high altitude. If that is true, then low-ROC climbs would be consistent with the 18:40 BFOs in all climb modes: FLCH, VNAV, or V/S.
@MJ
I, and others here, have carried out wide and deep research to understand the conditions that are required to ensure the AES functions as required, to understand why ‘ACARS’ correspondence and FMS related parameters remained absent after the last routine aeronautical correspondence event at 17:07, the conditions during the SATVOICE calls, and to interpret key signal characteristics that were evident from the data recorded in the Log.
Before the Stratos Log became available I had partially disassembled the initial MH370 Data Communications Log into discrete messages (IIRC, I identified 22 messages in that relase of Log) and what those messages were likely to relate. One message was completely disassembled using transcripts ‘scraped’ from Richard Quest’s TV interview at Inmarsat (h/t Paul Sladen) and I went on to partially disassemble other records from that same source material. I present that as an example of the efforts made to improve understanding of what information was being made available.
Certain avionics data feeds are necessary for the AES to function, the ‘IRS’ data as an example. As an example, there is no evidence to support an idea that this 777’s ADIRS (Air Data Interial Reference System) was in any way impaired during the flight. There is evidence to assume that the ADIRS operated normally. For that ADIRS data to be transferred to the SDU, there’s a significant network of avionics capability that simply has to operate normally.
Considering ‘ACARS’ and FMS related parameters, the simple answer is that the functions within AIMS, concerned with managing FMC and ACARS related data flow to the AES were inhibited. There is a simple, single click, operation on the MFD menu system that inhibits use of SATCOM by the FMS and ACARS related functions.
Consider the ‘IFE’ communications, the presence of ‘IFE’ related activity on the datalink yet the absence of ACARS correspondence after the GES Log On at 18:25. The IFE system, and the protocol it employs over the SATCOM datalink, is not ACARS. Onboard the aircraft, the IFE correspondence to the AES involves the same physical buses as ACARS. Certainly, all evidence indicates that IFE, AIMS, and the AES was operating normally but the timing of certain events indicates restoration of service after power interruption.
It’s a reasonable conclusion that ACARS and FMC were inhibited from communicating with the AES, not because something was broken, but because a ‘tick’ in check box was cleared.
That the AES operation was restored at 18:25 and continued to operate without interruption until some time after 00:11, then restored operation again at 00:19 is indicative that the aircraft systems were operating normally.
The signalling involved with the SATVOICE calls was consistent in protocol analysis, the signalling exchange was maintained for a consistent period, and ultimately cleared as unanswered.
The GES Log published by Malaysia in May 2014 and the subsequent full ‘Stratos’ GES Log provided to this blog’s host in June 2017 are consistent. To contrive a scenario that these have somehow been contrived in a grand conspiracy is, quite frankly, blofledian nonsense. The full ‘Stratos’ Log XLS file includes worksheets detailing events for another six elements of Inmarsat’s network that are related to 9M-MRO’s satcom datalink activity throughout 2014-03-07. One worksheet even records information back to 2014-03-05, and that information is consistent with ‘Aircraft Records’ compiled in the DPRM Folder 5.
I’ll give you “the data is without a doubt unquestionable”. It have no doubts for its validity. Interpretation of the data, its usefulness, requires further comment: a) the confidence for accuracy derived from BTO has been refined over time as Victor describes above; b) the utility of BFO data has evolved, interpretation over a wide temporal range requires additional bounds, whereas interpretation over a narrow temporal range, or a single events, is improved.
It is a fact that, as you say, we don’t know why operation of the AES was interrupted between 17:08 and 18:25. We do know the circumstances of how that could have occurred, and we have some well argued conclusions for what the immediate effects of the power restoration.
In attempting to substantiate ‘unquestionable’, this comment has got long enough. The information is out there, it is tough stuff to fully comprehend, but that does not allow a stance where its discounted on intuition alone. Megabytes worth of garbage has posted questioning the satcom data, and its interpretation, on mere bogus intuition.
@Rob
@DennisW
“The next few months should get it done although I am still reluctant to bet on it.”
I wouldn’t bet on it. Stay with the stock market, instead.
