Sonar image of the debris field of the ARA San Juan (Armada Argentina).
About one year after its disappearance, one of Ocean Infinity’s (OI’s) autonomous underwater vehicles (AUVs) has detected the remains of the missing Argentine submarine, the ARA San Juan. From Ocean Infinity’s website:
Ocean Infinity, the seabed exploration company, confirms that it has found ARA San Juan, the Argentine Navy submarine which was lost on 15 November 2017.
In the early hours of 17 November, after two months of seabed search, Ocean Infinity located what has now been confirmed as the wreckage of the ARA San Juan. The submarine was found in a ravine in 920m of water, approximately 600 km east of Comodoro Rivadavia in the Atlantic Ocean.
Oliver Plunkett, Ocean Infinity’s CEO, said:
“Our thoughts are with the many families affected by this terrible tragedy. We sincerely hope that locating the resting place of the ARA San Juan will be of some comfort to them at what must be a profoundly difficult time. Furthermore, we hope our work will lead to their questions being answered and lessons learned which help to prevent anything similar from happening again.
We have received a huge amount of help from many parties who we would like to thank. We are particularly grateful to the Argentinian Navy whose constant support and encouragement was invaluable. In addition, the United Kingdom’s Royal Navy, via the UK Ambassador in Buenos Aires, made a very significant contribution. Numerous others, including the US Navy’s Supervisor of Salvage and Diving, have supported us with expert opinion and analysis. Finally, I would like to extend a special thank you to the whole Ocean Infinity team, especially those offshore as well as our project leaders Andy Sherrell and Nick Lambert, who have all worked tirelessly for this result.”
Ocean Infinity used five Autonomous Underwater Vehicles (AUVs) to carry out the search, which was conducted by a team of approximately 60 crew members on board Seabed Constructor. In addition, three officers of the Argentine Navy and four family members of the crew of the ARA San Juan joined Seabed Constructor to observe the search operation.
Ocean Infinity had committed to conduct the search operation for up to sixty days, and to take on the economic risk of the search, only receiving payment if the submarine was found.
Position Estimates from Acoustic Measurements
The debris field is centered at (-45.9499,-59.7730). Meanwhile, after analyzing the hydro-acoustic data from CTBTO sensors that was collected on Nov 15, 2017, the hot-spot was estimated to be centered at (-46.12,-59.69). That’s a distance of only 20 km, which is an impressive demonstration of the capabilities of the hydro-acoustic analysis.
There was a test explosion on Dec 1, 2017, that was used to calibrate and verify the position estimation model. The position error of the test explosion was 37 km. Therefore, the position error for the debris field was well within what was expected based on the test explosion.
An interesting description of the CTBTO’s hydro-acoustic analysis is presented here.
Implications for the Search of MH370
Unfortunately, the acoustics generated by the impact of MH370 on the ocean surface would not propagate along the “deep sound channel” (DSC) the way an underwater acoustic event does, so the impact likely was not detected by CTBTO sensors.
The San Juan’s debris field was detected earlier this month only after an AUV mission was conducted in a deep trench that was previously scanned from a higher altitude in September. The ability to skim about 100 meters above the challenging surface contour allowed the side scan sonar sensors in the AUV to collect data at the proper altitude to achieve the required image resolution. Because of its limited ability to glide above steep slopes, this resolution would have been very difficult to achieve using a “towfish”, which is a cable-towed underwater vehicle that is similarly equipped with side scan sonar sensors, but is much more limited in maneuverability than an AUV.
The initial seabed search for MH370 was conducted by the vessels Fugro Discovery, Fugro Equator, and GO Phoenix, each using a towfish. These vessels scanned about 120,000 square kilometers of seabed using this technology. Based on the failure of OI to detect the San Juan’s debris field until an AUV mission was conducted in a trench, we have to at least consider the possibility that that the debris field for MH370 was passed over in the initial search but was not detected due to challenging terrain.
In the coming weeks, we’ll learn more about what worked and didn’t work in the search for the San Juan. That knowledge should be applied to MH370 to determine if we can confidently eliminate all of the seabed that has been previously searched.
Update on November 18, 2018
The ATSB did provide statistics for estimating the confidence of detecting MH370 in the 120,000 square kilometers that were initially scanned primarily by towfish. Those statistics are presented in the figure below. About 97.4% of the total area had a detection confidence of 95%, based on the ability to detect a debris field with dimensions of 100 meters by 100 meters. About 2.1% of the area had a detection confidence of 70%, mostly due to difficult terrain, environmental conditions, or degraded data. About 0.5% of the area had no data. The aggregate probability to detect MH370 in the 120,000 square kilometers is therefore about 94%. Although this is high, it is not 100%.
Don Thompson and Mike Exner provided helpful comments for this article.
Area coverage statistics for the ATSB seabed search (ATSB).
Enhanced ROV photos and an improved sonar image of ARA San Juan debris are now available here:
https://drive.google.com/file/d/1n8VK4CurKF-OmL8vxQMunnoIfvHeOorA/view?usp=sharing
@Victor
Many thanks for the new post on ARA San Juan.
Seabed Constructor has now left the search area and is underway to Cape Town, South Africa.
Yesterday Ocean Infinity spent 2:20 hours at 45.9533°S 59.7741°W and then came back later to spend a further 5 hours just 375m north at 45.9500°S 59.7731°W. I expect that these were ROV missions to document the wreck.
@Victor
In the link below, the Seabed Constructor movements on 9th-10th November 2018 is marked in red with the locations that the vessel stopped to launch, pick up or listen out for AUVs marked with a date and time.
I have overlaid the AUV search lines on 8th-12th September 2018 marked in orange, with the 2nd AUV sortie highlighted in light grey.
POI24 is the location of the wreckage of the ARA San Juan. The wreckage is in a ravine 920m deep. The surrounding seabed is 741m deep. The ravine is 179m deeper than the surrounding seabed.
The bathymetry is courtesy of Don Thompson.
https://www.dropbox.com/s/7wzs8lvg21md132/ARA%20San%20Juan%20SEP%20and%20NOV%20Coverage.pdf?dl=0
I just read the WSJ article Victor Iannello added to his previous blog article. It raises many issues I had highlighted previously:
———
(1) The automated stall-prevention system on Boeing 737 MAX 8 and MAX 9 models […] can push it [the nose] down unexpectedly and so strongly that flight crews can’t pull it back up. Such a scenario, Boeing told airlines in a world-wide safety bulletin roughly a week after the accident, can result in a steep dive or crash — even if pilots are manually flying the jetliner and don’t expect flight-control computers to kick in.
(2) One FAA manager familiar with the details said the new flight-control systems weren’t highlighted in any training materials or during lengthy discussions between carriers and regulators about phasing in the latest 737 derivatives.
(3) checklists and procedures for the 737 MAX 8 entail more steps and take more time. Investigators and safety experts are convinced that as the emergency worsened, the Lion Air crew had barely seconds in which they could have diagnosed the problem and taken action to save the aircraft. Shortly before the plane crashed, according to local Indonesian media reports, one of the pilots told air-traffic controllers about difficulties controlling the plane.
———
@Andrew:
Item 3 is what I tried to convey when I said the crew has only a very short time to diagnose the problem – a diagnosis which is made brutally difficult by the many different and false alarms (stick shaker without stall) kicking in simultaneously, which has been described in the last paragraph here.
I still don’t understand
– the deviation from the decades-old principle of redundancy (1 malfunctioning AOA sensor is enough to nose-dive) And
– why there is no failsafe to do NOTHING in case of AOA sensor disagreement. That would be the logical and appropriate automation rule.
I guess the idea was to handle AF447-style AOA sensor disagreement.
+ It may help in AF447-style scenarios, but these should be handled by the crew themselves.
– On the other hand, it can artificially CREATE completely unnecessarily a potentially deadly problem.
For me the bottom line of these 2 items is to display AOA disagreement, but not to take automatic corrective measures, particularly in manual mode and especially when it’s not at all clear whether these corrective measures HELP or HURT.
@Peter Norton: I think you are over-simplifying the issue.
the deviation from the decades-old principle of redundancy
There is a difference between “redundancy” and “fault-tolerance”, which might involve automatically identifying the fault using two-of-three voting and reconfiguring the system. That was not how the B737 MAX is designed.
It may help in AF447-style scenarios, but these should be handled by the crew themselves.
– On the other hand, it can artificially CREATE completely unnecessarily a potentially deadly problem.
My guess is that Boeing performed a fault or failure analysis and determined that with one sensor indicating stall and the other not indicating stall, it would be safer (on average) to automatically correct for stall. You think otherwise. I think it is premature to decide that Boeing made the wrong design decision, although it is possible.
Update on November 18, 2018
The ATSB did provide statistics for estimating the confidence of detecting MH370 in the 120,000 square kilometers that were initially scanned primarily by towfish. Those statistics are presented in the figure in the post above.
About 97.4% of the total area had a detection confidence of 95%, based on the ability to detect a debris field with dimensions of 100 meters by 100 meters. About 2.1% of the area had a detection confidence of 70%, mostly due to difficult terrain, environmental conditions, or degraded data. About 0.5% of the area had no data. The aggregate probability to detect MH370 in the 120,000 square kilometers is therefore about 94%. Although this is high, it is not 100%.
@Victor
The recent OI statement, at the end…they say 6000-meters deep…they can go…that was my question. Converting that to feet 19,685-ft.
If I think “outside the box” MH370 could be deeper if the pilot could have been headed for Dordrecht Hole or similar deep spot. Near Arc6 at 22 South there is a 25000-ft deep canyon on Google Earth.
I know @DennnisW is still inside the box.
@Richard
Do you still think OI found the submarine in Sept (pre-discovery hit) or was it a fresh finding just now?
@TBill: Why not aim for the Mariana Trench? It is much deeper (36,000 ft), would have been closer than the estimated impact sites in the SIO along the 7th arc, and there would be no need to fly again over Malaysia or near Indonesia or Thailand?
@Victor
Apparently the perp wanted to overfly Malasia and head to SIO, possibly passsing close to Cocos to infer that was the intent.
P.S.- I feel the same way about 38S as I do the Marianas Trench. If this was a deniable crime (which is my hypothesis) then there would not be a direct flight to the middle of nowhere, because that looks like an intentional suicide. So I am thinking an effort to make this look like a overflight of a rational target.
@TBill: How does turning back and flying past several airports (Kota Bharu, Penang, Langkawi, Banda Aceh) with no descent make this a deniable crime? It has all the makings of a deliberate diversion. On the other hand, flying towards the Mariana Trench would have made the crime very deniable, as it would have been hard to prove the flight path was not caused by hypoxia.
@TBill
You asked “Do you still think OI found the submarine in Sept (pre-discovery hit) or was it a fresh finding just now?”
I still believe OI found the submarine on the 2nd AUV sortie on 10th September 2018 and it was misclassified as a fishing trawler, that had previously been reported as sunk 350m away. This was despite the fact that the target was exactly the dimensions of the submarine.
@Victor
Well my hypothesis has all or most of the answers, whether they are correct answers or not is another debate.
I would say the pilot definitely wanted it to be known that he stole the plane. That’s why he went up to FL430+ at IGARI and turned on the SDU at 18:25. We are expecting a hijacker to give us a message, and we got several, loud and clear and to the whole world. The unanswered Satellite calls sealed the deal. The pilot knew everyone knew he was still up there. Meanwhile the plan may have been to fake a safe landing diversion at Cocos, possibly going as far as a telling some folks that was the plan before the flight (hence @TimR’s story).
The crazy thing was Razak let the world believe the plane crashed, until that apparent lie could no longer hold water.
In the USA, we’d say it is clear what happened from circumstantial evidence. By SilkAir shows a different mentality of guilt in that part of the world, is my rationale.
More info on OI activity surveying the ARA San Juan debris field:
“The company Ocean Infinity, in charge of the ship Seabed Constructor that found the ARA San Juan, will send to the federal Justice and to the national Government 67,000 photos of the submerged sunken to 907 meters of depth.
Those photos in high definition that will be sent from South Africa, will be used to determine the causes of its collapse, which meant the death of the 44 members of its crew.
According to the Infobae website, if there are no technical or legal problems, on Wednesday, the 67,000 HD photos of ARA San Juan will arrive in Buenos Aires.
Five UAV vehicles (Autonomous Underwater Vehicles) and two ROV (Remote Operated Vehicles) took the 67,000 photographs that will be used to build in images a three-dimensional model of the ARA San Juan.
That model -which in the technical jargon is called Render- will allow us to observe how the hull of the submarine was sunk and how it is resting on the bottom of the sea. That information is key to determining the causes of the tragic shipwreck that began on November 15, 2017.
Ocean Infinity will send the photographs to the Ministry of Defense and to Marta Yáñez, federal judge of Caleta Olivia. In both cases, the Houston-based exploration company will submit the 67,000 HD photos with an explanatory dossier and a digital reconstruction of the ARA San Juan, which will allow you to see how your helmet (hull) was with a 360 degree perspective. So far, only three photos of the sunken submarine, corresponding to the bow section, the propeller and the sail, have been distributed.
The HD photographs will be the closest signs of which the Government and the Federal Justice will be able to find out the truth of the facts and adopt possible administrative and penal sanctions. At 900 meters depth it is impossible for an expert to monitor the situation of the hull and even less to get a sample of the sunken submarine.
When Minister Aguad receives the information from Ocean Infinity, a formal presentation will be made to President Mauricio Macri, who demanded investigative depth and speed to find those responsible for the collapse of ARA San Juan. On the other hand, Judge Yañez will analyze the 67,000 photographs together with her direct collaborators.
The magistrate assumed that it is impossible to recover the submarine leaning on a maritime ravine 907 meters deep, and looks forward to sending the images to decipher what happened on November 15, 2017.
Source: Tomorrow”
@All,
Here are my thoughts on lessons learned during Ocean Infinity’s successful search for the ARA San Juan. Items 3-6 are applicable to future searches for MH370.
1. If good acoustic detections are made by multiple CTBTO stations, localization can be very accurate. The method has been validated before. Conducting a calibration explosion nearby as soon as possible can improve the localization accuracy.
2. If good acoustic detections are made by CTBTO, the Friedlander frequency can be measured, leading to an estimate of the implosion depth which can then be used to eliminate any part of the error ellipse that is too shallow for an implosion to occur.
3. Start scanning at the best estimated position and generally work outward. This minimizes the search time required because the most likely places are searched first.
4. Take the time needed to thoroughly investigate difficult terrain before moving on to less likely areas. Maintaining the side scan sonar at an optimum altitude above the seabed is critical to achieve the sensitivity and resolution needed and to avoid large unilluminated areas.
5. As apparently done routinely by Ocean Infinity, investigate points of interest expeditiously using an on-site ROV. Investigate every possible contact thoroughly, and don’t delay investigations of points of interest unnecessarily, or you may be wasting time and money. [Richard Godfrey believes OI may have initially misidentified POI #24 as a sunken trawler. If so, they must not have employed the ROV on it, and the initial side scan sonar image must have been of poor quality.]
6. As I believe Ocean Infinity did, validate the retrieved data with a short delay using dedicated quality assurance people and software. This includes proving detection capability on every AUV mission. It is essential to know that every sonar is always working properly during data collection.
@DrB
You stated “Richard Godfrey believes OI may have initially misidentified POI #24 as a sunken trawler. If so, they must not have employed the ROV on it, and the initial side scan sonar image must have been of poor quality.”
The announcement by the Argentine Navy based on an OI report regarding the POI #2 was made on 14th September 2018.
Here is the excerpt posted previously:
https://www.dropbox.com/s/mip9uuwsvbl0xwx/ARA%20San%20Juan%20Initial%20POI.png?dl=0
Here is the full report:
https://www.dropbox.com/s/3stf97b3mkkywml/informe17.pdf?dl=0
OI had made no ROV deployment before this announcement.
The longest stop between 8th and 14th September 2018, anywhere, was 25 minutes, which is too little time for a ROV deployment.
The POI #2 was discarded as a fishing trawler that was reported to have sunk in the vicinity.
This decision was made without a ROV deployment.
Unfortunately POI #2 had the same dimensions as POI #24 and the ARA San Juan.
Perhaps OI can tell us whether POI #2 was at the same location as POI #24?
@Niels
Below is a link to a .pdf comparing the beaching distribution originating from between 20°S and 25°S and originating between 30°S and 35°S, as you proposed, using my data:
https://www.dropbox.com/s/r4yjbeh436bxazx/Beaching%20Map%20-%2020S%20to%2025S%20vs%2030S%20to%2035S%20vs%20Debris%20Locations%20plus%20Start%20Locations.pdf?dl=0
Below is a link to a .pdf comparing the beaching distribution originating from between 20°S and 25°S and originating between 30°S and 35°S, as you proposed, using David Griffin’s data. There are 9,067 beachings in the Western Indian Ocean originating from between 20°S and 25°S and 2,874 beachings from between 30°S and 35°S:
https://www.dropbox.com/s/hf3zmtfjsamdb74/Beaching%20Map%20-%2020S%20to%2025S%20vs%2030S%20to%2035S%20vs%20Debris%20Locations%20plus%20Start%20Locations%20DG.pdf?dl=0
@All,
David Griffin has updated his blog as follows: Edits to 16 Nov: Fig. 5a, 5b re-drafted, Fig 4b, 5c and tracks added, link to modelled beaching locations.
http://www.marine.csiro.au/~griffin/MH370/index.htm
@Peter Norton
In my view, it is premature to be jumping to conclusions about the cause(s) of the accident. We don’t even know exactly what problems the crew encountered; much of the information in the media is nothing more than speculation at this point. We also know very little about the design of the MCAS system and next to nothing about its fault tolerance. For example, the B777 has only two AOA vanes, just like the 737MAX, but the B777’s redundancy management logic also uses calculated AOA, based on inertial and air data. Does the B737MAX have something similar? I don’t know and I also don’t know the 737MAX system response to an AOA disagreement.
My advice would be to calm down and wait for more information to emerge. A preliminary report should be issued by the end of the month.
@RichardG
What is similar in your and David’s results is that the distribution of predicted beaching locations on mainland Africa does not strongly depend on starting latitude (S20-S25 vs. S30 – S35). It is therefore hard to use this distribution to discrimate between the two regions, by comparison with the distribution of found and reported items.