Unfortunately California residents pay the 2nd highest capital gains tax rate in the world (behind only Denmark). Most of it goes to paying for Governor Brown’s “bullet train to nowhere” which is now $24B over budget ($64B verus $40B) and seven years behind schedule. Ami and I are considering converting all our assets to bitcoins, and fleeing to South America (before April 15, of course). Understanding block chain technology is well beyond the capability of any US government agency – CIA, NSA, …
@DrB: I don’t think the variations in the vertical speed are related to the magnitude of the vertical speed, nor related to the precision of the vertical speed measurement. For a constant vertical speed descent, the plane is in a state of quasi-steady equilibrium. Variations in vertical speed are caused by disturbances such as turbulence, which are counteracted by the closed-loop control of vertical speed by varying pitch, which targets zero vertical acceleration. Level flight is a special case, in which vertical speed is zero. I believe the vertical speed variations in level flight would be similar to the vertical speed variations during a V/S descent, and relatively independent of the magnitude of the vertical speed for similar turbulence levels. Similarly, for a flight mode in which pitch is controlling airspeed (or Mach), I don’t think the variation in vertical speed depends on the vertical speed.
@DennisW, lkr
“Am I the only person here who wonders why [anyone] would worry much about mid-air collisions?”
If the intent was to ensure a ‘perfect’ disappearance, then the risk of a mid-air collision while crossing an airway is worth mitigating.
Ah, but for those pesky pings, perfection might have been attainable.
Reminder to self: check markup before hitting submit.
@Don Thompson,
Your January 19, 2018 at 2:44 pm post is in my opinion a very good summary of why the SDU data is to be believed.
I hope others who may be wondering if the SDU data was unreliable / interfered with, take your explanation onboard.
@Barry C
Thank you. And that wasn’t even the half of it!
I could add that I received logs of Inmarsat Aero traffic acquired by the developer of J-AERO software. One log file included traffic from one of 9M-MRO’s sister B777s, that also showed much consistency with the 9M-MRO log.
Victor wrote, “I believe the vertical speed variations in level flight would be similar to the vertical speed variations during a V/S descent …”
I question that statement. The atmosphere is much more uniform horizontally than vertically (i.e., it is much easier to mix horizontally than vertically). One can see that in the GDAS models, where quantities such as wind speed, humidity, and temperature gradients vary much more vertically than horizontally, E.G., at IGOGU, horizontal flight at 30,000 feet experiences a wind shear (knots/min) that is 2-6 times smaller than in a vertical descent of 2900 fpm. The models don’t have enough vertical resolution to say anything about temperature gradients on small scale, but when examined from ground to 44,000 feet, there is are fluctations from smooth behavior as a function of altitude and from day to day.
The Inmarsat data is real and well scrubbed, analyzed, filtered for anything and everything that looks odd or out of place. It is a very clean data set. There are no signs of error, deliberate or otherwise, except for a few OCR character errors spotted immediately in 2014. Some of the data took longer to understand (Offsets, statistics, etc.), but we have it well understood now. Don’s summary is excellent. Victor’s statistical analysis of the MH371 data and resulting improvement in the accuracy of the reference 7th arc is an important advance in our understanding.
I’ve been working with satellites, earth stations, modulators, demodulators, transmitters, receivers, antennas, electromagnetic fields and propagation, multipath, refraction, TEC,…. and lots of airplanes my whole adult life (>40 years). The BTO and BFO data are good. We should stop wasting time revisiting the question and move on to more uncertain questions.
@DennisW
And there’s me thinking California, the Golden State, the promised land where anyone can make their fortune (if they get lucky). It just goes to show how wrong you can be.
But be careful, South America looks a bit of a s++t hole kind of place.
Meanwhile at the Other Place, a poster has decided that the “real mission” of OI is to plant debris fields for the Search Which Will Comer After It….
@ALSM
We should stop wasting time revisiting the question and move on to more uncertain questions.
For sure. It has been a slow day. I am glad Don dealt with it the way he did. I too get frustrated by the data doubters.
@sk999 said The atmosphere is much more uniform horizontally than vertically (i.e., it is much easier to mix horizontally than vertically).
I agree. That’s why I qualified my statement with “for the same levels of turbulence”. I think you are more likely to hit some turbulence during a 20-minute descent than for 20 minutes of flight at level altitude for the reason you mentioned. That doesn’t mean you will hit turbulence for most of the ride down.
@Rob
But be careful, South America looks a bit of a s++t hole kind of place.