What is different is that the distributions you calculated seem to better match the distribution of find locations than the distributions calculated by David Griffin. This could mean that David’s simulations have a northerly bias; it could also mean that both your simulations and the distribution of find locations have a southerly bias (before we agreed on the possibility that a lot of items in the north have not been found or reported). At this moment I would not know what is more likely. I can only recommend both you and David to better validate your models. In your case the main question probably is what error you introduce by using historic, averaged drifter data; in case of David Griffin the main point is probably to check the “de-biasing” procedure used to compensate for systematic errors in the BRAN2015 oceanic circulation model. It would be best to “fix” the BRAN model and get rid of the need for de-biasing completely, as was more or less announced in the first CSIRO report.
@Andrew
Good post. However, speculation is fun and does little harm. My guess is that Boeing is not going to walk away from this one without some scars.
@TBill
You said:
I know @DennnisW is still inside the box.
Yes, I am. Tomorrow I have a doctor appointment in Santa Rosa. I am going to try to get there with zero gas remaining to simulate your scenarios. Maybe I will come away with renewed appreciation for your views.
@TBill,
I am comfortable with your whacko characterization. Good grief. Try to move things forward. Sure, there is a finite probability that the wreckage was missed. So what?? What makes the most sense to do next?
@DennisW,
You said: “What makes the most sense to do next?”
I think eliminating the known uncertain areas in the previous searches should be on the list to be considered.
@Victor,
Any inside knowledge on What OI’s planning to do next? Are they considering to return to the MH370 search zone? Thanks.
re: Drift Analysis ‘Discovery’
Even there is no definitive answer on where the plane landed based on drift analysis; it is quite certain that it landed at an almost optimal place – in terms of maximising “beaching” time. At least now I understand (possibly) why the plane landed so close to Australia (relatively) but not “in the middle of nowhere” – say ‘centre’ of SIO.
So may I take this opportunity to [again] suggest (the ‘crazy’ [“more professional/sophisticated” operation] idea – enhanced version) that perhaps there are also experts (say oceanographers) included in this disappearence’s “secret mission team” (if it exists at all) …
[Just for the sake of the (“optimal”) suggestion/idea – no other intention.]
@haxi: I don’t know OI’s plans regarding another search for MH370.
News websites are carrying a video clip of the scene in the ROV pilots’ station onboard Seabed Constructor as the ARA San Juan wreck site is investigated.
A best quality copy of the clip is archived at Vimeo.
@DrB
I think eliminating the known uncertain areas in the previous searches should be on the list to be considered.
Yes, I agree that should be considered. I’m not sure how to prioritize that activity. It would seem that scanning new areas would have a higher return on a probability of a find per unit time basis. Maybe at the end of the search from 25S to 20S if nothing is found?
@Victor
I hate to state the obvious but perhaps it is important to note:
How Long it Takes to Find Something
(1) ARA San Juan- Knowing semi-exact position – 1-yr to find
(2) AF447 – Knowing approx crash location – 2 yrs to find
(3) MH370 – Knowing only satellite Arc – ?? yrs to find
@TBill
I don’t understand what you’re trying to say, above.
However, the San Juan search was initially undertaken with unsuitable equipment: MBES and ROVs. Active search duration, with appropriate equipment, was in the order of 60 days.
IIRC, the underwater search for AF447 was conducted over 88 active, at-sea, days.
Both scenarios involved good fixes for an LKP, the CTBTO-IMS work for San Juan, and a position reports generated within 10 minutes prior to impact.
MH370 involves position fix in only one dimension, a line of position with a fixed constraint to the south based on fuel and no emphatic, single, constraint to the north. So, on one hand there is a long process of elimination that’s called a search.
Take a long moment before any reply including the word ‘glide’.
@Don
I am just saying the MH370 search is so much harder, and that lack of finding MH370 to date is understandable, and unfortunately we cannot draw too many conclusions– about location — from lack of finding MH370 so far.
@TBill
You stated “unfortunately we cannot draw too many conclusions– about location — from lack of finding MH370 so far.”
Nonsense! We can conclude to 97%, that MH370 was not between 25.0°S and 39.5°S.
25.0°S to 39.5°S has been searched at least to a search width of ± 20 NM (and in parts considerably wider), the remaining area from 23.0°S to 25.0°S should be searched at least to a search width of ± 20 NM.
As @Dennis keeps pointing out, there has never been a better probability of finding MH370.
@Don Thompson
You stated to @TBill “Take a long moment before any reply including the word ‘glide’.”
Or even … “dive – glide – dive”.
Laughing out Loud!
@Richard
Good I support searching where you want to search. I will be happy if you are correct.
@TBill
I would like OI to search from 25°S to 23°S. Unfortunately, I think they/we missed MH370 on the last round by a small margin. It was my fault, I got it wrong back then.
I do not want to be correct, I want MH370 found.
@Richard Godfrey
There are only 3% probability that all the search vessels didn’t miss MH 370 on the ground between 39,5 and 25 degrees? After you stated that SC found the San Juan earlier in September but didn’t recognize her as a submarine, I think I don’t understand your statement.
@Gerald
That is precisely the conundrum that needs to be solved.
Did OI miss or misclassify an early sighting of ARA San Juan?
If so, did Fugro, Go Phoenix or OI miss or misclassify MH370.
@Richard
Did SC search the complet 120000qkm already done by Fugro and Go Phoenix or did they just check the 25000qkm including the CSIRO hot spots?
Ok, after that OI went north up to 25S.
My proposal for a new search area would be starting from 25S to 20S because of your drift studies, even in 2016 Geomar Institute did a study which showed lots of northern probabilities for the impact, although this study was only based on the flaperon find.
Then ,in spite of the more southern areas, I would let SC check the CSIOR hot spots again.
Before a new search could start, OI defenitely will check their data on the old search, maybe they missed the plane.
@Gerald
Ocean Infinity searched a further 120,000 sq km in addition to the ATSB/Fugro/Go Phoenix 120,000 sq km.
The Ocean Infinity search included the remaining CSIRO search area, resulting from the ATSB First Principles Review.
Personally, I am only asking for a further search of 25°S northwards to 23°S with a search width of ±20 NM.
The GEOMAR study was a reverse drift analysis, which is flawed in my opinion.
Depending on the outcome of the technical analysis from Ocean Infinity of the ARA San Juan search, I think an independent audit of the search results from Fugro, Go Phoenix and Ocean Infinity might be a wise move before further money is spent on any search.
@Richard Godfrey
Richard, you had previously said ‘Out of a total of 190 simulator runs, 78 beachings were on mainland Africa, 73 beachings on Madagascar and 38 simulator runs ended up in mid-ocean gyres or heading out into the Southern Atlantic Ocean having rounded South Africa.‘
Did that beaching rate of just shy of 80 per cent surprise you? I thought that Henrik Rydberg had put the estimated beaching rate about an order of magnitude lower in his earlier paper.
re: Fuel ‘Wasting’
Another possible reason for fuel wasting [if so] is to maximise ‘night’ time flying, minimise ‘day’ time flying or optimise time of landing. If so, then it is more likely that it is a determined/deterministic flight from the very ‘beginning’. That is, no contingency, no plan A/plan B: implying no negotiation – no need for further communication – no practical need for SATCOM logon. Therefore, it is more likely a ‘straight-forward’ flight; more likely a “path of the simplest descent” flight after the FMT – more likely to go further remote – which should/would be further ‘south’ (within constraints).
BTW (sorry), one also need to re-examine the ‘real’ reason for the SATCOM logon (if ‘deterministic’ flight can be established).
Boeing returns fire:
Boeing CEO disputes MCAS details “intentionally withheld” from airlines
@Andrew
Boeing returns fire:
With the share price down 13 per cent since the MCAS story broke, Muilenburg is probably feeling the need to do something. Holding up a copy of the MAX AFM open to the section entitled ‘Maneuvering Characteristics Augmentation System‘ would settle the issue. I wonder why he hasn’t done that.
@mash wrote “one also need to re-examine the ‘real’ reason for the SATCOM logon”
Perhaps a slightly different question: why restore power to the L Main AC Bus, an action that incidentally caused the AES to establish a new Log On session?
@Mick
You asked “Did that beaching rate of just shy of 80 per cent surprise you?”
No! But now that you mention it, I started an analysis of my database.
I selected 280 undrogued transoceanic drifters from the year 2000 onwards.
The transoceanic drifters mostly ended by beaching somewhere. A few went round South Africa to the Atlantic, even fewer started back to Australia in the roaring forties.
I did not select drogued drifters, drifters prior to 2000 or drifters that stopped transmitting mid ocean. A clear case of selection bias.
@Niels, @Mick,
In the link below is a comparison of 240 drifters, that beached in the Western Indian Ocean between 1993 and 2018, compared with 159 simulations that beached in the same area, from a simulation database between 2000 and 2018:
https://www.dropbox.com/s/f67yw22d4wbiwvi/Beaching%20Map%20-%20240%20Drifters%20vs%20159%20Simulations.pdf?dl=0
@Richard Godfrey,
Do you understand why, for the very interesting 1993/2000-2018 comparison you just posted, your model:
(a) somewhat overpredicts in South Africa,
(b) somewhat underpredicts in northern Africa,
(c) has no beachings at all on the islands generally north of Madagascar?
Would the same hold true for the time period you actually used for making MH370 debris drift predictions?
@All,
Here is a new paper analyzing the MH370 drift model predictions:
https://drive.google.com/file/d/1-WWg0Pkf8XeF6fHHgqvdCM4vNLbgFFmc/view?usp=sharing
It presents my findings:
(a) Crash latitudes north of 20S are highly unlikely because there are either far too few beachings or they are far too early in arrival.
(b) Crash latitudes from 20S-23S are possible but unlikely because they generally arrive too early to be consistent with the Debris Reports even with long assumed reporting delays.
(c) Crash latitudes from 33S -35S are also unlikely because they arrive at or slightly later than the reporting dates, leaving no allowance for expected reporting delays.
(d) Crash latitudes from 23S-33S are likely.
(e) There is a slight, but not definitive, preference for 30S-33S over 23-25S, due to a noticeably better match for the Flaperon case and marginally better matches for several of the western zones (1,2, and 4).
This paper also presents my search recommendations:
“In my opinion, if a new search for MH370 were to be undertaken, these areas deserve consideration:
1) the unscanned and sub-optimally scanned areas within the previous search zones,
a) doing 25.0S-35.0S as a first priority, and
b) doing 35.0S-39.5S as a last priority, only after all other options have failed,
2) from 20.0S-25.0S within ±22 NM of the 7th Arc (and doing 23.0S-25.0S first),
3) from 30.0S-33.0S extending outward to ±75 NM, and
4) the unsearched remainder of 20.0S-35.0S out to ±75 NM.
The order of this list represents my current recommended prioritization.”
@DrB
The answer to (a) and (b) is that I did not select drogued drifters, drifters prior to 2000 or drifters that stopped transmitting mid ocean, when creating my simulation database.
The answer to (c) is that I have mapped the coast of Africa, Madagascar, Reunion, Mauritius and Australia in my software, but not the minor islands Rodrigues, Comoro Islands, Mayotte, Glorioso Islands or Aldabra.
You asked “Would the same hold true for the time period you actually used for making MH370 debris drift predictions?” For (a) and (b) no, the database used in the simulation is based on 280 undrogued transoceanic drifters from the year 2000 onwards. For (c) yes.
@RichardG
What would the beaching map look like if you only plot the beaching locations of undrogued drifters after 2000?
It is important to know if the notherly bias for beaching location of real drifters in this comparison with respect to your model is caused by the selection of types of drifters, or by the distribution of starting locations (which for the 240 real drifters is not bound to 7th arc I suppose?)
@RichardG
Another question: from these 280 drifters used to build your model, how many were close to the 7th arc between 20S and 40S (let’s say within 100 km) at a certain point in time?
@Mick Gilbert
RE: “Holding up a copy of the MAX AFM open to the section entitled ‘Maneuvering Characteristics Augmentation System‘ would settle the issue. I wonder why he hasn’t done that.”
It’s hard to know what Muilenburg actually said when media outlets cherry-pick comments that don’t tell the whole story. The Air Current article implies that Muilenburg said “That’s simply untrue” in response to claims that Boeing withheld information airlines. The article also claims that Muilenburg said ‘The relevant function [of MCAS] is described in the Flight Crew Operations Manual…’.
However, the following article from The Seattle Times imparts a different spin when it states that Muilenburg “…specifically denied reports in some media outlets that the procedure pilots need to deal with such uncommanded movements was not in the 737 pilot manual and that pilots were not trained on how to handle it.
‘That’s simply untrue,’ Muilenburg said.”
Boeing to hold regional meetings with airlines on 737 MAX, while execs try to reassure employees
@Don Thompson
“… why restore power to the L Main AC Bus, an action that incidentally caused the AES to establish a new Log On session?”
Why remove power [evidence, if any …] in the first place? (*** In fact, I am not too familiar with this “AC Bus” part.) It seems to me that the [main] ‘reason’ for that [presumably long] electrical power off/on is to explain the “Log On”? Why can’t the “Log On” be an ‘independent variable’? – say something done specifically to address the critical [communication [or even survival]] ‘problem’ …
*** But as an analogy, one does not need to understand/play around with main AC power supply to diagnose a ‘home-based’ PC external modem (data/power) connectivity problem (although that might be the [unlikely] cause).
@Andrew
Boeing has a $180B market cap and can afford to manipulate the PR any way it wants to with both in person and written press releases. You have to know they are shaping things with an appropriate mixture of truth and vagueness. We are dealing with professionals here. Not dumb shit journalists.
The reality is they are in serious difficulty and are deliberately filling the room with smoke.
@DennisW
Even so, Muilenburg would be a fool to tell employees that MCAS is described in the FCOM (as alleged by The Air Current) when the major US pilots’ unions have claimed they had no idea about the system and that it was not described in the manuals. My guess is that The Air Current got it wrong and that The Seattle Times version is more accurate. That is, Muilenburg ‘specifically denied reports in some media outlets that the procedure pilots need to deal with such uncommanded movements was not in the 737 pilot manual and that pilots were not trained on how to handle it’. That statement is true, but it only tells part of the story and glosses over the lack of information provided to the airlines and their pilots.
@Andrew
@DennisW
‘Boeing has a $180B market cap …‘
Boeing had a market cap of $211B the day the FAA released its Emergency Airworthiness Directive on 7 November. Their share price plot for the past fortnight looks a bit like JT610’s flight path over its last 30 seconds, the difference being that Boeing has a lot more than a comparative 5,000 feet of altitude to play with.
The loss of $31B in value over two weeks has galvanised the Board’s attention I would think. The long and short of it is that their handling of this has been appalling; they appear to be using the Malaysian government playbook – A Beginner’s Guide to Ballsing Things Up in Public. The notion that someone in their organisation thought that it would be a good idea to hold a global teleconference with all MAX operators is simply astounding. It’s difficult to manage the agenda and the message when you’re dealing with two or three customers face to face and some bright spark thought that they could do it with 30-odd customers on the phone! Think angry town hall meeting, take away the security and now turn the lights off. That little misstep alone shaved nearly $2B off their market value.
You can bet their new plan for their Field Service Reps to hold local briefings will be closed shop affairs – authorised attendees only, no recording devices, no press. Expect to see all MAX’s now fitted with AoA displays and disagree cautions as standard, no extra charge.
And yes, Andrew, you’re dead right, it’s difficult to know what Muilenburg actually said. I don’t think that he said that MCAS was documented in the FCOM. I suspect that he repeated what was in their FCOM Bulletin, that the recovery procedure is documented. To me that’s a bit like a car manufacturer, after one of their cars careens off a cliff, saying that yes, we did fit a pedestrian avoidance feature that we didn’t tell you about, and yes, under some circumstances it might get bad data and just turn sharply for no reason, but hello! steering wheel, people, it’s right there in front of you.
It is a testament to Boeing’s employee NDAs that the full contents of his internal email hasn’t leaked yet.
@Niels
You asked:
(1) “What would the beaching map look like if you only plot the beaching locations of undrogued drifters after 2000?”
(2) “What is the distribution of starting locations (which for the 240 real drifters is not bound to 7th arc I suppose?)”
(3) “From these 280 drifters used to build your model, how many were close to the 7th arc between 20S and 40S (let’s say within 100 km) at a certain point in time?”
There are currently 286 undrogued drifters in my database. The coverage is shown in the link below from 20°N to 50°S, 0°E to 160°E. The coverage comprises 4.4M km of track. My goal was to achieve an extensive coverage of the SIO:
https://www.dropbox.com/s/pjvth4c2c7tbgkd/286%20Undrogued%20Drifters%20Coverage%2020N%20to%2050S%2C%200E%20to%20160E.png?dl=0
Out of the 286 undrogued drifters, 204 were transoceanic and started between 0°S and 50°S, 20°E to 160°E as shown below:
https://www.dropbox.com/s/gvhoysfcl8traew/204%20Undrogued%20Drifters%20Start%20Locations%200S%20to%2050S%2C%2020E%20to%20100E.png?dl=0
The end locations of these 204 undrogued transoceanic drifters is shown below:
https://www.dropbox.com/s/0irq59kv24awfnq/204%20Undrogued%20Drifters%20End%20Locations%20from%200S%20to%2050S%2C%2020E%20to%20100E.png?dl=0
158 undrogued drifters passed through the 7th Arc area from 10°S to 40°S ±5° of longitude (there are actually 187 passes as some undrogued drifters pass the 7th Arc multiple times).
The start distribution of the 240 drogued and undrogued drifters previously shown is not bound to 7th arc.
Out of the 204 undrogued drifters (discounting those still in transit, picked up, stopped transmitting or batteries died), there were only 35 confirmed beachings on the coast of Africa or nearby islands between 10°N and 40°S, 20°E to 70°E:
https://www.dropbox.com/s/1155zixa6bppbe0/35%20Undrogued%20Drifters%20End%20Locations%20Confirmed%20Beached%2010N%20to%2040S%2C%2020E%20to%2070E.png?dl=0
@Mick Gilbert: Without trying to apportion blame, the drop in market cap is the market’s overreaction to Boeing’s liability and loss of goodwill. Boeing’s order book is strong. Its pipeline of new products and services is strong. Whatever deficiencies in the documentation, training, and functionality of MCAS in the 737 MAX will quickly be remedied. My unprofessional observation is the drop in Boeing share price could be a buying opportunity.
@Victor Iannello
Yes, you may well be right, Victor, regarding both the overreaction and the buying opportunity. At the end of the day, the market is the final arbiter of value. You can bet that Airbus will be printing a few extra boxes of neo brochures.
@Mick
In the same timeframe as Boeing’s loss of $31B (-15.6%) since 7th Nov 2018, Airbus has lost $4.9B (-5.9%).
The market is not just against Boeing.
@DrB
You stated: “(2) from 20.0S-25.0S within ±22 NM of the 7th Arc (and doing 23.0S-25.0S first)”
I would change the priorities: “(1) from 23.0S-25.0S within ±22 NM of the 7th Arc.”
Re: Boeing Stock
Boeing stock started the year at about 290, so it is hard to make too much of the current 317. I was going to say something about the stock price earlier last week but I decided without doing some correlations to other stocks going down in the current correction, it is hard to say much.
@RichardG
Thank you, interesting. So it seems drogued vs. undrogued has influence on the beachings location distribution. Can we understand that / why would drogued drifters tend to go more to the north in the area that we are looking at?
re:JT610
> Andrew: “I also don’t know the 737MAX system response to an AOA disagreement.”
Don’t we know that based on the EAD stating:
« This emergency AD was prompted by analysis performed by the manufacturer showing that if an erroneously high single angle of attack (AOA) sensor input is received by the flight control system, there is a potential for repeated nose-down trim commands of the horizontal stabilizer. This condition, if not addressed, could cause the flight crew to have difficulty controlling the airplane, and lead to excessive nose-down attitude, significant altitude loss, and possible impact with terrain. »
> Andrew: “I doubt that rectification will include significant changes to the system that
> would require re-certification. I suspect the ‘fix’ will be relegated to pilot training,
> better documentation and possibly a requirement for the mandatory fitment of the
> AOA indicator.”
I agree that we don’t have the full picture yet, but as this might take quite some time:
Based on the above, why shouldn’t the “fix” be to do NOTHING in case of “an erroneously high single AOA sensor input” in manual flight rather than “repeated nose-down trim commands of the horizontal stabilizer”? Wouldn’t you consider this safer? In my view, anything else would violate the fail-safe paradigm as it could create a critical problem that wasn’t there in the first place. In my view, this is incompatible with a fail-safe mechanism.
> Victor Iannello: “There is a difference between “redundancy” and “fault-tolerance”,
> which might involve automatically identifying the fault using two-of-three voting
> and reconfiguring the system. That was not how the B737 MAX is designed.”
@Victor Iannello: I am unsure what you are intending to say, because this was exactly my point. Depending on the case, redundancy is one possible means to achieve the goal of fault-tolerance. In this case, according to the EAD, 1 of 2 erroneous AOA sensor readings can lead to “excessive nose-down attitude” in manual flight. In my view, more redundancy would be required (i.e. at least 2 agreeing sensor readings) or/and a fail-safe response to do nothing except signaling an AOA DISAGREE alert. It seems there is neither. I agree that we need more information, but this is my interpretation of the picture we have at the moment.
@Peter Norton: You keep saying that if there is AoA disagreement in detecting excessive AoA, the “fail-safe response” is to do nothing other than flagging the AoA disagreement. I don’t know that this is true. I would be extremely surprised if the Boeing system designers did not do a thorough failure analysis and concluded that it would be better (on average) to automatically trim down with the assumption that the pilots would identify and properly recover if that automatic response created a runaway trim condition.
@Niels
You asked “Can we understand that / why would drogued drifters tend to go more to the north in the area that we are looking at?”
63% of all drifters in the Southern Indian Ocean lose their drogue. 72% of drifters confirmed as beaching have lost their drogue, many in the course of beaching where the drogue catches on the sea bed or rocks in the shoals.
The link below compares the 78 drogued drifters with the 29 (not 35 as stated earlier) undrogued drifters that were confirmed as beached in the Western Indian Ocean in the years 2000 to 2018:
https://www.dropbox.com/s/qaulzgn4y7ueznu/Beaching%20Map%20-%20Drifters%20-%20Drogued%20vs%20Undrogued%20-%20from%207th%20Arc.pdf?dl=0
The reason that more drogued drifters beach than undrogued drifters is that the drogue gets caught up in the seabed, rocks or reefs and the drifter does not get swept back out to sea.
One possible reason that more drogued drifters beach in the north is that there are strong on-shore winds all year round in the north (at 0°S, 43.5°E the 2015 average was 14 knots, 11 months on-shore), whereas in the south there are variable winds cross-shore (at 30°S, 31.4°E the 2015 average was 14 knots, 10 months cross-shore). Another reason is the tidal range in the north (Mogadishu, Somalia, max. tidal range = 2.98m) is also on average higher than the tidal range in the south (Durban, South Africa, max. tidal range = 2.19m). So once a drogued drifter gets blown ashore by the wind or lifted ashore by the tide or both, it gets stuck because of the drogue, whereas the undrogued drifter can more easily escape back out to sea. For example, in the linked maps, you can see far more drogued drifters beach in the Comoro Islands, than undrogued drifters.
@Richard Godfrey
First up, thank you for that beaching data and map.
Regarding, ‘The market is not just against Boeing.‘
True but compared to the CAC40, which is down around 3.2 per cent, Airbus’s treatment by the market (I’m only seeing them down 4.6 per cent) has been nowhere near as damaging as Boeing’s 14.6 per cent loss (against the backdrop of the Dow being down about 6.5 per cent, not forgetting Boeing’s significant impact on the Dow itself).
The fortnight from 7 November looks like Boeing’s worst for the past 5 years. Sliced or diced, this MCAS issue and Boeing’s handling of it has not been good for them.
@DennisW Thank you for your comments from November 17, 2018 at 8:18 pm.
Yes, running out of fuel certainly does not facilitate ending at a pre-determined location or performing a controlled ditching.
@Richard Godfrey
I have a question/comment about your drift studies if I may. It does seem to me that with so few data points (found debris), the difference between beaching locations from the 20-25S band and the 30–35S band is minor.
I see only two debris locations that are predicted by starting in the 20–25S band that are not predicted from the 30 -35S band.
(I am looking at these charts: https://www.dropbox.com/s/r4yjbeh436bxazx/Beaching%20Map%20-%2020S%20to%2025S%20vs%2030S%20to%2035S%20vs%20Debris%20Locations%20plus%20Start%20Locations.pdf?dl=0)
Is that a correct interpretation?
Is it not possible that these are just ‘outliers’ and this drift analysis is not able to discriminate latitude along the 7th arc due to the small number of known debris locations?
One example of how widespread beaching can be from the same starting location (nearly) can be seen here:
https://www.bbc.com/news/world-asia-35638091
re: ‘Qualitative’ Search Location Discriminant – Motive/Motivation
Assuming hiding the plane is the major objective [otherwise why spending so many hours flying (until fuel exhaustion) to the SIO, and taking so high a risk crossing the peninsula initially (aside: high risk taken probably indicating already in determined ‘self-destruction’ mode at this ‘early’ stage)], the ‘best’ impact location should/would be more ‘remote’ [low chance to find] in general (i.e., no going/turning ‘back’). Similarly, in accordance with this perspective/scenario, there is no reason not to [provide an opportunity for a] ‘slide’ if the ultimate/unifying aim is to sink the plane ([almost] intact) to the bottom of the ocean.
@Shadynuk.
One can’t put a lot of weight in predicted low-probability events (debris predicted to be at a particular beaching location, or not) with a limited number of trials (200 for Richard’s model). The uncertainties are large in this case. David Griffin’s model is statistically more certain (but not necessarily more or less accurate) because of the large number of trials (76,200).
@Mash.
@Don Thompson,
We know the SDU was without power from about 17:21 until 18:24. Perhaps the power was removed to maximize the confusion and to minimize the chances of being intercepted when crossing Malaysia. Then, once out of radar range, the power was restored (now with ACARS disabled) so it would become apparent that the aircraft was still in the air for a very long time, but at an unknown and distant (i.e., beyond radar range) location. Perhaps ZS wanted not only to disappear the aircraft, but to prove he was the smartest guy in the room in the process. He may also have been curious about how long it would take MAS to realize he was gone and place a satellite phone call to the aircraft. He found out at 18:40 and, if he was still listening, at 23:14.
@Peter Norton
RE: ‘Don’t we know that based on the EAD stating:…’
Yes, we know from the EAD that ‘if an erroneously high single angle of attack (AOA) sensor input is received by the flight control system, there is a potential for repeated nose-down trim commands of the horizontal stabilizer.’ However, we don’t know if the system was intended to operate that way, or if there was a ‘failsafe’ mechanism that didn’t work for some reason. For example, there might be some kind of comparison between the AOA inputs that should put the system in a ‘safe’ condition in the event of an AOA disagreement. What if that mechanism failed to work as intended due to a software coding issue? I’m not saying that happened; at this point we do not know.
RE: ‘Based on the above, why shouldn’t the “fix” be to do NOTHING in case of “an erroneously high single AOA sensor input” in manual flight rather than “repeated nose-down trim commands of the horizontal stabilizer”? Wouldn’t you consider this safer? In my view, anything else would violate the fail-safe paradigm as it could create a critical problem that wasn’t there in the first place. In my view, this is incompatible with a fail-safe mechanism.’
Reports suggest that MCAS was required for certification purposes, to help the pilot lower the pitch attitude in the event of a stall, especially at high thrust settings. If that’s the case, then ‘do NOTHING’ in the event of an AOA disagreement is probably not an option unless the system can discriminate between the sensors to determine which one is erroneous. If the system can’t already do that, then Boeing might look at using calculated AOA as a third input, similar to the B777. However, I suspect that would involve a lot of work and re-certification of the system. Ultimately, the ‘fix’ will depend on another failure analysis and risk assessment, informed by the results of the JT610 investigation. One would hope that Boeing and the FAA will be a LOT more conservative the second time around.
@All,
To see how easy, in hindsight, the ARA San Juan search could have been, see this:
https://drive.google.com/file/d/1KhhiigHcjAc3GPHPL8kM8imresi0J3gw/view?usp=sharing
Simply extrapolating the planned course at the last reported speed until the inferred implosion time from the CTBTO recordings gives a location only 4.4 NM from the wreck, much closer than any other method. The information needed to do this was available in December 2017. I suspect this calculation might have been done as one consideration in selecting the search area, but I have not seen it discussed publicly.
@all
I was hoping we had finally transitioned as a group to the “Inmarsat Winter” – the only thing the Inmarsat data can tell us is the plane went South at the FMT and came down rapidly after fuel exhaustion.
I see we still have all sorts of things people are still reading into it.
@Shadynuk
You asked:
(1) “I see only two debris locations that are predicted by starting in the 20–25S band that are not predicted from the 30 -35S band. Is that a correct interpretation?” – I think your interpretation is correct, only if you apply an inappropriate precision, but in general your interpretation is incomplete.
(2) “Is it not possible that these are just ‘outliers’ and this drift analysis is not able to discriminate latitude along the 7th arc due to the small number of known debris locations?” – I do not think that 10 out of 23 simulation data points, from a start latitude range of 20°S to 25°S, can all be dismissed as outliers.
The key thing I wanted to show in the graphics you reference, is the much wider spread of beaching locations from 20°S to 25°S, when compared to 30°S to 35°S. There have been only 11 MH370 confirmed or likely debris locations on mainland Africa, but they are widespread (debris found 3,712 km apart). A similar dispersion is seen in the example you show from the Tsunami in Japan (debris found 4,295 km apart).
7 debris locations were in the central area around Mozambique between 20°S and 30°S. Both start latitude ranges in the simulation have little problem to demonstrate this area can be reached. There are 13 simulation runs from the start latitude range of 20°S to 25°S and 8 simulation runs from the start latitude range 30°S to 35°S beaching in this central area.
1 debris location was in the north (above 10°S) and the only start latitude range, which shows beachings in this area is between 20°S to 25°S (4 in total). There were zero beachings for the start latitude range 30°S to 35°S.
3 debris locations were in the south (below 30°S) and the only start latitude range, which shows beachings in this area is between 20°S to 25°S (6 in total). There were zero beachings for the start latitude range 30°S to 35°S.
I think that is fairly conclusive, that the start latitude range between 20°S and 25°S should be searched to at least a search width of ±22 NM. In addition, as this is the only area not yet searched, as @DennisW has pointed out, the chances of finding MH370 are high.
Flight Global is reporting that a 737 MAX-8 operated by Sunwing had faulty air data, leading to replacement of the left ADIRU.
Regarding JT610 it looks like the the NTSC has released the FDR data from the accident flight and the preceding flight. There’s an embedded presentation in the following article – https://m.detik.com/news/berita/4312425/knkt-beberkan-data-fdr-lion-air-pk-lqp-di-dpr
The slide quality isn’t great but it appears to shows an ongoing battle between the MCAS trimming down and the crew trying to retrim (by using the electric trim I’m assuming given the bursts). I can’t be sure but it also seems to show a major disagreement between the two AoA sensors on both the accident flight and the preceding flight. That would suggest that the change out of the AoA vane did not correct the problem. You’d have to suspect the associated air data module or the ADIRU may have been where the fault lay.
I’m sure there will be more comprehensible coverage in the western press in due course.
@Andrew: re:JT610. I see your point. Thank you for broadening the perspective.
> DrB says:
> To see how easy, in hindsight, the ARA San Juan search could have been, see this:
> https://drive.google.com/file/d/1KhhiigHcjAc3GPHPL8kM8imresi0J3gw/view?usp=sharing
Yes. This is very painful to look at.
There are parallels to the search for AF447.
If memory serves me right:
(1) Both have been hiding in plain sight.
After a 2-year search, AF447 was found just 6,5 nm from its LKP.
p. 30+31+39: http://www.ifremer.fr/web-com/sar2011/Presentations/SARWS2011_FERRANTE.pdf
(2) Both crash sites have been missed during the initial search and were only identified in a second pass much later.
(3) In both instances, the first miss was the result of a wrong assumption:
SanJuan: POI was assumed to be a fishing vessel and dismissed.
AF447: search area was dismissed based on the false assumption of working pingers
“The search that listened for the acoustic pings from the aircraft’s data recorders almost certainly covered the location where the wreckage was eventually found. That’s an important point. A different analysis might have excluded this location on the basis that it had already been covered. But Stone and co chose to include the possibility that the acoustic beacons may have failed, a crucial decision that led directly to the discovery of the wreckage. Indeed, it seems likely that the beacons did fail and that this was the main reason why the search took so long.” ¹
Michael Purcell (WHOI): “we went back into areas that they had already surveyed but that required another look because of the terrain” ²
—
¹ http://www.technologyreview.com/s/527506/how-statisticians-found-air-france-flight-447-two-years-after-it-crashed-into-atlantic
² http://www.hydro-international.com/content/article/lessons-learned-from-the-af447-search
@Peter Norton.
1) Nothing below 1000m in the oceans can be described as ‘plain sight’.
2) The wrecksite of AF447 was surveyed once with side scan sonar. The only prior sweep over the wreckage location involved the attempts to localise the ULB transmissions. Localising ULB transmissions, and imaging by ensonification are two entirely different techniques.
3) The wrecksite of San Juan was not misidentified earlier in the search. It was, simply, missed. Early in the search for San Juan, OI used the AUV’s to make PoI investigations. No PoI dives, whether supervised/tended AUV or ROV were conducted in the locality of PoI 24. In the case of the search for AF447, many of the contributors to that search have sought to describe their work as pivotal in defining the final survey site. WHOI worked with UMASS on an oceanography solution, a WHOI team conducted the successful deep ocean search. My opinion is that it was the oceanography work that won out.
The San Juan wreck lies in a relatively narrow canyon/gully. From renderings of the bathymetry, the AUV tasked to conduct the dive in the area surrounding the San Juan wrecksite would have traversed across the canyon, not along it. I estimate the traverse across the canyon to be 1500-2000m, the canyon sides fell away sharply from the surrounding faces of the sloping continental shelf. It appears that, while traversing the canyon, the AUV altitude off the seafloor wasn’t maintained so as to adequately image the seafloor. The AUVs were, typically, navigating survey lines at 800m spacing at 50m altitude, the canyon floor may be 150m below the edge of the canyon. If the gradient of the canyon wall fall-away exceeds the underwater vehicle’s RoD, imaging is compromised – a ‘data holiday’.
@all,
A non embedded version of the NTSC power-point presentation showing DFDR derived graphics for PK-LQP ia at:-
https://tinyurl.com/pk-lqp-dfdr
@Barry Carlson: Thank you, Barry. I wonder if the resolver of the faulty AoA sensor was replaced as part of the maintenance. It appears to me that one or more of the higher order bits were not working, although the lower order bits seem fine.
@All
Re JT61: Some observations based on the FDR data for the accident flight in the KNKT presentation:
1. There was a large difference (approx 20°) between the two AOA vane outputs throughout the flight. The difference existed before the commencement of the take-off roll.
2. The left stick-shaker activated as the aircraft rotated during the take-off. It remained active for most of the flight, apart from a short period when the flaps were extended again during the climb (see below).
3. The flaps were retracted at the normal point, but shortly afterwards they were extended again for a short time before being retracted. They remained retracted for the remainder of the flight.
4. There were several automatic (STS) and manual (crew) trim inputs as the aircraft accelerated during the initial climb, which is normal.
5. Automatic (MCAS) nose-down trim inputs began as soon as the flaps were retracted the first time. These were counteracted by manual (crew) nose-up trim inputs.
6. Automatic nose-down trim inputs (MCAS) ceased for a short time when the flaps were extended again. They re-started as soon as the flaps were retracted the second time and continued for the remainder of the flight.
7. The average stabiliser position was about 5.0 units after flap retraction, with short period movements of about 1 unit up and down, due to manual (crew) nose-up inputs that opposed the automatic (MCAS) nose-down inputs.
8. The stabiliser position changed from about 5.0 units to 0.0 units (ie the nose down limit) at the end of the flight due to continued automatic (MCAS) inputs. The manual (crew) inputs were sporadic and of short duration during this period and, while they interrupted the automatic inputs for a short time, they were insufficient to return the stabiliser to the previous position.
9. The rate of descent increased rapidly when the stabiliser moved towards the nose down limit at the end of the flight, with a commensurate increase in airspeed.
During the previous flight from DPS-CGK, the AOA discrepancy was also about 20° throughout the flight and the left stickshaker was active from take-off until landing. Automatic (MCAS) trim inputs ceased during the climb. There were no further manual trim inputs after that point, which suggests the crew used the cutout switches to de-activate the trim.
The pilot of the accident flight kept control of the aircraft and maintained an altitude of about 5,000 ft for approximately six minutes by using manual trim inputs to oppose the automatic MCAS inputs. However, something changed during the last minute of flight and, apart from a few sporadic inputs, the pilot stopped actively opposing the automatic trim. Why?