I have spent a ton of time down there hiking and rock climbing, and I am known as Perito Dennis. A Perito is a Spanish term for a knowledgeable and respected scholar, as in Perito Moreno after whom the glacier is named. It is not uncommon for friends there to tell people to go ask the “perito” (me) the answer to difficult questions.
So far they (my South American friends) have been unable to produce someone who can defeat me at chess, but it is ongoing. In any case you can become an instant citizen there by walking into a country with a threshold level of cash (especially if you have friends there who will vouch for your character and continued “employment”).
Plus that, as Duncan has calculated, South America is the place to be as global warming escalates.
I do want to spend some time addressing Richard’s query (overlooked by everyone but me) relative to the equivalence of gravitational and inertial mass. I feel like my recent foray into geometric calculus (which makes tensors a trivial untaking) has prepared me well for that undetaking. Dr. Parkinson, who I know well from GPS days, has an experiment going on at Stanford which obliquely addresses that issue, but he is no longer talking to me since I told him the only use for magneticism was sticking things to your refrigerator.
@VictorI,
My personal experience in aircraft and my knowledge of atmospheric physics both indicate the vertical speed variation is noticeably more pronounced when rapidly descending than when in level flight. You feel this in the seat of your pants, because it modulates the net effect of gravity and the vertical accelerations of the aircraft. One reason for this is that the temporal frequency of disturbances caused by vertical density stratification is dependent on vertical speed. If you have a control system attempting to maintain a constant vertical rate (such as V/S), its ability to correct disturbances depends on the temporal frequency of the disturbance. Higher vertical speed means higher temporal frequency disturbances, which leads to incomplete control and incomplete removal of those disturbances. Your notion that the vertical speed stability is independent of vertical speed is incorrect. It will depend on the vertical speed. That is why I used the percentages of vertical speed in my example, because that will be the approximate form of the actual vertical speed fluctuations – they will depend roughly on the vertical speed.
@DrB: Higher vertical speed means higher temporal frequency disturbances, which leads to incomplete control and incomplete removal of those disturbances.
You are neglecting mass of the aircraft, which reduces the response of the system, i.e., the variation in vertical speed, when the frequency of the disturbance increases. Also, the frequency of the disturbance is probably more closely related to horizontal speed than vertical speed, since for a typical descent, horizontal speed is about 20 times higher than vertical speed, and the scale of the turbulent eddies is probably similar in the horizontal and vertical directions. Additionally, at a higher vertical speed, you would pass through the turbulent layer faster.
I question your analysis that relates vertical speed variation with vertical speed. If speed is controlled by pitch, irrespective of the vertical speed, I would expect variations in vertical speed to be signficant.
Thank You all for your responses.
I don’t know if anyone has seen this although I suspect some have but it is (IMO) a well written account on the search for AF447. I understand that OI has people in it’s team that worked on that mission.
I imagine the same principles will be used to help narrow down the search for Mh370. Is this a fair assumption?
https://arxiv.org/pdf/1405.4720
Trying to find answers to questions that haven’t been pored over time & time again is tough. Trying to not speculate too much is probably even harder.
Can we examine why we still haven’t had a public release about the French report into the Flaperon. The official Malaysian Government information page stated that the French would be providing further information in due course:
“The French-led investigation team examining the wing part
or flaperon of the vanished Malaysia Airlines (MAS) Flight MH370 aircraft has
concluded the first phase of inspection work.
Australian Joint Agency Coordination Centre (JACC) said the French authorities
would, in consultation with Malaysia, report on the progress in due course”
The full statement can be seen here:
http://www.mh370.gov.my/index.php/en/380-1st-phase-french-investigation-on-mh370-flaperon-completed-jacc
Can anyone give any indicators over why a public release of the report of the Flaperon didn’t happen & clarify whether there were meant to be further phases to the French investigation?
There has been much speculation over the Flaperon over the years & surely this report would go a long way to quelling some of those assumptions.
(If this is something that has been debated at length before & is of no interest for discussion then please ignore & move on).
On 20th January at 07:22 UTC, Seabed Constructor was at 36.4496°S 83.8694°E, 4,189 km into the journey from the last stop and with 637 km to go. The average course is 88.7350°T and the average speed 13.4 knots.
Seabed Constructor is expected to arrive at 36.3°S near the 7th Arc on 21st January at around 09:00 UTC, which is 16:00 Local Time.
@MJ
Your link from arXiv is a paper authored by Laurence Stone, Colleen Keller, and others at Metron Scientific Solutions. Whether Metron’s expertise might have been useful in the case of MH370 has been much discussed here. DennisW will no doubt comment. I hope the content of the paper isn’t a recent find for you, if so you really have been in the wrong bubble for the last 4yrs.
The Office of the Paris Prosecutor (maybe not such a good translation) opened an investigation into the loss of MH370, at the behest of the Wattrelos family. La Réunion is French sovreign territory, an overseas département. The flaperon was taken by the French judicial team, not the BEA, for investigations to be carried out by French national technical laboratories. Malaysia, as the lead accident investigator, did liase with the French at some level. Malaysian DCA personnel were photographed entering the French technical facilities. The DPRM (RMP) compendium of reports was compiled so as to be shared with foreign law enforcement agencies, certainly the French as its leak appears to have originated in that country. My own expectation is that Malaysia has received reports on technical analysis from the French. As the French interest is an ongoing terrorism related investigation, not an air safety investigation, there is no obligation to make any public releases.
@Don Thompson
“Ah, but for those pesky pings, perfection might have been attainable.”
Indeed. Such a grotesquely audacious plan, but a plan with an embarrassing (for him) flaw.
@MJ. As Don Thompson said, “My own expectation is that Malaysia has received reports on technical analysis from the French.’
Operational Search for MH 370, October,2015 page 103,”A number of other tests and examinations were made in France on the flaperon, including marine
specimen examinations (barnacles attached to the flaperon), failure analysis of metal and composite components and flotation testing. Details of all examinations were provided to the ATSB by the French judicial authorities to assist with the search for MH370 (the public release of any reports on the flaperon examination is the responsibility of the French judicial authorities).”
October, 2017
@DennisW
“I do want to spend some time addressing Richard’s query (overlooked by everyone but me) relative to the equivalence of gravitational and inertial mass.”
Yes, it’s about time that one was put to bed.
I myself have a couple of intriguing problems yet to make much headway on: Dirac’s theory on the nature of inertia, and Heisenberg’s uncertainty principle. Both waiting for the weekend when I get stuck in a lift.
I have the same problem as you, Dennis. Because we’re clever, people think we should know the answers to everything. and get cross or scathing if we don’t. I have an aunt who was rather chuffed at the search being called off, even took a poorly disguised delight in it. The word is schadenfreude.
But little does she know. The game’s not over yet
Yes Don.
I figured you & the other experts would have of course seen it as the paper is nothing new. The idea is that rather than posting a lengthy explanation Everytime someone asks how the search will be carried out once the General wreckage site is found then having a link to a similar scenario may be easier. My query related to whether the principle methods laid out in the paper are still valid today.
@David
“Details of all examinations were provided to the ATSB by the French judicial authorities”
Yes. But problem is IMO the ATSB also never published any of those details or conclusions made by the French on the most probable failure/seperation sequence. They must have made some kind of conclusion IMO.
This all is not public and thus cannot be independently checked.
Why is the ATSB not sharing this information with the community?
Anyway, I hope they share(d) it with OI.
@Ge Rijn
Take some time to understand the constraints under which entities, like ATSB, operate as set out by the standards and recommended practices of Annex 13 to the Convention on International Civil Aviation.
The non-disclosure clauses set out that the “State conducting the investigation of an accident or incident shall not make [certain] records available for purposes other than accident or incident investigation, unless the appropriate authority for the administration of justice in that State determines that their disclosure outweighs the adverse domestic and international impact such action may have on that or any future investigations.”
Translation: if ATSB makes an analysis that may be valuable to publish, and the lead investigator (Malaysia) agrees, they can publish that information concerning the analysis. The French ‘took the 5th’, and obviously decided not to grant publication. It’s entirely within their gift to do so. That is why neither ATSB, nor Malaysia, has shared anything related to the flaperon.
Again, whatever is ‘IYO’ is poorly informed.
@MJ,
The Metron paper primarily described the work to define the specific area where the final, successful, phase of seabed search was undertaken. BEA published 3 interim reports and a final report concerning AF-447, plus other reports dealing with the seabed search as a specific subject.
I understand, without any sight of the ‘contract’, that upon locating a debris field Ocean Infinity’s team will map out the debris field area in preparation for a later recovery effort. The objective would be to scan the debris field at maximum resolution with side-scan sonar and/or HD electro-optical imaging so that key items ma be identified for recovery.