One further comment to add to those above:
10. During the accident flight there was a small difference of about 10-20 kt between the left and right computed airspeed outputs that began as the aircraft rotated during the take-off.
@Andrew,
Looking at flight JT610, on power-up at the gate, the RH AoA vane is drooped about -5° – normal, but the LH AoA vane is indicating about +8° – abnormal, and adopts an even greater disparity, about 20° with the RH side, possibly on push back and taxi. The manner in which this LH disparity is recorded [to me] indicates an intermittent [interface coupling / dry joint?] problem in the electronics – ADIRU, as the LH vane later follows faithfully the RH vane [but with a large offset], which wouldn’t be the case if there was a bearing, synchro coupling problem.
In any case, the problem encountered seems no different to the previous days DPS – CGK flight, which was successfully handled. So, whatever maintenance did wasn’t related to AoA disparity, and I suspect that the accident flight’s crew thought they were dealing with UAS.
During pre-flight, would you normally check that AoA indications showed as matched – within small margins?
Peter Norton,
Don Thompson wrote, “In the case of the search for AF447, many of the contributors to that search have sought to describe their work as pivotal in defining the final survey site. WHOI worked with UMASS on an oceanography solution, a WHOI team conducted the successful deep ocean search.”
As one with a personal involvement with what transpired during the marine search phases for AF447, I concur with Don’s assessment. A lot of backtracking work went into determining from where the recovered bodies from the aircraft had originated – undertaken by UMASS on behalf of WHOI. At the end of the day, after much discussion prior to the final phase of the search, Mike Purcell made the decision to proceed as per the systematic search pattern already underway using AUV’s.
It was only after the finding of the debris that the UMASS-Dartmouth – Finite Volume Community Ocean Model (FVCOM) backtracking worked, as even over the relatively small distances involved, the blended winds data required a precise total leeway factor associated with the bodies. For obvious reasons that factor changed with the actual drift period involved.
In hindsight, the basic rule of searching outward from the Last Known Position wasn’t followed, and not detecting the ULB pingers ‘muddied the waters’.
@Don Thompson
ad (1): I was speaking from a 2d map perspective to point out a
similarity to Dr.B’s quoted remark.
ad (2): re “The wrecksite of AF447 was surveyed once with side scan
sonar.” I didn’t claim the opposite, but some areas were surveyed twice
with sonar, how do you know the wrecksite was not among them ?
ad (3): re: “The wrecksite of San Juan was not misidentified earlier in the search. It was, simply, missed.” What is your evidence for that?
My statement was based on the information provided by Richard Godfrey. According to his information, the “wreckage location was part of the 2nd AUV sortie from 9-11 September”, described as a “metallic wreck with same dimensions of the sought contact”, but “dismissed as a fishing trawler that was previously reported sunk 330m away”.
———-
Richard Godfrey says:
November 17, 2018 at 10:15 am
The wreckage location was part of the 2nd AUV sortie from 9th to 11th September 2018 depicted below:
http://www.dropbox.com/s/zo3lpld3mozr3s6/ARA San Juan Initial Coverage.pdf?dl=0
This sortie raised the first POI described as a metallic wreck with the same dimensions of the sought contact. It was dismissed as a fishing trawler that was previously reported sunk 330m away.
http://www.dropbox.com/s/mip9uuwsvbl0xwx/ARA San Juan Initial POI.png?dl=0
———-
Richard Godfrey says:
November 18, 2018 at 1:25 pm
re: “Richard Godfrey believes OI may have initially misidentified POI #24 as a sunken trawler.”
[…]
The POI #2 was discarded as a fishing trawler that was reported to have sunk in the vicinity. This decision was made without a ROV deployment.
Unfortunately POI #2 had the same dimensions as POI #24 and the ARA San Juan.
Perhaps OI can tell us whether POI #2 was at the same location as POI #24?
———-
> DrB says:
> Richard Godfrey believes OI may have initially misidentified POI #24 as a sunken
> trawler. If so, they must not have employed the ROV on it, and the initial side scan
> sonar image must have been of poor quality.
The initial side scan sonar image would be among the most interesting information for a long time with regards to MH370’s lower confidence coverage areas. Will it be released?
@Barry Carlson
RE: “During pre-flight, would you normally check that AoA indications showed as matched – within small margins?”
AOA indication is a customer option that is available on some newer Boeing aircraft. Some operators have them, but many do not. B737s that have AOA indications on the pilots’ PFDs also have an AOA DISAGREE alert message that is annunciated on the PFDs if the L & R AOA values disagree by more than 10 degrees for more than 10 continuous seconds. Pilots are not required to check the AOA indications during the pre-flight. My understanding is that the Lion Air aircraft did not have the AOA indication option or the AOA DISAGREE alert.
> Barry Carlson: re:AF447 “In hindsight, the basic rule of searching outward from the Last Known Position wasn’t followed”
This is another parallel between AF447 and SanJuan.
In both cases searching outward from the LKP would have yielded better results as the actually employed search area prioritizing.
This leads back to an idea pondered by others and myself here:
Given that
– both the Inmarsat data and the drift studies don’t conclusively pinpoint specific latitudes and
– most crashes occur very close to the LKP
why wouldn’t the 7th arc (between a conservative northern and southern end point) be searched from the inward out?
(This had been my search strategy suggestion right from the beginning in 2014.)
@Andrew said: During the accident flight there was a small difference of about 10-20 kt between the left and right computed airspeed outputs that began as the aircraft rotated during the take-off.
Which you would expect if an incorrect AoA correction to airspeed was applied on the left side.
However, something changed during the last minute of flight and, apart from a few sporadic inputs, the pilot stopped actively opposing the automatic trim. Why?
Yes, I think that is what we’ll need to understand from the CVR, when found. What it seems they did not do is cutoff the electric trim and use the manual trim wheel.
@Peter
why wouldn’t the 7th arc (between a conservative northern and southern end point) be searched from the inward out?
The information pointing to a LKP has evolved over time in the mainstream analtical community (not with respect to my opinion). Intially BFO, BTO, and flight dynamics were all in the mix. Although it should have been obvious that BFO and flight dynamics were given much too great a weight in early analytics (even today the DSTG mistakenly clings to a belief of a mysterious position corrupting BFO factor).
I think hindsight would result in a modified search strategy much like you are suggesting – a wider range of latitudes at the expense of search width from the 7th arc.
@Victor, @Andrew,
A comparison of the FDR data between the previous flight of PK-LQP and the crash flight is quite revealing:
The same problem occurred in both flights and in both flights the left stick shaker was activated at take off.
In the previous flight the pilots switched off the MCAS after around 6.5 minutes into the flight. They even switched it back on briefly around 12 minutes into the flight, but got the same effect and switched it off again. The pilots had obviously diagnosed the problem.
In the crash flight, the first action from the pilots was to extend the flaps again, but that did not help so the flaps were retracted again after around 4 minutes into the flight. The MCAS was not switched off. After about 10 minutes into the flight, there is a sudden jump in both the local and foreign force on the control column pitch and MCAS runaway occurs. The pilots had apparently not diagnosed the problem.
https://www.dropbox.com/s/vs5ckwuh3bgo3ep/JT610%20Data%20Comparison.pdf?dl=0
@Victor
RE: ”Which you would expect if an incorrect AoA correction to airspeed was applied on the left side.”
Yes.
@Richard Godfrey
RE: ”In the crash flight, the first action from the pilots was to extend the flaps again, but that did not help so the flaps were retracted again after around 4 minutes into the flight.”
Theoretically, MCAS shouldn’t operate when the flaps are extended. I’m wondering if the automatic trim inputs that continued when the flaps were extended again were the normal operation of the STS rather than MCAS. The nature of those inputs is different to many of the other automatic inputs; they are of very short duration, more like ‘blips’ than 10 second bursts. The longer duration bursts didn’t recur until the flaps were retracted the second time.
@Richard
The pilots had obviously diagnosed the problem.
Yes, one could conclude that even without the FDR data. That is one of the things that makes this event so odd and so maddening for me. Don’t pilots talk to one another? Is there no feeling of an obligation to pass information to the following flight crews. Ami and I do a much better job of communicating issues with our family vehicles.
@Richard G, @DennisW, @Andrew: Good luck with a jury trying to digest the material and deciding whether the 737 MAX had design flaws, whether Boeing provided insufficient documentation, and whether a competent crew should have been able to diagnose the condition and recover. This case will be a great one for the lawyers.
@Victor
I agree with your summary.
Even worse, the accident may have been a combination of design flaws AND/OR insufficient documentation AND/OR lack of training AND/OR pilots not following procedures AND/OR pilots unable to diagnose the condition and recover. As you say the only winners will be the lawyers.
On 31st October 2018, 2 days after the crash, the Indonesian Government had already ordered Lion Air to suspend the Technical Director Muhammad Asif from duty, together with a number of technicians who had cleared the plane for departure. This was before the discovery of the FDR on 1st November 2018 or media reports about Boeing withholding information on 12th November 2018.
@Peter Norton
1) I strongly disagree with any characterisation that suggests search targets on the ocean floor lie in ‘plain sight‘. Even in 2018, for the purposes of a wide area search, our technological prowess affords a view of deep ocean sea bed that is comparable to terrestrial navigation through a dense fog, at night, illuminated with only a dim flashlight.
2) AF447: in the immediate 6 weeks after the crash (Phase 1 according to BEA), a good proportion of the candidate area was searched with hydrophones with the goal of locating a ULB ultrasound source. No ULB sources were detected. Phase 2 involved IFREMER’s deep towed side scan sonar equipped UV to the south-east of the LKP. Phase 3 involved Phoenix International’s ORION deep towed side scan sonar equipped UV and 3 REMUS 6000 AUVs to the north-west quadrant of the LKP but ignoring an area close to the LKP. For Phase 4, the 3 Remus 6000 AUVs were deployed into a previously unsearched area, closer to the LKP than previous seafloor searches. I wouldn’t regard listening for ULB detections as a ‘search’, the objective is to localise a recorder’s ULB beacon, identifying the recorder is yet another challenge. Source: Sea Search Operations, October 2012, BEA.
3) ARA San Juan: Seabed Constructor began wide area search by deploying AUVs into Site-1, Area 1. From east to west AUV-2/Dive 01, AUV-6/Dive 03, and AUV-1/Dive 05 covered the first 3 of the 5 segments in that area. PoI 24 lay within a trench/canyon at the north of the AUV-1/Dive 5 tasking. No ROV nor supervised/tended AUV missions reviewed that area. Source: Ocean Infinity’s Daily Status Reports hosted at Armada Argentina’s web server.
OI’s Daily Status report recorded 5 PoIs within the easterly segment of Area 1, surveyed by AUV-2/Dive 01. To investigate those PoIs, AUV-2/Dive 15 was tasked with Seabed Constructor supervising/tending the AUV along a path in the NE corner of Area 1, a considerable distance from PoI 24. I’ve emailed Richard concerning my observations on the survey of Site 1, Area 1 and the absence of any ROV or AUV investigations in proximity to PoI 24.
During AUV-1/Dive 05 I suspect the AUV did simply miss the wreck site as it navigated over the southern lip of the tranch/canyon. Mapping PoI 24 onto the processed images of the bathymetry shows the location on the southern side of the canyon. I do expect that OI had good bathymetry data to assist mission plans, but the AUV aided INS navigation does suffer inherent drift during the mission, and the AUV uses a forward looking sonar to avoid obstructions ahead. AUV-1 may have descended into the canyon too late, when approaching the wrecksite from the south, or, conversely, ascended too early thus losing resolution or acquisition of the seafloor. Alternatively, the plan may have excluded a descent into the canyon at that time.
Concerning 7th arc search and searching out from a central point (or line of position): consider the practicalities of deep tow search. Unnecessary winching of the deep tow UVs was avoided. The ships equipped for deep tow operations required 10,000m tow cables with power and fibre optic ‘umbilicals’ woven in, and specifically purposed ‘traction’ winches. The cables weren’t durable enough for repeated, full, winch-in/out operations. The deep tow UVs were kept at depth throughout most of the 6 week swings, hence the long survey lines parallel to the 7th arc. Fugro’s AUV surveys then filled in the gaps, assuming the gaps were at depths <4500m.
Much of Ocean Infinity's work, south of the Diamantina Escarpment, was conducted with the AUVs traversing the 7th arc so as to align with the long valleys and ridges that lie transverse to the 7th arc.
@Andrew,
I found that American & Delta specified the AoA indication on their 737-NGs, with the requirement carried forward by AAL for its 737 MAX. Some sources made reference to previous accidents where AoA sensor failures had been identified as causal (but not necessarily on 737s).
@ DennisW
RE: ”Don’t pilots talk to one another? Is there no feeling of an obligation to pass information to the following flight crews.”
Where I work an incident such as JT43 would be treated very seriously by the airline’s engineering, flight operations and safety departments. The commander would be required to submit an Air Safety Report that would be investigated by the relevant departments and passed to the civil aviation regulator. The commander would also be required to include ALL the relevant details in the aircraft’s maintenance log, to help the engineers to diagnose and rectify the problem. Most pilots I know would call or visit the Chief Pilot in person to make a verbal report after such an incident. The Chief Pilot would then follow up with the engineering department to make sure the problem is resolved.
It will be very interesting to find out how much of that occurred at Lion Air after the JT43 incident. Very little, it seems.
@Andrew
Thx for that. That seems how things should be.
Video clip showing a relevant exercise at NASA Ames Human Factors Research Lab (as it was then called).
@Peter Norton
I previously stated “The POI #2 was discarded as a fishing trawler that was reported to have sunk in the vicinity. This decision was made without a ROV deployment. Unfortunately POI #2 had the same dimensions as POI #24 and the ARA San Juan. Perhaps OI can tell us whether POI #2 was at the same location as POI #24?”
Meanwhile, Don Thompson has convinced me, by his carefully tracking the AUV sorties, that several POIs (including POI #2) were investigated with AUV 2 in Dive 15. This was a short dive lasting 13:46 hrs from 12 SEP 2018 20:47 UTC to 13 SEP 2018 10:33 UTC mostly covering the canyon to the South East of the canyon where ARA San Juan was found. The nearest POI in this sortie was at 45.9491°S 59.6875°W, where Seabed Constructor circled above the AUV in the canyon from 13 SEP 2018 08:07 UTC for 46.5 minutes. This nearest POI is 6.7 km away from the location where ARA San Juan was discovered at 45.9533°S 59.7741°W.
I agree with Don’s conclusion, that the ARA San Juan was missed the first time, because it was in a deep canyon against the southern steep wall.
@DennisW
No sweat. The airline where I work (and many airlines like it) has gone to great lengths to establish a safety culture where employees can openly report report deficiencies, expose hazards and raise safety concerns without fear of retribution. The company has a policy, signed by the CEO, that states the reporting of unpremeditated or inadvertent errors will not result in disciplinary or punitive action being taken against the reporter. The underlying ethos is that we’re all human and humans sometimes make mistakes. From a safety perspective, the overarching principle is that the organisation needs to know about such occurrences so that we can all learn and find ways to improve the system. At the same time, however, reckless behaviour, dereliction of duty, persistent breaches of company standards or deliberate failure to report occurrences are not tolerated.
Unfortunately, some less enlightened airlines in certain parts of the world (some of them major players) persist in punishing employees for committing errors, in the belief that doing so weeds out the transgressors and improves safety. Nothing could be further from the truth, as such behaviour only drives problems underground, where they can’t be ‘fixed’.
@Don Thompson
Wow, that’s old (but still very relevant)! NASA’s research back in the 1970s was the cornerstone of modern CRM training. We’ve come a long way from the days where the captain was ‘god’ and could not be challenged, but even so, it’s still the case today that some crews perform far more effectively than others when problems arise.
@Don Thompson
Two more snippets along the same lines:
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20020087647.pdf
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19870013202.pdf
There’s plenty more on the subject!
@Andrew,
Yes, that clip is ‘old’! 1985, I think, and recorded a short while after I visited to commission some updates on the the sim. A colleague, visiting later on a support or tech training task, ‘made up the numbers’ in the simulated ATC centre for the TV crew.
BBC recently broadcast a compilation of Horizon’s reports on air safety over the years, I acquired a full copy.
The centre now uses a CAE built B747-400 simulator, replacing that Link 727.
@Victor
MH370 Legal Case Dismissed in USA but…
https://www.reuters.com/article/us-malaysiaairlines-mh370-lawsuit/u-s-judge-dismisses-litigation-over-missing-malaysia-airlines-flight-idUSKCN1NS25B
Mary Schiavo, a lawyer for some of the plaintiffs, said in an email on Friday her clients were preparing for a June 2019 trial in Kuala Lumpur over the plane.
———-
Barry Carlson: Looking at flight JT610, on power-up at the gate, the RH AoA vane is drooped about -5° – normal, but the LH AoA vane is indicating about +8° – abnormal, and adopts an even greater disparity, about 20° with the RH side, possibly on push back and taxi. […] During pre-flight, would you normally check that AoA indications showed as matched – within small margins?
Andrew says: AOA indication is a customer option that is available on some newer Boeing aircraft. Some operators have them, but many do not. B737s that have AOA indications on the pilots’ PFDs also have an AOA DISAGREE alert message that is annunciated on the PFDs if the L & R AOA values disagree by more than 10 degrees for more than 10 continuous seconds. Pilots are not required to check the AOA indications during the pre-flight. My understanding is that the Lion Air aircraft did not have the AOA indication option or the AOA DISAGREE alert.
———-
@Andrew:
Does that mean that during taxi
– there was no alert on JT610
– but with this option installed, there would have been an alert, which would very possibly have saved the plane from taking off ?
@TBill
Is there a link to the “61-page decision” ?
@Ventus45
I do not see a link to the 61-page decision at this time.
@Ventus
https://ecf.dcd.uscourts.gov/cgi-bin/show_public_doc?2016cv1129-22
@Andrew,
Re: AoA disparity: Thanks for your comprehensive reply.
At this stage we are only privy to limited DFDR data, and what visual/aural warnings were generated may explain the pilots’ actions.
@Peter Norton
RE: “Does that mean that during taxi
– there was no alert on JT610
– but with this option installed, there would have been an alert, which would very possibly have saved the plane from taking off ?”
I assume so, but that’s based on reports found on the Internet. I have not flown the B737 and have limited knowledge of the logic that generates the AOA DISAGREEMENT alert, other than it requires more than 10° of AOA disagreement over a continuous period of more than 10 seconds.
A somewhat damning report about safety at Lion Air:
https://www.nytimes.com/2018/11/22/world/asia/lion-air-crash-safety-failures.html
@Andrew: Thank you. That’s a very damning story about Lion Air, which I think will likely share some (but not all) of the blame.