@Don Thompson, @ALSM: Let me join others in agreeing that you make a very persuasive case that the Inmarsat data has been very well investigated and there is no reason to believe the data has been manipulated. Some of the subtleties of the data have only recently been revealed, including characteristics that were unknown to Inmarsat before the disappearance. To believe that the data was altered by a malicious party, and to believe that party had an understanding of the data before the disappearance that exceeded Inmarsat’s understanding, and to believe that party could have correctly predicted that Inmarsat would eventually correctly interpret the data, is simply not a reasonable theory.
@Rob
Please tell your Aunt with Schadenfreude, that according to Heisenberg’s Uncertainty Principle, MH370 is more likely where it is, than elsewhere.
For those of you wondering why the French do not release the flaperon report it is because the French are treating it as a murder case.
Anything related to mechanical failure or some aircraft related problems would be released by the French authorities. Just like AF447.
@Don Thompson
Here you go again..
That was not my point. Whatever reason by whom does not matter.
What matters is the details of that report were never published.
And the I wonder why not. Is it because;
“unless the appropriate authority for the administration of justice in that State determines that their disclosure outweighs the adverse domestic and international impact such action may have on that or any future investigations”
What could be so important about the flaperon investigation not to disclose that information?
@IR1907: In January 2017, it was reported by AFP that the French background check of passengers and crew aboard MH370 found no cause for suspicion. So if this criminal case continues, who do the French believe are the suspects?
@Victor : I agree unreservedly with your statement …
“@Don Thompson, @ALSM: Let me join others in agreeing that you make a very persuasive case that the Inmarsat data has been very well investigated and there is no reason to believe the data has been manipulated.”
@Victor
“So if this criminal case continues, who do the French believe are the suspects?“
In my view, the weight of evidence points to Zaharie Shah.
It is common knowledge, that most investigators believe that Zaharie Shah is the culprit.
Means, Motive and Opportunity.
@Ge Rijn
Refer above.
The Annex 13 protocol is well established. The reality is that ATSB, necessarily with Malaysia’s agreement, appears to have considered the non-disclosure protocols of Annex 13 and decided that it’s useful to release information authored within its jurisdiction, and concerning its work that is pertinent to the search. That decision can only be made by, and for the work of, the ATSB.
Overall there never has been any conclusive statement on an impact attitude of the plane or an in-flight cause of the damage and seperation of any piece recovered.
Malaysia in fact reported the most detailed on found debris. But left it by only mentioning mainly tension-induced damage and hardly any compression damage on any pieces they investigated.
They did noy provide any conclusions from this regarding attitude of the plane during damage and seperation of those pieces.
The French did not provide any details at all on the damage to the flaperon (to the public).
Only some conflicting floating-analysis which did not provided conclusive answers too. Rather raised more questions.
Then the ATSB published a report on the outboard flap section. Rather detailed. But still no conclusive answer on how this piece got damaged the way it was and how it seperated. During impact or in-flight? No answer.
There conclusion was it was most probably retracted when it seperated..
Which to me is the same as saying; we don’t know at what point or how it seperated. Even not quite sure it was retracted or not.
Serving as factual information this kind of information has hardly any value to OI’s search effort.
The only and important value that’s in it, is there is no proof at all coming from this debris investigations, that justify constrainst only close to the 7th arc. On the contrary IMO for those analysis (especially Malaysian) suggest no high speed nose down impact did occure regarding only tension-induced damage, trailing edge damage and the consistent amount of pieces recovered that would specifically seperate during a ditch-like impact.
IMO this is very important to OI to calculate within their search-effort.
I hope they’ve got all the information on the flaperon from Malaysia or the ATSB and all other information we are not aware of.
@Don Thompson
I must say I now regret using the words “my intuition” earlier in this thread.
OI’s gotta be close by now.
@All: Seabed Constructor should reach the southernmost portion of the search area by 11:00 GMT Sunday. That means, if all goes well, by Friday morning we should know if the wreckage is within the 5,000 sq km area that CSIRO designated as most likely. Thanks to @Don Thompson and @Richard Cole for keeping us posted.
Back in August, David Griffin said: We think it is possible to identify a most likely location of the aircraft, with unprecedented precision and certainty. This location is 35.6 [degrees south], 92.8 [degrees east].