@Peter Norton: There is a long list of ways this accident could have been avoided. For instance, higher reliability AoA sensors, better fault identification logic, better maintenance service, better fault indication, better documentation, better training, better flying skills, etc. All of these (and others) contributed, but how can you attribute any single item to the crash?
@all
re: Flight Path/Destination ‘Reverse Engineering’ [after FMT]
By Assuming:
1. 100% successful operation (no ‘accidents’).
2. (most probable) operation purpose. [say hide plane]
3. major consideration/success factors.
4. major operational/time/distance/flight mode/fuel/… constraints.
5. 100% optimisation/optimised (professional) solution.
Perhaps one can [better] reconstruct the ‘true’/’genuine’ flight path/destination – the 100% ‘masterpiece’.
[At least a deterministic (not probabilistic) scenario’s solution.]
@mash
Your suggestion suffers from a serious negative attribute – it does not make any sense.
As Peter Norton pointed out a few posts ago, the only reliable information we have relative to the LKP is the BTO value associated with the 7th arc. Following your suggestion constitutes a repetition of a strategy that did not work i.e. using “assumptions” to refine the LKP. We should have done exactly as Peter suggested – searched outward from the 7th arc bounded by reasonable latitude constraints. The latitude constraints suggested by the debris findings (also hard evidence, not an assumption) are 35S to 20S.
@Andrew: thank you
@Victor Iannello: I don’t. Andrews posts have shown that the issue is more complex. The root cause will be determined by the investigation. Seeing the FDR data, I was just wondering aloud if a symptom could have been observable prior to take-off, and if that could have prevented the take-off and thus saved many lives.
I wouldn’t expect an alert given the minimum speed for an AOA DISAGREEMENT alert, but if this B373 MAX had the AOA indication option installed, I wonder whether the crew could/would have noticed the 20° disparity during taxi, following Barry Carlson’s comment?
I assume it would also have made it easier to diagnose the problem in the air. That is why I agree with Richard Godfrey in saying:
“If a system such as the STS relies on a single AOA sensor input, it should be mandatory to install the AOA indication option, so that in the event that one of the AOA sensors fails, the pilot is given an alert that AOA DISAGREE and can quickly diagnose the problem. I think it is bad practice that Boeing offer the AOA indication as an option. It should be mandatory.”
Whether the AOA indication option would have made any difference prior to take-off is certainly a separate question. I don’t know how knowledgeable this person is, but according to this comment I found on avherald’s JT610 page, the AoA vanes are useless on the ground:
“AOA probes are completely useless on the ground below about 60 kts. They can be fully up, or fully down depending on the wind. It is not until good airflow starts over them during the take off run that they move to the correct position. If you go and touch one on the aircraft on the ground […] you will see that they are very easy to move, and then stay where you leave them.”
Also interesting, from the same page:
“As someone who make a living maintaining aircraft, I recently had the joy of installing not 1, not 2 but 3 different AOA vanes on an aircraft (I will not name the manufacturer) to replace a damaged one. 2 “overhauled” vane were faulty… Try to prove to your boss that you need a 3rd one! You have to be pretty bullet proof in your troubleshooting. I am also extremely careful at certifying maintenance after anything involving stall protection systems, airspeed and altitude indication. I don’t hesitate to require a test flight WITHOUT PASSENGERS after ambiguous troubleshooting results and duplication of problems. The pressure to keep these airplanes in the air is real and the urge to save money is even bigger but try a crash to see what will happen to the company bottom line…”
@DrB
I’m ready your paper “Analyses of Predictions by MH370 Debris Drift Models” with interest. The representation of the results of the two models in the format you have chosen is helpful. I’m still discussing with David Griffin about the de-biasing process he used to compensate for systematic errors in the BRAN2015 circulation model. I feel I can only give a proper feedback on your paper after we concluded those discussions.
One point David explained is that they are working on improving the BRAN model, but that it is a “massive task”.
@Peter Norton
RE: ”Whether the AOA indication option would have made any difference prior to take-off is certainly a separate question. I don’t know how knowledgeable this person is, but according to this comment I found on avherald’s JT610 page, the AoA vanes are useless on the ground:…”
That’s a very good point; one I should have considered in my earlier reply. The AOA vanes can sit at different positions while the aircraft is on the ground and will remain that way until the airspeed is sufficient to align them with the airflow. The AOA DISAGREEMENT alert logic is likely to use either a speed signal or a weight-on-wheels signal to avoid nuisance alerts on the ground. On that basis, the crew of JT610 would not have received an alert while the aircraft was taxiing, even if the disagreement alert option had been installed.
@DennisW
Thanks for the link to the court judgement.
@Andrew: Yes, this was my assumption, that the AOA DISAGREEMENT alert depends on a minimum speed (and possibly also weight-on weels signal as you say). Wasn’t this also why the AF447 stall warnings stopped*, due to airspeed falling below the minimum?
If the AOA vanes are useless below 60kt, the AOA indication option would not have helped during taxiing, even if it were installed.
* « As the stall worsened, the aircraft’s airspeed bled away and it began to descend, the angle of attack increasing to 41.5° before the airspeed fell below 60kt and the angle of attack became “non-computed data” – an invalidation which shut off the stall alarm. » ¹
In addition to the avherald quote above (about AoA vanes being useless below 60kt), this is the second reference to 60kt.
One more:
« If you’re sitting on the ramp the AoA vanes will typically read full throw (~45-60 degrees nose down). They converge on reality during the initial acceleration down the runway at about 60 knots. This is, not un-coincidentally, about the same speed that the airspeed indicators come alive (this is part of why pilots call “airspeed alive” during the initial roll). In short: you absolutely need valid airspeed to get valid AoA. » ²
As a side-note:
« The French cockpit union SNPL […] insists the design of the stall warning “misled” the pilots. “Each time they reacted appropriately the alarm triggered inside the cockpit, as though they were reacting wrongly. Conversely each time the pilots pitched up the aircraft, the alarm shut off, preventing a proper diagnosis of the situation.” » ³
I know this is not the whole story, but it certainly wasn’t helpful.
—
¹ http://www.flightglobal.com/news/articles/airbus-clashes-with-pilots-over-af447-alarm-360681
² http://www.airliners.net/forum/viewtopic.php?t=770127&start=200#p11112717
³ http://www.flightglobal.com/news/articles/stall-warning-controversy-haunts-af447-inquiry-360336
@Victor. “Good luck with a jury trying to digest the material and deciding whether the 737 MAX had design flaws, whether Boeing provided insufficient documentation, and whether a competent crew should have been able to diagnose the condition and recover.”
Yes when it comes to blame distribution there will be plentiful contest. However I think it likely that the prime responsibility is with the ground crew and/or the culture in which they worked.
No design flaw was responsible for the aircraft being released to fly in when defects indicated it was unairworthy.
You should not release an aircraft with a CAS disparity of 17 knots at take off and landing (previous flight) much less when unusual steps have been needed to stop the aircraft pitching down. Such defects must be found and fixed before next flight. If there is doubt they have been fixed there should be a local maintenance test flight with a pilot well versed in system details, briefed on the nature of the problem and with a clear understanding of the corrective steps available.
Similarly you should not release an aircraft which has had a persistent unexplained down trim without being sure that the cause has been found and corrected.
And you should not release an aircraft which has had a persistent stall warning via stick shaking on the captain’s side for an entire flight.
Gawd to all three.
Even if the ground crew, which I hope included professional aircraft engineers, had little information in their manuals about the new MCAS arrangement they should not release an aircraft in that state.
@Andrew. As I raised earlier, supposing the aircrew to be a part of the acceptability loop in release of an aircraft is to muddie responsibilities such that the ground crew come to see themselves (and are seen by the aircrew) as subordinate. If then aircrew are subject to other influences: commercial pressures, direction from a chief pilot, ego, such weakening of technical airworthiness responsibility can have the result we now see.
What should constrain airline owners and directors unfamiliar with the concepts and disciplines on which aviation is based are outside regulatory constraints and inspections though unfortunately these can be subverted.
As to FDR data detail I add a couple of points. Leaving aside whether the pilots tried trim deselection (I am a bit lost as to how then they would retrim if nose down at that point) there was the likely outcome of selecting 5 deg of flap again. That relieved stick pressure previously.
You may be able to throw some light on the term computed airspeed which I take it is another term for TAS, derived from CAS and air density. In that case there may not be a simple cosine relationship between a two CASs since AoA is already taken into account by nature of a pitot via the total pressure it perceives. Distortion, leaking or a wonky static could be amongst explanations.
As to why there was less corrective trimming at the end the impression I have is that it was all hands to pull on the control column, intuitively, when the nose down predicament became obvious.
I gather we can expect a preliminary report from Indonesia next Wednesday.
@Andrew and @Peter Norton
As per Richard’s previous correction to me.
The Boeing FCOM Bulletin of November 6, 2018 states:
AOA DISAGREE alert (if the AOA indicator option is installed).
The NY Times article comments on training, noting that Garuda’s pilots used the (Boeing Flight Services) 737 MAX simulator in Singapore, while Lion Air use their own three 737-900ER simulators.
The NY Times article is illustrated with images of the Boeing 737 Max test aircraft during its visit to the 2016 Farnborough airshow. At that time the aircraft, regitration N8704Q, had the AOA Indication option configured. That aircraft was later delivered to Southwest Airlines, Southwest’s 737 NG and MAX aircraft are not configured for AOA Indication.
@David,
Surely maintenance recovered the QAR data from the previous flight, reviewed it to check what was causing the RH stick-shaker and MCAS/STS trim problems.
Would seem though, they weren’t aware of the AoA disagreement, and/or didn’t relate it the LH/RH CAS discrepancy.
Or did they just do …?
Should have been “LH stick-shaker”.
@Barry Carlson
@David
Surely maintenance recovered the QAR data from the previous flight, reviewed it to check what was causing the RH stick-shaker and MCAS/STS trim problems.
I have been thinking the same thing. However, somewhat astoundingly, neither the stick shaker activation nor the need to cut-off the automatic stab trim on JT43 appear to have been written up in the tech log! Engineering are unlikely to be chasing down issues that they haven’t been made aware of.
When they are parcelling out blame at Lion some of it will undoubtedly go to the JT43 flight crew. You get the feeling that had they written up the runaway stab issue on their flight things may have turned out differently. Even if the aircraft had been released to fly JT610 in exactly the same condition the incoming crew would have almost certainly been ‘primed’ for a runaway stabiliser rather than just an unreliable airspeed.
@David
RE: “As I raised earlier, supposing the aircrew to be a part of the acceptability loop in release of an aircraft is to muddie responsibilities such that the ground crew come to see themselves (and are seen by the aircrew) as subordinate. If then aircrew are subject to other influences: commercial pressures, direction from a chief pilot, ego, such weakening of technical airworthiness responsibility can have the result we now see.”
Disclaimer: I can only speak from my own experience working for reputable organisations where there is no ‘muddied’ responsibility between engineers and aircrew, and where engineers certainly do not see themselves as subordinate. At those organisations, a pilot would NEVER try to influence an engineer to abrogate his or her responsibility for the airworthiness of an aircraft. If anything, the system occasionally works the other way around, where pilots, who are normally averse to risk, are under pressure to accept aircraft with deferred defects that are not operationally acceptable to them.
I fully accept there are organisations that don’t subscribe to the standards held by more reputable operators. The following comments describe how the system should work.
The release of an aircraft by engineering and the acceptance of that aircraft by the commander are separate functions. The commander does not influence the engineer’s decision to release the aircraft; he or she decides whether to accept the aircraft after it has already been released. Any engineer worth his or her salt would NOT release an aircraft unless satisfied that it is airworthy.
Once the aircraft is released, the commander should review the maintenance log. If a reported defect has been ‘rectified’ and signed off as serviceable, the commander might question the maintenance action that has been carried out. For example, many faults that occur while airborne cannot be repeated on the ground. The engineers carry out the required diagnostic tests, find no fault and subsequently clear the defect as ‘ground checked serviceable’. If that same fault has occurred on a number of previous sectors, then clearly there is a problem that needs further investigation. The engineers should pick that up and take the necessary action, but if they don’t (usually due to time pressure), the commander should start asking questions and is under no obligation to accept the aircraft until he or she is satisfied.
The commander might also decide that a defect should be rectified before flight, even though it has been deferred under the MEL. The MEL outlines the commander’s responsibilities in that regard. For example:
“MEL conditions and limitations do not relieve the commander from determining that the aircraft is in a fit condition for safe operation with specific unserviceabilities.”
“The decision of the commander to have allowable inoperative items corrected prior to flight will take precedence over the provisions contained in the MEL. The commander may request requirements above the MEL, whenever in his judgement such added equipment is essential to the safety of a particular flight under the conditions prevailing at the time.”
“The MEL does not take into account multiple unserviceabilities. Therefore, before dispatching an aircraft with multiple inoperative MEL items, it must be assured that any interface or interrelationship between inoperative items will not result in a degradation in the level of safety and/or unduly increase crew workload. When assessing the effect of multiple unserviceabilities, in particular those involving related systems, it is essential for the commander to exercise good judgement based on the specific circumstances at the time, including enroute airport availability, navigation facilities and weather conditions.”
Defects that are acceptable from an engineering perspective might not be operationally acceptable; that’s the commander’s call and would normally be made in consultation with engineering and flight operations management.
RE: “Leaving aside whether the pilots tried trim deselection (I am a bit lost as to how then they would retrim if nose down at that point) there was the likely outcome of selecting 5 deg of flap again. That relieved stick pressure previously.”
The B737 has a manual trim wheel that can be used to manually adjust the stabiliser, albeit much more slowly than the electric trim. The crew’s decision to reselect flap seemed to stop the action of the MCAS, as it should have done. The outcome might have been very different if the crew had not retracted the flap the second time.
RE: “You may be able to throw some light on the term computed airspeed which I take it is another term for TAS, derived from CAS and air density. In that case there may not be a simple cosine relationship between a two CASs since AoA is already taken into account by nature of a pitot via the total pressure it perceives. Distortion, leaking or a wonky static could be amongst explanations.”
My understanding is that computed airspeed is simply CAS, ie the speed computed by the ADIRU from the pitot-static and AOA data.
I don’t understand your point. The ADIRU uses AOA data to correct the static pressure data to remove the position error caused by the changing AOA. In this case, it seems that the erroneous AOA data was the cause of the difference between the L & R computed speed shown in the FDR report. It probably would have resulted in IAS DISAGREE and ALT DISAGREE alerts on the pilots’ PFDs, as described in the Boeing FCOM Bulletin.
@Don Thompson
RE: “The Boeing FCOM Bulletin of November 6, 2018 states:
AOA DISAGREE alert (if the AOA indicator option is installed).”
Sorry, I don’t understand your point. We believe the Lion Air aircraft was not fitted with the AOA indicator option/AOA DISAGREE alert. @Peter Norton asked if the crew would receive an AOA DISAGREE alert while taxiing if the aircraft DID have the option installed.
RE: “The NY Times article comments on training, noting that Garuda’s pilots used the (Boeing Flight Services) 737 MAX simulator in Singapore, while Lion Air use their own three 737-900ER simulators.”
It would be interesting to find out how Garuda’s differences training for the MAX differs from that of Lion Air and what prompted them to include 737 MAX simulator time in that training, if indeed that is the case. My understanding is that most operators only require their pilots to undergo a CBT course before flying the MAX.
@Andrew
@Don Thompson
It would be interesting to find out how Garuda’s differences training for the MAX differs from that of Lion Air and what prompted them to include 737 MAX simulator time in that training, if indeed that is the case.
Yes, I’d query that also. Garuda operate only one MAX, it was delivered late last year.
More broadly on that NYT article, while there is no doubt that the lack of an effective safety culture at Lion Air will figure prominently as a contributing factor in the JT610 crash, the article has just a whiff of hatchet job about it.
It states that, ‘In recent years, its flagship carrier has suffered at least 15 major safety lapses, including a crash that killed 25 people …’
I guess what constitutes ‘recent years‘ can be relative but you have to go back 16 years to 2002 to get the ‘15 major safety lapses‘. Lion Air only commenced operations in 2000. I guess that ‘Since commencing operations 18 years ago the airline has suffered at least 15 major safety lapses …’ doesn’t have quite the same Wow! factor to it.
And if you look at the ‘major safety lapses‘ they include a fuel spill, a very hard landing, hitting a cow landing at Jalaluddin Airport and the recent taxiing accident where a B737-900 clipped a light pole with its left winglet while being marshalled.
With regards to the latter, two weeks ago Southwest did essentially the same thing at Saint Louis-Lambert International Airport except they carved a chunk out of an adjacent WOW Air Airbus A321neo’s winglet. Now, I don’t think that Southwest has a safety problem at all but you could probably paint them that way if you were so inclined. They have a fleet 6 times larger than Lion but in a six month period from May to October this year they had eight serious inflight issues that included an inflight engine problem and subsequent engine fire after landing, a few flap malfunctions, two hydraulic failures and a couple of serious instrumentation issues. And that doesn’t include the uncontained engine failure and fatality in April.
Getting back to Lion’s safety record, nearly half of the ‘15 major safety lapses‘ were overruns/runway excursions (tropical weather and runway/airport conditions tend to also be factors with these) and four involved older McDonnell Douglas MD-80/90s. The quoted ‘crash that killed 25 people‘, the 2004 crash of Lion Air Flight 583 at Surakarta involved both; an overrun and an MD-80. That crash was not overly dissimilar to the crash of American Airlines Flight 1420, also an MD-80, at Little Rock, Arkansas five years earlier which killed 11 people.
As to the NYT’s reporting of the 2013 descent into water on approach of Lion Air Flight 904 at Bali (a crash recently replicated by Air Niugini Flight 73 at Chuuk, Federated States of Micronesia), they state,
‘Official accident reports accused the 24-year-old first officer of lacking “basic principles of jet aircraft flying” and advised Lion to “ensure the pilots are properly trained.”‘ The accident report does not accuse the FO of lacking “basic principles of jet aircraft flying” at all. You might infer that from the report but it doesn’t state it. As to the report advising Lion to “ensure the pilots are properly trained.”, well, that phrase does not appear in the accident report so I don’t know where the NYT is “quoting” that from. It is also interesting that of all the photographs of the JT904 crash that are available, the NYT elected to go with one taken four days after the crash when the wreck (that was still largely intact after the crash) had been towed by savage crews to the nearby breakwater and cut up for recovery by crane.
Now, none of this is to diminish the serious piloting shortcomings that managed to put JT904 into the drink. Lion had managed to pair a 6,000 hours-on-type captain who had a noted habit of pressing approaches that were not properly stabilised with a 900 hours-on-type FO who was regularly graded at or below the minimum standard. Throw in the some ordinary weather on approach and what could possibly go wrong?!
And I’m not suggesting that Lion Air aren’t likely to be found partly culpable for JT610, they should be. I just think that the NYT article is hyped.
Any old how, I’m sure that Boeing enjoyed reading it.
… towed by savage crews …
Mmm, that would be ‘salvage’ crews but given the conditions some of them might have been a bit on the shirty side.
@Barry Carlson, Mick Gilbert. I was unaware that the write up was incomplete Mick. However I find it hard to imagine that there was insufficient there to raise the likelihood that there was a defect which could not be deferred or which at the least needed further inquiry of the previous crew. Even the “computed airspeed” divergence should warrant immediate fixing.
@Andrew. Thank you for those your advice about the system as it should work and how the aircrew can add requirements. As you say, “Any engineer worth his or her salt would NOT release an aircraft unless satisfied that it is airworthy.” Implicit is that he does not sign off on defects just because the test equipment does not find them and has the skills, equipment and information needed to work to his tasks limits, plus access to the aircrew and technical assistance.
My concern in this case is that with the aircrew accepting a defective aircraft, the ground crew in that culture treated that edict as acceptance that the aircraft was sufficiently airworthy. Release by consensus as it were.
On the AoA effect on airspeed, what I took it Victor and you meant was that the lower computed airspeed recorded for much of the flight was the higher airspeed multiplied by the cosine of the 21 deg difference in the AoA sensors, in other words the ADIRU converted airflow perceived to be at-an-angle, to aircraft forward airspeed. The difference of say 17 knots is similar.
I have my doubts that position error from changing AoA would be that big.
@Andrew. Incidentally neither AoA in-flight reading looks right to me. The captain’s is too high, the co-pilot’s looks too low.
@David
David, I don’t know if you have seen the photographs that purport to be of the Aircraft Flight and Maintenance Log Book for PK-LQP that have been circulating since very soon after the accident. This one shows the Defects raised after JT43; and this one shows the corrective actions that were taken to address the defects.
Perhaps Andrew or Don can offer a more informed analysis of what we’re looking at but what I can see is that:
a. There were only two defects raised after JT43; the ‘IAS & ALT DISAGREE SHOWN AFTER TAKE OFF’ and ‘FEEL DIFF PRESS LT [LIGHT] ILL [ILLUMINATED]. There’s no mention of the uncommanded nose-down trim, no mention of the stick shaker activation, no mention of having to use the STAB TRIM CUT OUT switches, no mention of having to trim manually for virtually the whole flight.
b. Both defects are cleared as ‘OPS TEST ON GROUND FOUND SATISFIED’ and ‘TEST ON GROUND FOUND OK’
There have been a number of reports of another entry by the JT43 crew that reads;
‘A: PK LQP, B737 Max 8
D: 28.10.2018
O: Airspeed unreliable and altitude disagree shown after takeoff. STS was also running to the wrong direction, suspected because of speed difference. Identified that CAPT instrument was unreliable and handover control to FO. Continue non normal checklist of Airspeed Unreliable and Altitude disagree. Decide to continue flying to CGK at FL280, landed safely rwy 25L
R: DPS CGK LNI 043
E: AFML’
I haven’t seen any photographs of that entry so you should perhaps treat it accordingly. However, if the second write up is a true record you might well be struck by the fact that while it refers to completing the non-normal check-lists for Airspeed Unreliable and ALT DISAGREE there’s no mention of the Runaway Stabilizer NNC. Near as I can tell there are only two non-normal occurrences that call for moving both the STAB TRIM CUTOUT switches to CUTOUT and that’s Runaway Stabilizer and Stabilizer Trim Inoperative so you’d think that one of those non-normal occurrences would rate a mention.
The other thing that I’m struck by is that these records are available; one or two people, presumably employed in Lion Air’s Jakarta Engineering base, created digital records of the tech log before and after corrective actions had been taken. I’m still trying to get my head around what that says about the people that took the photographs (Is it a standard tin-plating practice? Was it a once off because they were concerned about that aircraft? What does any of that say about work culture?)
@Mick Gilbert
RE: “And I’m not suggesting that Lion Air aren’t likely to be found partly culpable for JT610, they should be. I just think that the NYT article is hyped.”
Sorry, but I don’t. In their short, 18-year history, they’ve written off six aircraft, severely damaged eight others and killed over 200 people through accidents. You won’t catch me flying on Lion Air or any of the other airlines in that group!
@David
RE: “Implicit is that he does not sign off on defects just because the test equipment does not find them and has the skills, equipment and information needed to work to his tasks limits, plus access to the aircrew and technical assistance.”
I agree, however, it’s a sad fact of life that good old-fashioned troubleshooting skills are fast disappearing in this age of electronic boxes. In most cases, the engineer simply consults the onboard maintenance computer of fault isolation manual, resets a system and then runs a test. Further troubleshooting might involve the re-racking or replacement of an LRU. If the system passes the test, the defect is signed off as having been rectified. A lot of the older engineers despair at the lack of troubleshooting skills of their younger colleagues.
RE: “On the AoA effect on airspeed, what I took it Victor and you meant was that the lower computed airspeed recorded for much of the flight was the higher airspeed multiplied by the cosine of the 21 deg difference in the AoA sensors, in other words the ADIRU converted airflow perceived to be at-an-angle, to aircraft forward airspeed. The difference of say 17 knots is similar.
I have my doubts that position error from changing AoA would be that big.”
No, the ADIRU uses AOA data to correct the static pressure for position error. If the AOA data is erroneous, there will be an error in the ‘computed’ airspeed and altitude. The difference isn’t great, but it’s enough to cause IAS and altitude disagree messages.
RE: “Incidentally neither AoA in-flight reading looks right to me. The captain’s is too high, the co-pilot’s looks too low.”
The AOA measured by the AOA vane is also affected by position error and is not the true AOA. I believe the ADIRU corrects the measured AOA to determine the true AOA (don’t ask me how!). I think the data shown on the FDR report is the measured, or indicated AOA.
@Andrew
You won’t catch me flying on Lion Air or any of the other airlines in that group!
Don’t get me wrong, I wouldn’t be getting on a Lion flight either. I’m not a fan of Indonesian aviation, period. In a previous job I used to fly regularly to a couple of mines in Kalimantan. The usual routing was a B737 from Jakarta to Balikpapan and then either an Indonesian licence built CASA 212 or a helicopter, depending on which mine. Travel days were never particularly relaxing.
@Mick Gilbert. That makes for sorry reading. For example, a test to confirm that the pitot or static systems were the cause of the airspeed mismatch before flushing should have been a first step.
Likewise with the elevator feel computer. Not that much use running a test after cleaning the plug if the test equipment disclosed no fault before.
This is not taking full advantage of test equipment yet treating it as having authority beyond its scope.
“However, if the second write up is a true record you might well be struck by the fact that while it refers to completing the non-normal check-lists for Airspeed Unreliable and ALT DISAGREE there’s no mention of the Runaway Stabilizer NNC. Near as I can tell there are only two non-normal occurrences that call for moving both the STAB TRIM CUTOUT switches to CUTOUT and that’s Runaway Stabilizer and Stabilizer Trim Inoperative so you’d think that one of those non-normal occurrences would rate a mention.”
Yes. My suspicions are that to have recorded that would be to provide evidence that the decision to continue the flight was poor airmanship, at the least (IMO). Lack of a proper write up is itself a serious misdemeanour (IMO) and may have contributed.
Perhaps photographic duplication of records is standard practice. However disclosure of them suggests a motive.
I must admit to jumping the gun with the preliminary report still coming and of course little if any of this is relevant to MH370 but the apparently egregious nature of the actions and inactions which led to this accident are salutary. If borne out lets hope they lead to some house cleaning rather than sweeping under the carpet.
@Andrew. Thank you again. I can see that with programming and electronics playing an increasing part the functions of test equipment will extend. However if it is to be used as an authority its limitations need to be clear. It may be that if some has a checking function integral to airworthiness it needs to be certified with uses prescribed in a way similar to aircraft, though I daresay that is not a novel observation.
One other thing about the likelihood that continued use of 5 deg flaps might have saved the day, that would beg the question as to how that flight would have been written up.
It is curious that he would try that. Almost as if he had it in mind to do so were the previous flight’s trim difficulty to recur. I can understand him retracting the flaps briefly having tried it to check he had found a fix but why not redeploy again when the problem did recur?
Prospects of finding the voice recorder seem to have receded. Do you have any other thoughts as to how that initial flap re-deployment came about?
@Mick
Yes, the age of political correctness makes it difficult to say what we really mean. Likewise, the bounds of professional courtesy make it difficult to say what you really mean relative to what you think about the opinions of others. I was born about 50 years too late.
@David said: On the AoA effect on airspeed, what I took it Victor and you meant was that the lower computed airspeed recorded for much of the flight was the higher airspeed multiplied by the cosine of the 21 deg difference in the AoA sensors, in other words the ADIRU converted airflow perceived to be at-an-angle, to aircraft forward airspeed. The difference of say 17 knots is similar…I have my doubts that position error from changing AoA would be that big.
I believe the pitot (total pressure) sensor is designed to be fairly insensitive to angle of attack over a fairly broad range. As @Andrew says, it is the (incorrect compensation for) position error that causes the disagreement of airspeed and altitude when one of the angle-of-attack sensors is faulty.
DennisW:“We should have done exactly as Peter suggested – searched outward from the 7th arc bounded by reasonable latitude constraints. The latitude constraints suggested by the debris findings (also hard evidence, not an assumption) are 35S to 20S.”
We have been discussing this in 2014 elsewhere and at the beginning of this year here on Victor’s blog:
———-
DennisW: Re: Search strategy
I can’t make sense of scanning full width (+/- 25nm) as the search is extended North. Why not scan a selected arc length (say to 26S) along the arc, and sequentially expand the width? If you postulate a bell shaped probability of terminus distance from the arc, that would result in scanning the highest probability area remaining as a function of time.
Peter: Re: sequentially expanding the width
I made the same suggestion back in 2014.
I guess this would optimize the order (highest probability areas closest to the arc being searched first).
Absent a very high degree of certainty of finding MH370 in a particular segment (which necessarily corresponds to a very low degree of certainty of NOT finding a MH370 in this particular segment, which I think never was the case), I think it would have been more judicious to cover a longer portion of the 7th arc at the expense of more limited width and – if even needed by then – to sequentially expand the width afterwards (not 1:1 however, but more like at a ratio of 1:3 or 1:4 inside/outside).
———-
I guess it was not done this way, because hopes were always put on specific arc segments that appeared more promising ?
With the last Inmarsat data points suggesting an accelerating dive, I would expect the highest probability closest to the arc (with an offset to the outer side). Historic data would legitimate such an approach (see subsequent comment).
“Mathematically Locating Ocean Downed Aircraft”
https://sites.math.washington.edu/~morrow/mcm/mcm15/38724paper.pdf
Distance between crash site and last known position (LKP) at cruising altitude (except JT610):
3 nm ³ 1994-03-23 Aeroflot 593
8 nm ³ 1995-12-06 Khabarovsk 3949
5 nm ³ 1997-12-19 SilkAir 185
8 nm ² 1999-10-31 Egypt Air 990
4 nm ³ 2001-11-19 IRS Aero 9601/02
0 nm ² 2002-05-25 China Airlines 611*
2 nm ³ 2002-12-21 TransAsia 791
17 nm ³ 2005-08-16 West Caribbean 708
3 nm ³ 2006-08-22 Pulkovo 612
9 nm ³ 2007-01-01 Adam Air 574
17 nm ² 2007-01-01 Adam Air 574
6 nm ² 2009-06-01 Air France 447
5 nm ³ 2009-07-15 Caspian 7908
15 nm ¹ 2014-12-28 AirAsia 8501
?? nm 2016-05-19 Egypt Air 804
<2 nm ¹ 2018-10-29 Lion Air 610
* 0nm: LKP within surface debris field
¹ distance from LKP to crash site, source: avherald
² distance from LKP to crash site, source: https://i.imgur.com/kUzCwVp.jpg (measured)
³ “distance traveled from the location of the aircraft at the beginning of the emergency to time of impact”, source: https://www.bea.aero/enquetes/vol.af.447/metron.search.analysis.pdf (p.37)
@David
RE: “Almost as if he had it in mind to do so were the previous flight’s trim difficulty to recur. I can understand him retracting the flaps briefly having tried it to check he had found a fix but why not redeploy again when the problem did recur?…Do you have any other thoughts as to how that initial flap re-deployment came about?”
I doubt the decision to re-deploy the flaps was related to the previous flight’s reported problems. After all, the crew supposedly did not know about MCAS or that it would stop trimming when the flaps were deployed. Further, the alleged maintenance log page for the previous flight does not mention any link between the flap configuration and the automatic trimming.
My guess is that the pilot decided to re-deploy the flaps simply because the automatic trim was apparently normal up until the point where the flaps were retracted the first time. The pilot might have thought ‘well, the aircraft wasn’t trimming the wrong way before I retracted the flaps, so let’s deploy them again and see what happens’.
The flaps were only re-deployed for a short period before they were again retracted, but during that time the aircraft had climbed to 5,000 ft. The pilot might have thought that 5,000 ft was a reasonable altitude to see what would happen if the flaps were again retracted and if the problem could be controlled. I can’t explain why the he did not extend the flaps again when it became apparent that the automatic trim was still trimming nose down. He was initially able to keep the aircraft under control at about 5,000 ft, so perhaps he was considering the next steps to take when disaster struck.
@Victor, Andrew. Thank you. The preliminary report next I think.
@Peter Norton & @DennisW,
Earlier, I replied with some points relevant to the deep ocean search: running survey lines parallel to the 7th arc vs transverse to the arc.
Concerning 7th arc search and searching out from a central point (or line of position): consider the practicalities of deep tow search. Unnecessary winching of the deep tow UVs was avoided. The ships equipped for deep tow operations required 10,000m tow cables with power and fibre optic ‘umbilicals’ woven in, and specifically purposed ‘traction’ winches. The cables weren’t durable enough for repeated, full, winch-in/out operations and repeated turns, if traversing the arc, would have incurred significant loss of productive time. The deep tow UVs were kept at depth throughout most of the 6 week swings, hence the long survey lines parallel to the 7th arc. Fugro’s AUV surveys then filled in the gaps, assuming the gaps were at depths <4500m.
Much of Ocean Infinity's work, south of the Diamantina Escarpment, was conducted with the AUVs traversing the 7th arc so as to align with the long valleys and ridges that lie transverse to the 7th arc.
@the mathematicians here:
Does Mike Chillit’s method hold any water?
https://bit.ly/2PXscGz
I presume not, as something so easy would have been found by thousands of experts on day 1. But is there a simple way to explain (to us layman folks) where/why this method, which seems logical at first glance, fails ?
@Perfect Storm: His methodology holds ZERO water. His construction of the second, “congruent circle” is arbitrary. Whether he realizes it or not, he is constraining the final end point to be on a line that intersects KLIA and is perpendicular to the line that runs between the sub-satellite point and KLIA. That gives a specific location on the 7th arc. But there is absolutely no reason for the end point (or any other point along the path) to be on that line.
You can decide for yourself whether Chillit is incompetent or a fraud.
That said, irrespective of his false analysis, it is possible the plane is where he says.
I’ll say that the Daily Star has twice in recent weeks asked me for comments on their articles on MH370. I refused to have my name associated with their trash, and I told them so. I can’t stop them from lifting quotes from my blog articles, but anybody that works with them is contributing to the problem.
@Perfect Storm
Concerning Chillit.
No.
But it might be illuminating for him to demonstrate how his method works out for all the known locations when the aircraft communicated with the GES, where the centres of his congruent circles might be placed. Maybe you could press him to show that?
For future reference: UK ‘news’ titles including the word Daily in the title should be treated sceptically, perhaps excepting the Daily Telegraph. Their revenue stream is based on placement of a personalised ad on the web page that is rendered to your device/web browser.
@Victor Iannello: Thanks to your explanation I see now that the geometry just looks nice, but is not based on a logical mathematical concept.
@Don Thompson: Good point. If I find a contact address, maybe I’ll point him to this flaw. I admit I’d be curious to see what he has to say about that.
@Perfect Storm: On Twitter, many of us have exposed his math, science, and logic errors. His response is to block us. At this point, we just ignore him.
In order to generate clicks, the Daily Star has been running a series of stories on MH370 that should be immediately dismissed. It’s a great match–individuals proposing false theories that are desperate for attention, and an online publication that has little regard for the truth but needs viewers.
Concerning Chillit . . .
Somehow he has been able to get his rubbish into a number of UK “dailies”, and now he is a “Mathematician” and had developed this new “elaborate math” to determine where the plane rests. Oh, he is also an “expert” in drift studies !
I have tried contacting the writers of some of these articles, and tried to explain the same issue that Don comments on, probably without success.
If the method works for the BTO on the 7th arc, then it should also work for all other BTOs, and particularly for all those where the position of the aircraft is known from the ACARS reports prior to 17:07. If the methodology worked then it would also accurately plot the position of the aircraft at the other arc crossings.
Of course it doesn’t do this.
Note that Chillit keeps revising his position incrementally, and it really comes as no surprise that he nudges towards points that Richard has previously shown to have high probabilities.
@Victor Iannello, Don Thompson: I searched for a contact address but could only find his twitter. I don’t have a twitter account and was about to ask if one of you could post him a link to our discussion here, but if he just blocks you, then it’s obviously useless. It’s a pity, it would have been interesting to see his response.
@Perfect Storm
MC’s email address was posted in a tweet.
Ignore the leading ‘@’
@Don Thompson: Thank you. Thanks to your link I just had an email exchange with him. He gave me an explanation about Euclidean Plane Isometry, but unfortunately I didn’t get permission to share it, and that is something I must honor. It’s a pity, because I would have liked to see if better educated people than myself can make sense of it. No luck. I’m still grateful to all of you for having tried to make sense of what limited information is available publicly (the 4 slides).
@Perfect Storm,
Typical. He uses all the big words to sound like he is very knowledgeable, and to confuse the listener. Mostly these are mis-used and irrelevant. He also tells “porkies” . . . . like his statement that the BTO for Kuala Lumpur had never been released before. Rubbish.