We no longer have a need to speculate. We will soon know whether or not CSIRO’s predictions will bear fruit.
Translation of Ge Rijn’s last
“Because neither the ATSB nor Malaysia, in their debris analyses, made any conclusion for an impact scenario, even though it would be wholly improper at this stage for an investigating team to do so, the absence of any conclusion reinforces Ge Rijn’s opinion for the dynamics of the final impact.”
Oh well, if Larry Vance publishes a story at least he’ll get one positive review.
According to this link (BEA France) Mh370s disappearance is classed as an accident:
https://www.bea.aero/fr/les-enquetes/les-evenements-notifies/detail/event/disparition-en-mer/
Although:
“The BEA is the French civil aviation safety investigation authority. Its investigations are aimed solely at improving aviation safety and are not intended to identify fault or liability”
Although:
“The BEA is the French civil aviation safety investigation authority. Its investigations are aimed solely at improving aviation safety and are not intended to identify fault or liability”
So on what grounds would the French prosecutors use to classify the “accident” instead as 1 with criminal intent?
@Michael John
Annex 13 SARPs define only two terms: accident and incident.
I expect parties outside the safety investigation can use whatever noun they care to, to describe the occurrence.
@Victor
Four French nationals have been kidnapped and murdered by a Malaysian pilot. Ofcourse the French are not going to make their analysis public due to high sensitivity.
@Victor
Many countries and parties have carried out background checks and came up with interesting results. What you are reading in the public is for political consumption.
@Michael John
I would not read much into it. Technically they can not use any other word.
@IR1907: I also have little faith in press reports. I have no idea what the French criminal investigators are pursuing. I also have no faith that the investigation won’t get mired in geopolitics.
However, if you have knowledge about the criminal investigation that is being conducted in France or elsewhere, please share it.
On 20th January at 21:29 UTC, Seabed Constructor was at 36.3737°S 87.8481°E, 4,545 km into the journey from the last stop and with 281 km to go. The average course is 88.7277°T and the average speed 13.45 knots.
Seabed Constructor is expected to arrive at 36.3°S near the 7th Arc on 21st January at around 09:00 UTC, which is 16:00 Local Time. Sunset is at 20:13 Local Time.
@Richard
re “Please tell your Aunt with Schadenfreude, that according to Heisenberg’s Uncertainty Principle, MH370 is more likely where it is, than elsewhere.”
Than you Richard, I’ll tell her the next time she rings me. I’ll tell her the best brains from around the world are on the case, and we won’t rest until we find it where we should have found it all along. Just as Heisenberg predicted. She won’t be best pleased.
“Le 7 mai 2014, le parquet de Paris a ouvert une information judiciaire pour “détournement d’avion aggravé par la mort d’une ou plusieurs personnes”, puis une seconde pour “détournement d’aéronef en relation avec une entreprise terroriste”
http://french.xinhuanet.com/2015-08/06/c_134485079.htm
So, the French judicial investigation is about “hijacking of an aircraft aggravated by the deaths of one or more persons” and “hijacking of an aircraft in relation to terrorist undertaking”.
Long ago, the Australian Senate asked the ATSB to respond to criticisms that they had (according to some independent experts and other bidders) not selected the “technically best” search contractor.
The then Chief Commissioner of the ATSB, Mr Martin Dolan, told the senators that he had selected the “best value for money for the Commonwealth” tenderer which met the ATSB’s “Statement of Requirements”.
With that “technically best” issue never fully or definitively resolved in the public domain, it may be considered extant.
The ATSB’s degree of confidence that the Fugro search had not “missed” the debris field is not verified to this day. An opportunity, to possibly do so, is at hand.
Looking at Richard Cole’s latest graphic, and considering that there is now BOTH such a “high condidence” that the debris field must be between 33 and 36 south, AND that “it must be VERY close to the arc, one is forced to consider whether or not it REALLY IS in the area already scanned by Fugro.
Thus, a sensible initial strategy (from OI’s point of view) might be to scan not just the two Orange Primary Search Area(s) of 5,000 square kilometeres, but to also RESCAN the area between them, previously scanned by Fugro.
If nothing else, it would enable a direct comparison between the actual operational efectivenes of both technologies over the same ground, but it may also be interesting from a contractual point of view.