You can view his twitter feed even although you don’t have an account. All the nonsense is there, going back over about 4 years.
Ask him to use his methodology to confirm the aircraft location at 17:07. The BTO is known, and the Lat and Long are available from one of the last ACARS reports. All publically available information.
@All,
Sorry to be silent for a while. After our initial discussions I decided to give my paper a final brush-up before publication. Now it is ready, and you will get a link to it tomorrow.
All I ask for in return is that you read it and (if relevant) cite it according to scientific standards, so I get credit for the work. Sorry in advance for the difficult math it contains. If some of you find this part annoying, look at the figures and read the introduction, discussion and conclusion. I did my best to write these parts clear and simple. However, difficult math is unavoidable to arrive at a rigorous solution to the problem.
@DennisW,
You have several times talked about some equations describing the entire problem, which you could not solve. I believe that I have written down some similar or perhaps equivalent equations. After some troubles, I managed to solve them using a combination of different mathematical methods. If I disregard solutions outside the fuel range, solutions with velocities outside the engineering limits of the airplane, crazy zigzag solutions, and only count multiple roots leading to the same location as one solution, I find exactly four independent solutions to the equations. One of these is identical to the solution proposed in the scientific paper by Inmarsat in 2014. However, the failure to find the airplane in the vicinity has practically proven it wrong.
@Richard Godfrey,
Sorry to disappoint you concerning the writing of a new software code for debris tracing. However, instead I used the ideas from our old optimization software as an important input to analyzing the satellite data and solving the equations I just mentioned.
@Perfect Storm
You might be interested in this Reddit thread a few months ago where Mike Chillit’s satellite technique was discussed:
https://www.reddit.com/r/MH370/comments/94nfi2/planes_resting_place_or_only_conspiracy_express/
@Victor
Of course the unlucky fact is that the POR Inmarsat satellite did not pick up MH370, suggesting MH370 stayed westerly of that satellite range. From Inmarsat’s coverage map, it looks like if MH370 had gone much further north than 20 South and Arc7, it would have been getting close to the overlap region where both satellites could have seen it. This would seem to counter-indicate the Xmas Island theory of Jean-Luc…can I say that?
@TBill wrote “… POR Inmarsat satellite did not pick up MH370…”
Your comment may be construed to suggest that a ‘satellite’ has some active role in selection of the optimal operating region for the AES, and that the ‘satellite’ (actually, the GES) is somehow monitoring AESs that may be approaching its service area, but not yet logged on.
Such notions are misplaced.
The function to determine the optimal Ocean Region executes entirely within the AES.
So your comment might be better written to describe that, after 18:25, there is no record of the AES attempting to execute a handover to the POR region, indicating that it did not approach the overlap region between IOR and POR.
@Viking
@DennisW, You have several times talked about some equations describing the entire problem, which you could not solve.
I think it is more accurate to say I have had brief love affairs with integrated Doppler and Geometric Algebra/Calculus. I could never derive a “clean” integrated Doppler approach, and the Geometic Algebra did not yield anything convincingly different than Gibbs (it was fun, however, and I have become a fan of it, and David Hestenes, for physics education).
Look forward to seeing your work.
MC reminds me of the Godfather of Fake News:
https://www.bbc.co.uk/news/resources/idt-sh/the_godfather_of_fake_news
In a story in The Australian, it’s reported that:
A presentation to the Indonesian parliament on the Lion Air crash, has revealed the pilot basically fought with the aeroplane to the end of the flight….He explained that the MCAS trimmed at a faster rate than the pilot-commanded trim, making it increasingly difficult to control the aeroplane…It is believed the pilots on Flight JT610 had prepared for an unreliable airspeed event prior to takeoff, and believed that was the issue when they requested to return to Jakarta.
Although we don’t have the full remarks from Indonesia’s NTSC, it appears that there were no remarks relating to how a pilot is trained to identify and recover from runaway trim, nor any mention of any potential shortcomings in maintenance or risk management.
@Don Thompson
Thank you. But do we (or Inmarsat) know where we might have expected a change to POR stellite? That could give us some hints where MH370 might not have gone.
@TBill: Since the satellites orbit nominally above the equator, the closest satellite to an AES is determined by the longitudes. IOR is nominally at 64E and POR at 178E, which means that for longitudes west of 121E, IOR is closer. The 7th arc crosses the equator at around 109E. That means, at all locations on the 7th arc, IOR is closer than POR. So in a nutshell, we would not expect a change from IOR to POR at any point along the trajectory of MH370, as it would mean changing to a satellite that is lower on the horizon.
@Victor @Don
Good info…thank you.
TBill: I concur with the comments by Don and Victor. However, it should also be noted that there are several factors affecting the AES selection of a satellite other than elevation angle. For example, the prior MH371 flight switched back and forth between the POR and IOR birds a dozen times during that flight. It was in view of both for the whole flight. These changes are normal, particularly if traffic is congested on one satellite more than the other. What we know is that MH370 only used the IOR satellite, and the AES can only logon to one GES at a time, even if the AES transmissions are received and re-transmitted by the adjacent satellite (POR in this case). Due to the directivity of the AES HGA, the signal received by the adjacent satellite will be very weak, thus allowing the reuse of the frequency in the POR network.
Oh, maybe it was a mistake, but I decided to challenge MC with a little test of his Isometric Reflection nonsense, in an email to him a little earlier today.
He has responded via Twitter in a way that makes me really angry. He is totally un-trustworthy. But I already knew that.
@Brian Anderson: Yes, it was a mistake. He is incorrigible.
Somebody on Twitter once gave me this advice: “Never play chess with a pigeon. He’ll just crap on the board and strut around like he won.”
@TBill: Thanks for the reddit link. Here is the discussion on aviation.stackexchange:
https://aviation.stackexchange.com/questions/57441
Same final outcome, it was closed as “badly defined/justified method a random person on the internet came up with”.
I was hoping for a public discussion here or over there, but I think that ship has sailed.
The way Viking teases and hypes his work before publication has me on the edge of my seat. I hope it’s as good as he makes it sound like.
@Perfect Storm: We don’t discuss nonsense here, and Chillit’s “math” is nonsense, as anybody with any skill will quickly tell you.
Re JT610, reported in The Seattle Times:
Lion Air pilots struggled against Boeing’s 737 MAX flight-control system, black box data shows
@All,
Here is the link to my paper: http://arxiv.org/abs/1811.09315
I will be very busy tomorrow, so my answers may not come immediately.
@Perfect Storm,
I can hardly resist the temptation to comment on MC’s model. I agree it is nonsense, but fundamentally it is not a totally crazy idea. The problem is that he does not know how to do the math correctly. He has committed a mistake equivalent to making a non-monotonous substitution in an integral.
@Andrew: Thanks for that link to the article. That’s the best one I’ve seen written so far. It’s also eerily similar to the discussions we’ve had here.
@TBill, @Don Thompson,
I really enjoyed your comments on the POR satellite above. Several years ago I was surprised that the received power at the satellite goes up as the travelled distance increases. Since there is both diffraction loss and atmospheric loss this is counterintuitive unless the flight is directed within a narrow angular interval, where the combination of weather-induced phenomena and the optimized reception pattern (for optimum commercial coverage – represented by the Inmarsat satellite coverage) work together.
This favors routes near Christmas Island.
@Victor
My thoughts exactly. I was interested to read Peter Lemme’s comments regarding the certification basis and the difference between “hazardous” and “single point catastrophic failures”. It seems the failure analysis assumed that pilots would automatically invoke the runaway stabiliser checklist in the event of a problem with MCAS, even without specific knowledge of the system. That’s all very well and would probably work in most circumstances, except when a crew is distracted, confused, task saturated or simply doesn’t perform so well.
@DennisW,
I also tried an integrated Doppler method and arrived at roughly the same solutions as in my paper but with less precision. It is most easy to solve (the integrated Doppler) by rephrasing the problem as an angular integral and performing a numerical integration of the angular velocity.
@Andrew. Reportedly an Australian Boeing 737 check and training captain’s view, including, “What is an issue is why the Lion Air plane was allowed to fly in the first place…. In Australia, it would have been grounded.”
https://www.airlineratings.com/news/passenger-news/lion-air-pilots-aware-new-737-flight-system/
Re the effect of flaps and your, “……except when a crew is distracted, confused, task saturated or simply doesn’t perform so well.”
I notice thrust was reduced about 10 sec before raising flaps the second time. Subsequently thrust was adjusted (experimented with?) with before full thrust was selected finally. How that fitted in with the turn back intention is unclear but it might indicate confusion.
@Mick Gilbert. Would you expect those Flight and Maintenance Log defect write-ups to be in English?
@Viking: Thanks for your paper. I stand corrected. Looking forward to read it.
re: “I can hardly resist the temptation to comment on MC’s model. I agree it is nonsense, but fundamentally it is not a totally crazy idea. The problem is that he does not know how to do the math correctly. He has committed a mistake equivalent to making a non-monotonous substitution in an integral.”
Oh no. You have figured out how to intrigue me. Please tell me more. Can you correct “the mistake”? If the idea is “not totally crazy”, is there a way to improve it so that it could tell us something about the flight path ?
Many thanks
@David
RE: “What is an issue is why the Lion Air plane was allowed to fly in the first place…. In Australia, it would have been grounded.”
I agree with that comment. The statement that “all airlines were briefed on the new feature and the detail is in the pilot manuals” is odd, given the reported reactions of the US pilot unions.
@Perfect Storm,
Quick comment: Follow the suggestion on angular velocity I just gave to DennisW.
@Viking: Thanks. I hope you can elaborate at another moment.
What would the “corrected MC method” be able to tell us ?
@Viking,
I’m particularly interested in the points you make regarding the point of closed approach to the satellite, para 2 in your paper.
It seems to me that there is ample evidence that the BTO data at and beyond 18:28 should be reliable. Using this data a number of us have independently determined that the point of closest approach to the satellite can be deduced from the minimum in the polynomial linking these data points, or in fact the derived LOS distances from the satellite. The time for this minimum can also be determined, and we have assessed that at about 19:52 utc.
This is also in agreement with the very early rough assessment based on the very fuzzy graph of elevation angles, the infamous Lido slide, which preceded any real BTO information.
Of course one has to decide if the BTO trend was still decreasing at 19:41, or increasing at that point.
How can you show that it was increasing at 19:41 ?
@David
RE: “Would you expect those Flight and Maintenance Log defect write-ups to be in English?”
Forgive my intrusion, but I would say ‘yes’, simply because Lion Air employs a lot of non-Indonesian pilots.
That said, I question the authenticity of the log page that has been put up on the Internet. The page is supposedly for the sector from DPS to CGK (ie JT043). However, the Maintenance Release and PIC Acceptance blocks appear to have been signed by the DPS engineer and the PIC of JT043. Perhaps Lion Air has a different system, but those signatures would normally be on the previous log book page, a copy of which would have been left in DPS when the aircraft departed. The Maintenance Release and PIC Acceptance blocks on the alleged log book page should be blank, because it appears that no corrective action had been taken when the image of the page was recorded. The signature blocks would have been completed by the CGK engineer when the aircraft was released for service in Jakarta, and the PIC of JT610 when he accepted the aircraft in Jakarta.
PK-LQP Preliminary Report. I haven’t managed to download it yet; the site is very sloooooow.
@Andrew. Yes re sigs. Yes re sloooow.
The photographs that Mick posted wrote up the 28.19.2018 DPS to CGK flight as 093, not 43. Departure (out) time was 1400, arrival (in) 1602, flight time 2:02. I can find no 093 on that route. 43 departure DPS is 1930, arr CGK 2020, flight time 1:50, presumably time zone change.
Mmmmm.
@David
I think that ‘9’ might just be a handwriting thing; it could be a malformed ‘4’. The times could also be the result of a delay; I don’t find that particularly odd. The things I do think are strange are the two signature blocks at the bottom, as I mentioned previously. Perhaps they do it differently at Lion Air!
Viking: Regarding your comments above:
1. “Several years ago I was surprised that the received power at the satellite goes up as the travelled (sic) distance increases.” I assume by this statement you mean that as the AES to S/C range increases, the AES uplink signal strength received at the S/C increases. If that is what you mean, it is wrong in the case of the I3F1 L-Band recv global beam. The AES power flux density decreases as 1/R^2. In addition, the global beam antenna gain is max at the sub-satellite point and drops off by -2.5 dBIC dB at the horizon. So, for a constant AES EIRP, the AES signal received at the satellite decreases for 2 reasons as the AES moves away from the sub satellite point.
2. “Since there is both diffraction loss and atmospheric loss this is counter intuitive unless the flight is directed within a narrow angular interval, where the combination of weather-induced phenomena and the optimized reception pattern (for optimum commercial coverage – represented by the Inmarsat satellite coverage) work together.” This is misleading at best; mostly nonsense. For 1.6 GHz, the atmospheric loss is <0.5dB, and the diffraction loss at the elevation angles involved with MH370 was zero.
It should be noted that the AES EIRP is "power agile", meaning it is continuously adjusted by the system to compensate for all link losses, regardless of the cause. The system adjusts the AES EIRP to maintain a near constant receive signal strength at the satellite LNA.
@David
The Preliminary Report states that the PIC of JT043 only reported the IAS & ALT DISAGREE alerts and FEEL DIFF PRESS light, as shown on the log book page that’s been circulated. He also made a verbal report to the engineer, but there’s no mention of what was discussed. The report criticises the JT043 PIC’s decision to continue to CGK and the lack of detail in his defect reporting (rightly so!).
@Andrew. Much as you say, local press summary has NTSC reaction mild and explanatory. 6 problems from Oct 26.
https://news.detik.com/berita/d-4320593/sejak-26-oktober-lion-air-pk-lqp-punya-6-masalah
https://news.detik.com/berita/d-4320622/beri-sinyal-pan-pan-begini-masalah-lion-air-pk-lqp-saat-di-denpasar
https://news.detik.com/berita/d-4320624/hidung-pesawat-naik-turun-ini-kondisi-lion-air-pk-lqp-sebelum-jatuh
https://news.detik.com/berita/d-4320690/knkt-ada-beda-data-jumlah-pramugari-lion-air-pk-lqp
https://news.detik.com/berita/d-4320760/knkt-lion-air-pk-lqp-denpasar-jakarta-harusnya-tak-lanjut-terbang
https://news.detik.com/berita/4320769/2-rekomendasi-knkt-untuk-lion-air-pascajatuhnya-pk-lqp
https://news.detik.com/berita/d-4320834/aneh-ada-beda-data-stall-di-pilot-dan-kopilot-lion-air-pk-lqp
@Andrew. “The times could also be the result of a delay; I don’t find that particularly odd.” The recorded times were more than 5 hours ahead of 43’s schedule. I assume they were the local time at one end since there is an hour o time zone difference between the two.
Even so the defects recorded match what the NTSC says.
But then there are the sign offs at the bottom you mentioned.
Not yet in the clear I think.
@David
The times recorded in the log book are UTC. According to the log book the aircraft departed DPS at 1400 UTC (2200 Local) and arrived CGK at 1602 UTC (2302 Local). The scheduled departure time is 1930 Local, arrival time 2020 Local.
@Viking
It looks to me you are using uncompensated target Doppler values in your paper, which looking at the values originate from the sample route calculation in the Ashton paper. In my understanding the raw uncompensated Doppler values are unknown (we only know Fup+fcomp by approximation). Your notes after proof partially address this; as a suggestion: the compensation algorithm is for example described in the DSTG book (Bayesian methods…). Also interesting is the paper by Henrik Rydberg. In addition I’ve done some combined BTO/BFO analytics. I will send Henrik’s and my paper to you by mail.
@Andrew. Thanks. That explains that bit.
Some hours back with the report mid-ether I posted a local Indonesian account of the NTSC briefing on it. Unfortunately it had enough URLs to warrant Victor’s moderation.
Now overtaken I recommend disregarding them when and if they appear.
For those that haven’t time to wait for the NTSC’s slow servers, here is an alternative download site for the PK-LQP Preliminary Report.
Can’t believe the crew of JT043 made no tech log entry of the STAB trim switches being flicked off or that the stick shaker operated continuously! If they had, the JT610 crew would have quickly isolated the trim too.
@Perfect Storm,
I do not think the corrected MC method can tell us anything new. However, it may provide a simple crude overview to display solutions.
@Brian Anderson,
You can see from the Inmarsat paper that 19:41 has a sign change of the Doppler shift compared to the rest of the points for the southern solution.
If one uses a higher order polynomium to fit the data, one may manage to get it on the same curve as the rest for both BTO and BFO, but that will not be following a great circle.
My point is that any scenario (accident, suicide, terror, scam, etc.) will usually contain a relatively long segment with flight along (or practically along in case of magnetic heading) a great circle with constant speed and height. The starting point is to determine beginning and end for this ‘straight’ part of the flight, and analyze it first. Then the rest can be pieced together.
Figure 2 in my paper is designed to use a combination of BTO and BFO information to look for deviations from a ‘straight’ flight, and I find a clear deviation at 19:41.
@airlandseaman,
I do not know how well the SDU can compensate for the power loss. However, I am quite sure it is unable to compensate weather related issues. If you look at weather satellite results the plane passes through the tropical inter-convergence zone near a thunderstorm at 21:41 on the route I propose. This leads to damping of the signal.
@Niels said: In my understanding the raw uncompensated Doppler values are unknown (we only know Fup+fcomp by approximation)
Yes, this is a fundamental flaw in @Viking’s paper. We only know the “residual Doppler shift”, i.e., the shift that remains after the pre-compensation by the AES. (The other components of the BFO are path independent.)
The AES pre-compensation algorithm is well-known, as described in the Inmarsat JON publication, the ATSB publications, and the DSTG Bayesian analysis. Since June 2014, many people on this blog have been fairly accurately calculating the BFO for hyptothetical paths.
My recommendation is that you (@Viking) consider the full BFO model in your paper (among other things). Otherwise, the wrong sensitivities of the BFO to horizontal speed, track, and vertical speed will be calculated. For instance, the BFO is sensitive to the N-S horizontal speed, but almost completely insensitive to E-W horizontal speed, as it is almost completed compensated by the pre-compensation algorithm. On the other hand, the vertical speed is completed uncompensated by the AES, and therefore the full Doppler shift due to this component of AES velocity appears in the BFO.
@Viking
Your paper starts off being well presented, but descends into a story built around a cast of thousands.
1. Your route fits the BTO but diverges from the BFO significantly towards the end.
2. Your end point does not fit the drift analysis. I am running a simulation just to find out how big the error is.
3. Kate Tee is an unreliable witness, who has changed her story. Even the GPS Log from the boat is not certain, as we are not sure which time zone was set.