Under the terms of the contract OI have with Malaysia (which we are not privy to) I wonder if there is a clause dealing with the situation where OI might find the debris field in an area already searched by Fugro ?
Perhaps there is a special “reward” payment / provision, for such an outcome ?
@Ge Rijn. Re the flaperon report’s release you said, “This all is not public and thus cannot be independently checked.
Why is the ATSB not sharing this information with the community?
Anyway, I hope they share(d) it with OI.”
On the second bit, I quoted their reason. Like commercial information if the owner is unwilling to release it…..
Also, the ATSB is not the investigator.
On the first, IMO the notion that the public should be able, independently, to check progress of an investigation, or a sub-investigation like wreckage location, surely that would conflict with the efficient use of specialist investigation resources? Continuing explanations and answers to many well- meaning but ill-informed questions and to those with entrenched positions, and ratbags, would divert effort from the investigation.
This is as distinct from selected members of the public being invited to assist, if necessary under non-disclosure agreements.
Other than keeping the public generally informed I doubt that there is or should be any obligation to share technical information during an investigation except via factual information and interim reports. Where would the demand stop? Take a look at the continuing questioning of the flaperon’s authenticity after a reasonable basis for that assessment was disclosed.
If the public had access to the French investigation details do you think that would satisfy all, rather than beg more questions? Does not the release of those reports to the investigators not meet the French obligations, like disclosure of commercial information by such as Boeing?
On the last bit I think it likely that all information which would help OI with its search was and is being disclosed, under non-disclosure agreement if needed, that being in the interests of the investigation.
@ventus45: That’s not going to happen, nor should it. The probability of detection was greater than 95%. Considering the likely size of the debris field, my guess it was better than 95%. At this point, OI has a better chance of finding the debris field by moving northeast along the 7th arc than re-scanning areas, regardless of the “confidence” expressed that the wreckage is in the 5,000 sq km.
@All: In a previous post, I thanked @Don Thompson and @Richard Cole for keeping us posted on the progress of Seabed Constructor. In fact, @Richard (Godfrey) has graciously been providing comments on this blog regarding the progress of Seabed Constructor. Thank you, Richard.
@Ge Rijn
The ATSB has no authority to release details of the French analysis of the flaperon. It’s that simple.
@David
Absolutely – well said.
Is there an efficient way to test drift models? Turns out, yes. When an underwater volcano explodes, it emits a large amount of pumice that floats to the surface and might act like floating bits of airplane debris. And, there’s a lot of it. Here’s an article describe the eruption of a volcano on the Havre Seamount in 2012, about 400 nm North of New Zealand:
http://www.sciencealert.com/almost-nobody-noticed-largest-underwater-volcano-eruption-ever-recorded-havre-seamount
Almost 2 years later, about when MH370 disappeared, bits of pumice washed up on the shores of Australia.
Here’s an article that compares the observed drift of pumice rafts from satellite images with predictions made using ARIANE, which is the same modeling software used in the Geomar study:
http://www.nature.com/articles/ncomms4660
The authors seemed to think that it worked well, but did not run the simulation past 180 days.
@ventus45
“technically best” means the bid that scored, by objective measures, highest against the requirement. That was Fugro. Unsuccessful bidders did whinge (and I don’t think they were ‘poms’): DOS out of the Seychelles and Williamson & Associates of Seattle.
OI’s search strategy may not be immediately apparent as Constructor doesn’t keep station over an AUV during the AUV’s mission (there are, of course, 8 operating simultaneously) but it may well become evident.
@sk999
Thanks for the Havre Seamount paper. I found it to be a difficult read. Maybe I am just having a bad day or the ravages of age are taking a predictable toll.
The next few weeks should be very interesting. For some reason I am feeling apprehensive (scared). How weird is that? I don’t have any skin in the game.
@David
Indeed well said and agreed upon mostly. I know the limitations about publishing information and all kinds of hindering effects it can have if you do this in public. Nonetheless not sharing any details about such an important matter (in this case the flaperon) gives more reason to speculate and less change to more complete information IMO.
Malaysia published a report on the debris they investigated.
The ATSB did the same on the outboard flap section.
Not very detailed reports but far better then nothing.
If the French or Malaysia in their name had published a report on the same level on the flaperon a lot of speculation could have been avoided and more information gained with help of the community IMO.
Now some people keep wondering why no (damage)information has been made public at all only on the flaperon.
But I see it’s a bit of a non-issue now. I also think OI has gotten that information and that’s what counts now.
@Ge Rijn
But I see it’s a bit of a non-issue now. I also think OI has gotten that information and that’s what counts now.
I don’t know that OI has the resources to interpret the additional information. This is not trivial stuff. Maybe they do. Maybe they have the IG under contract. I don’t know. However, saying that OI has this or that is meaningless.
@Rob
@Richard
Boys, it’s all in good fun I know and while quantum physics is a long, long way back in the rear vision mirror for me, I’m pretty sure that you’re butchering poor old Werner’s Uncertainty Principle. Thankfully, it has near zero relevance here; if it did it would render the fundamentals of path modelling from the satellite data, viz the simultaneous resolution of BTO and BFO to yield both position and velocity vectors, impossible.
As to MH370 being more likely where it is than elsewhere, that’s more Copenhagen Interpretation (part Heisenberg to be sure) than Uncertainty Principle.
I’d hate for Rob’s aunt to think that “the best brains from around the world” have got Werner wrong.
@Mick
Heisenberg, Godel, and Chomsky, walked into a bar…
https://www.johndcook.com/blog/2013/12/03/heisenberg-godel-and-chomsky-walk-into-a-bar/
@DennisW
Very good, Dennis, although I prefer the Schrödinger, Gödel and Chomsky version posted in the comments.
On 21st January at 07:38 UTC, Seabed Constructor was at 36.3262°S 90.6739°E, 4,978 km into the journey from the last stop and with 28 km to go. The average course is 88.7318°T and the average speed 13.45 knots.
Seabed Constructor is expected to arrive at 36.3°S near the 7th Arc on 21st January at around 08:45 UTC, which is 15:45 Local Time. Sunset is at 20:13 Local Time.
On 21st January at 07:38 UTC, Seabed Constructor was at 36.3262°S 90.6739°E, 4,978 km into the journey from the last stop and with 28 km to go. The average course is 88.7318°T and the average speed 13.45 knots.
Seabed Constructor is expected to arrive at 36.3°S near the 7th Arc on 21st January at around 08:45 UTC, which is 15:45 Local Time. Sunset is at 20:13 Local Time.
@Mick Gilbert
Thanks Mick. I agree, It’s all good fun and don’t worry, I’ll go easy on the old girl.
Seabed Constructor has arrived at the 7th Arc at around 36.3°S.
On 21st January at 09:22 UTC, Seabed Constructor was at 36.3168°S 91.1465°E, which is 5 NM to the South East of the 7th Arc at 20,000 feet.
Seabed Constructor has stopped at the 7th Arc at around 36.3°S.
On 21st January at 11:52 UTC, Seabed Constructor was at 36.3092°S 91.7895°E, which is 25 NM outside the 7th Arc at 20,000 feet to the South East of the 7th Arc.
Can this ship fly too?:)
The search is about to start. I wish OI all the luck in the world.
@Ge Rijn
The reference for the 7th Arc has been calculated based on a MH370 altitude of 20,000 feet and co-ordinates supplied on the assumption that the 7th Arc at 20,000 feet would be clamped at sea level.
@DennisW said: Maybe they have the IG under contract.
No. Some of us voluntarily help when we can.
@Richard
Ofcourse I understand. It was just a little joke.
Thanks for your up-to-date information.
Article with Blaine Gibson’s views to date (although well known for a long time).
Interesting to me he also went to Tasmania. No finds there or anywere he searched in Australia. Not much of a suprise but still he went there himself again to check it out once more.
Much respect for his efforts:
http://www.abc.net.au/news/2018-01-21/mh370-blaine-gibson-confident-in-renewed-search-off-wa-coast/9346068
@DennisW
I don’t think it would be meaningless if the scientists onboard O.C. have all the information on the flaperon including the French conclusions.
This is a key piece we know very little about for this information has been hidden from us all.
If OI knows it could be even be decisive on the seach-width and where to look also.
We’ll see.
Ah Blaine Gibson. From an amateur Indianna Jones to expert analyst in a swell 4 years.
I admire the guy genuinely although I think the whole media spotlight has gone a little too much to his head. I just feel he is putting too much false confidence into the families. I read Don Thompson quoting a 70% success rate (An article in AIN) I personally prefer this approach.
[Comments here are closed. Please continue the discussion under the new article.]