4. The barnacle growth can fit a wide range of latitudes and timings and is not a helpful differentiator.
5. Anyone quoting Chillit is suspect. Quoting Wise is not much better.
6. You quote Inmarsat and Victor heavily, but have not fully understood either.
7. Your maths checks out as far as my knowledge goes but the data is wrong as Niels and Victor have pointed out. He initially assumed a constant earth radius, only to later make an oblate correction. A bit of a contortion! (pun intended).
8. I am not sure Brian Anderson is convinced either.
@David
@Mick Gilbert. Would you expect those Flight and Maintenance Log defect write-ups to be in English?
Apologies, a tad late to the dance today. I see that Andrew has addressed that.
The Preliminary Report seems to attest to the veracity of the photos and the pilot’s flight condition report as submitted through Lion’s A-SHOR system.
My initial take-aways are:
• The Captain and First Officer had over 9,400 hours on B737s between them, they weren’t inexperienced.
• 443 cycles on the aircraft in just 94 days from first flight, 74 days from registration. Lion certainly fly their airframes hard. That almost certainly means that there’s pressure on Engineering to get them back into the air. It probably also means there’s pressure on flight crews to meet schedules.
• Each time Engineering touched that aircraft they seemed to make things worse; after each ‘fix‘ the next flight had more or more serious problems.
• One of the big issues for JT43 when they experienced their MCAS event was control column forces. The FO ‘commented that the control column was too heavy to hold back‘. At the onset of JT43’s MCAS event the control column forces rise rapidly to around 85 (out of 100) of whatever the units of measure are. For JT610, until the terminal 90 seconds or so, the forces never get above 50 units and tend to bounce around the 25 mark.
• There’s already been comment on the appallingly (perhaps criminally?) deficient write up of the defects arising on JT43. Reading between the lines, that might have been to avoid attention being drawn to their decision to continue on to Jakarta rather than retuning to Denpasar. What more can you say about that?
• When the Captain of JT610 accepted that aircraft he was aware of the IAS and altitude disagreement problems and the fact that the speed trim system was ‘… running to the wrong direction, suspected because of speed difference …’. He had no idea that the stab trim cutouts had been necessary on the previous flight.
• There’s evidence that JT610’s crew then did what any professional crew would do and briefed for an unreliable airspeed event on take-off. I don’t think that it is happenstance that JT610 ‘… requested approval to the TE controller “to some holding point”’ as well as advising ‘… the TE controller that the intended altitude was 5,000 feet.‘ and that they re-retracted flap at 5,000 feet. There is level flight pitch and power data for holding, flaps up at 5,000 feet in the Flight With Unreliable Airspeed tables.
• After take-off, the IAS DISAGREE appears to have manifested first. They didn’t encounter their first MCAS induced problem until after they had requested to hold at 5,000 feet. The MCAS trim issue was subsequent to the initial unreliable airspeed issue. They’re effectively presented with the problem that they had briefed for; there’s a lot to be said for the ‘priming’ effect and how that shapes their perception of everything that happens subsequently.
• The decision to re-extend the flaps after the first MCAS induced upset ‘masks’ that problem for the remainder of their climb. That’s possibly the phase of the flight where the MCAS induced trim would have made the control column forces difficult to control. Glass half full – it didn’t kill them then. Glass half empty – they missed the same imperative as JT43 to do something urgent about it like move the stab trim switches to cutout.
• Between 23:25:05 UTC and 23:29:45 UTC air traffic control gave them something like six or seven instructions relating to heading changes. For a crew dealing with a flight control problem I can’t imagine that that would have been helpful.
• Between 23:30:58 UTC and 23:31:09 UTC there appears to have been a change in the pilot flying from the Captain to the FO. Things go horribly awry around that time.
Viking says: “the corrected MC method (…) may provide a simple crude overview to display solutions”
@Viking: how would the corrected method look like geometrically?
Victor pointed out that “he is constraining the final end point to be on a line that intersects KLIA and is perpendicular to the line that runs between the sub-satellite point and KLIA. That gives a specific location on the 7th arc. But there is absolutely no reason for the end point (or any other point along the path) to be on that line.”
How can this be corrected (geometrically) ?
Viking: Re your staement above: “I do not know how well the SDU can compensate for the power loss. However, I am quite sure it is unable to compensate weather related issues. If you look at weather satellite results the plane passes through the tropical inter-convergence zone near a thunderstorm at 21:41 on the route I propose. This leads to damping of the signal.”
Sorry, your are wrong about how the link is maintained, and weather attenuation. The AES does adjust for “weather”, which at L band is virtually zero attenuation anyway.
preliminary report, p. 36 (45): “Have a SAFE fight!”
what bitter irony considering the unsuccessful fight …
@Victor
Reuters News story- MH370 debris and NOK backj in the news
https://af.reuters.com/article/commoditiesNews/idAFL4N1Y340D
prelim.report p. 53 (62): Boeing: “The function is commanded by the Flight Control computer using input data from sensors and other airplane systems.”
Boeing being deliberately vague here ?
I mean, the intention was to clarify the MCAS function …
Isn’t one of the key questions how many sensors and what other input data (if any) get to decide the nose down command ?
From The Australian quoted by Victor above:
« A presentation to the Indonesian parliament on the Lion Air crash, has revealed the pilot basically fought with the aeroplane to the end of the flight. [NTSC] head Nurcahyo Utomo explained the data retrieved from the flight recorder. […] “after the trim down, the pilot commanding electric trim continued to fight until the end of the flight.” He explained that the MCAS trimmed at a faster rate than the pilot-commanded trim, making it increasingly difficult to control the aeroplane. “It was noted at the end of the flight, the automatic trim increased, but the trim of the pilot was shorter” »
@Andrew:
Since the crew did not use the cutout switches (or they malfunctioned?), I would have 2 theoretical questions about “MCAS trimming at a faster rate than the pilot-commanded trim”:
(1) Is the column trim switch enough to override MCAS input on a long term basis? Even when “the automatic trim increased”? I mean, could the crew have safely landed somewhere just using the trim switch, or would there have been a moment (when “the automatic trim increased”) where the trim switch would not have been enough anymore to counteract MCAS AND trim, inevitably leading to a crash if the cutout switches are not used ?
(2) In case it is possible to override MCAS on a long term basis by using the trim switch alone: Can the trim switch only stop further MCAS nose-down trim or also successfully apply nose-up trim (i.e. reduce the nose-down trim that had been applied by MCAS) if used long enough? In other words, if the pilot had failed to entirely counteract the MCAS nose-down trim, is it still possible afterwards (for example when the other pilot takes over control) to make up for that using the trim switch alone?
I just realized that neither the Boeing FCOM Bulletin (prelim.report p.52 (61)) nor the Boeing Multi Operator Message (next page) explicitly answers this point.
@Victor, and all the other IG members,
I’ve just read MC’s total drivel on his recent Tweets. What a nasty deluded person.
I apologise for having inadvertently dragged you into this argument, and I assure you I am not continuing any discussion with this F***wit, ever, again, ever . . .
Of course all his accusations are complete nonsense, as are his geometric constructions and math.
[but it is clear that he does read this blog . . . ]
@Brian Anderson: He has a long history of interpreting corrections to his math, physics, and logic as personal attacks. I assure you that his behavior will not change. Nobody that is informed in this subject matter agrees with the methodology of his path reconstructions (although I do believe it’s possible the plane will be found near his location). He is best ignored.
@Peter Norton
RE: “(1) Is the column trim switch enough to override MCAS input on a long term basis? Even when “the automatic trim increased”? I mean, could the crew have safely landed somewhere just using the trim switch, or would there have been a moment (when “the automatic trim increased”) where the trim switch would not have been enough anymore to counteract MCAS AND trim, inevitably leading to a crash if the cutout switches are not used ?”
The reason the “automatic trim increased” is because the pilot reduced both the frequency and duration of his opposite direction trim inputs at the end of the flight. Consequently, the stabiliser incrementally moved closer to the nose down limit, to the point where the pilot lost control. It is NOT a case of the trim switch not being enough to counteract the MCAS.
The MCAS is designed so that it stops whenever the pilot uses the control column trim switches and restarts 5 seconds after the trim switches are released. Theoretically, the pilot could maintain control by applying opposite direction trim every time MCAS operates. However, landing the aircraft would not be an easy task and the pilot would need to stay on top of the MCAS trim changes to prevent a runaway.
RE: “(2) In case it is possible to override MCAS on a long term basis by using the trim switch alone: Can the trim switch only stop further MCAS nose-down trim or also successfully apply nose-up trim (i.e. reduce the nose-down trim that had been applied by MCAS) if used long enough? In other words, if the pilot had failed to entirely counteract the MCAS nose-down trim, is it still possible afterwards (for example when the other pilot takes over control) to make up for that using the trim switch alone?”
Yes, the trim switch works normally and overrides any MCAS inputs, provided the pilot uses the trim long enough to move the stabiliser in the opposite direction. Boeing’s FCOM Bulletin states: “The nose down stabilizer trim movement can be stopped and reversed with the use of the electric stabilizer trim switches…”
@Viking
Thx.
I would characterize your paper as a serious effort to be sure.
Some of the feedback you have already received I would agree with. Although I don’t think any of it dramatically alters your conclusions.
My biggest concern is with Richard’s drift analytics which I have a very high regard for. It would be hard for me to endorse a terminus North of 20S at this time.
@Andrew: thank you. I hope they’ll find the CVR. I had been wondering whether the crew could have been “overwhelmed” by the MCAS trim, technically unable to resist. But reading your answer, they should have survived even if in panic-tunnel-vision they had forgotten about the cutout switches, by just using the column trim switch regularly (… or maybe just extending the flaps).
Speaking of which … as discussed previously extending the flaps seems to have stopped the MCAS nose-down trim. The pilot apparently noticed that, since he re-extended the flaps a second time after take-off once the MCAS nose-down trim initially kicked in. So with flaps extended for landing, why “landing the aircraft would not be an easy task and the pilot would need to stay on top of the MCAS trim changes to prevent a runaway” ?
@TBill. Thanks for the Reuters’ article about new MH370 debris, to be delivered personally to the Malaysian transport minister tomorrow. Perhaps those pieces then will be looked into.
However, though there has been no final report issued, I have my doubts. The vortex generator base and the item accompanying it have received an interest conspicuous by its absence. Reportedly these have been available from Madagascar for many months. Even if not surely they are available for inspection?
@David: The vortex generator base and the other item remain in Madagascar as evidence in the investigation looking into the death of the Malaysian diplomat.
I don’t think the ATSB will look into the new debris, as the MH370 investigation team has been reassigned. I don’t know if Malaysia has the capability or willingness to do much with the parts. I wouldn’t get your hopes up.
On a positive note, at least one of the new parts has been identified (not by me), and the part does provide some additional evidence about the nature of the impact. More on that soon.
@Peter Norton
Yes, the CVR will answer a lot of questions if/when it’s found. They should have been able to maintain control of the aircraft.
Good point re the flaps inhibiting the MCAS – you are quite correct. I should have given that some more thought before pressing ‘Submit’. The ND trim problem would have ceased when the flaps were extended during the approach. The subsequent approach and landing would have been ‘normal’, apart from the persistent stickshaker and IAS/ALT disagreement, neither of which would have prevented a safe landing.
@all
concerning p.1-3 of the report: I was surprised to see so MUCH radio communication.
I don’t know if “aviate – navigate – communicate” is to be interpreted sensu stricto, but if find myself in a critical situation, having lost control over my aircraft and not knowing yet how to solve the problem, wouldn’t that me my utmost priority ?
I imagine constantly talking to ATC in that situation was distracting the pilots from talking to each other, which they should have done to solve the problem, no ?
I remember reading Sullenberger maintained radio silence after deciding “We’re gonna be in the Hudson” to concentrate on flying.
@Andrew,
I sense that just holding the trim wheel would have stopped any MCAS induced trim, either NU or ND. Manually turning the trim wheel would also over-ride the electric driven trim.
You might remember the AF447 scenario where the THS wound itself to essentially max NU trying to accommodate the elevator demand, and it could have been stopped at any time by just holding the THS trim wheel.
My suspicion is that the manual provision will always over-ride the electric/hydraulic drives.
Things that just turn themselves regularly, seem to have a mesmerizing affect, and the reason they do so seems to be forgotten in the heat of the moment.
@Barry
Does it matter?? 189 people are dead. Ask yourself why.
@Dennis
I agree, it’s immaterial at this point in time.
I’m not try to put blame on the accident crew; they were misinformed regarding the aircraft’s previous flight. However, we are talking about the unknowns of the Boeing MCAS / STS auto trim system.
If all details of its operation were fully known, this discussion wouldn’t be taking place.
@Barry
I agree.
@Perfect Storm
After this comment I will follow the advice by many Bloggers concerning MC’s model. However, I think you deserve an answer to your question. You can fix many problems by giving up the constraint that the mirror satellite must be on equator.
@Barry Carlson:
re: holding the trim wheel
I’m also looking forward to Andrew’s reply.
« I sense that just holding the trim wheel would have stopped any MCAS induced trim. [On] AF447 the THS wound itself to […] max NU trying to accommodate the elevator demand, and it could have been stopped at any time by just holding the THS trim wheel. My suspicion is that the manual provision will always over-ride the electric/hydraulic drives. »
Are you sure it’s that easy to stop the trim wheel with your hands?
In theory apparently it should be. From AA’s 737 OpMan:
« Manual rotation of the trim wheels can be used to override autopilot or main electric trim. The effort required to manually rotate the stabilizer trim wheels may be
higher under certain flight conditions. Grasping the stabilizer trim wheel will
stop stabilizer motion. »
But having seen the videos of how fast the wheel spins and reading this discussion where multiple people describe it as a “knee-capper” and “this looks to be about as safe as sticking your hand right on a running drive belt in an automobile”, I have a hard time imagining how these – quote – “sawblades” can be touched, much less stopped. Of course, if I am in a life-or-death situation, I might not have a choice (if I forgot about the other options, such as the cutout switches, or if for some reason nothing else works).
Interesting part from the same discussion:
>> I remember we were told that one of the reasons Boeing put that big trim wheel
>> in the cockpit like that was so that in the event of a runaway trim malfunction,
>> you could try to grab and hold the wheel until you got the circuit breaker pulled.
>> I never had the “opportunity” to test that theory, and I’m glad I never did.
>
> It is possible, but you must be brave. I’ve done it in the sim and any half-a**
> attempt will only get you hurt. Give it 100% and you’ll stop the wheel –and the
> runaway trim. But it is easier, quicker and safer to the pilots, to just use the “stab
> trim cutout switches.”
@DennisW
Thanks for you very constructive comments. I completely agree that agreement with drift analytics is the only serious concern. This is also the reason why I initially entered mostly into debate on this subject.
Viking: “You can fix many problems by giving up the constraint that the mirror satellite must be on equator.”
@Viking:
How does a mirror satellite help? I don’t understand that.
If you know any geometrical method that is helpful in any way, then please, by all means, present it here.
@Victor. Blaine Gibson apparently was misinformed about the availability of the vortex gen base for collection. I wonder how that came to be?
https://thewest.com.au/news/mh370/malaysia-agrees-to-collect-mh370-wreckage-from-madagascar-which-was-subject-of-zahid-raza-assassination-ng-b88861923z
At one point Blaine thought there to be a connection between the Consul’s assassination, his custody of the two items and/or his impending forwarding of them to Malaysia. Blaine later cooled on that yet there is an appearance that this prolonged murder investigation has seen to it that they have not been part of the Malaysian investigation.
It seems possible that Malaysia could have got access. They could at least have appeared to have tried, yet they have made no mention of these items in any report.
I do not see how the ATSB concluding the task allotted to it should halt continuing Malaysian responsibility for investigation, though that is flogging a dead horse it seems.
Irrespective, it is encouraging that these new finds might yield added evidence.
@Richard Godfrey
I am very happy that you will make new simulations with my end position as input.
@Perfect Storm,
As I said I will not enter into more discussions on MC’s model at this stage. However, you may get some qualitative inspiration from this link to a well-proven method from electrostatics:
https://en.wikipedia.org/wiki/Method_of_image_charges
@Peter Norton
RE: “I don’t know if “aviate – navigate – communicate” is to be interpreted sensu stricto, but if find myself in a critical situation, having lost control over my aircraft and not knowing yet how to solve the problem, wouldn’t that me my utmost priority ?
I imagine constantly talking to ATC in that situation was distracting the pilots from talking to each other, which they should have done to solve the problem, no ?”
Absolutely. The utmost priority in any emergency situation is to FLY THE AIRCRAFT. The incessant instructions from ATC would not have helped the situation. Mind you, the crew didn’t help themselves either. At no point did they declare an emergency or communicate the severity of the problem to ATC. If they had declared a MAYDAY, ATC would have immediately understood the aircraft was in distress and could have cleared all other aircraft out of the way. Controllers are trained to deal with such situations and, in my experience, do everything they can to assist, including keeping quiet when told to ‘standby’. However, they need to know there’s a problem! The crew of JT610 did mention a flight control problem and later advised ATC they could not determine their altitude, but I don’t think that was enough for ATC to appreciate the severity of the problems the crew encountered.
@Viking: Thank you for the link. I am not asking you to discuss his model as apparently it does not have a deeper meaning.
I was just asking if there is *any* geometrical method that might be helpful in any way?
If so, I don’t see a reason for not presenting it.
Anything that is helpful … well, is helpful.
@Richard Godfrey
Concerning citation of various sources: I think anyone who has made a significant contribution deserves a citation, even if you disagree with his/her conclusions.
At this moment I will only comment in detail on Kate Tee: I read her blog quickly very early in 2014 (May/June, I think), and did not find any reason to bother. However, after completing my CI solution in early 2017, I wanted to have a second look. Surprisingly, I found almost perfect agreement with results from her GPS log. This led me to also check her distance estimate with a diffraction calculation, and again I found nice agreement. Finally, I realized that she had been extremely honest about her uncertainty on the exact time from the beginning.
In conclusion these observations completely changed my mind on her observations. Now I am convinced that she is honest and reliable, but unfortunately without scientific training.
@All
Is somebody able to check her observation of two flights crossing each other from the opposite direction with the new timing? I guess they might for instance be from Singapore and Perth. If this also fits, I think it will make a big difference.
@Barry Carlson
@Peter Norton
RE: “I sense that just holding the trim wheel would have stopped any MCAS induced trim, either NU or ND.”
Theoretically yes, but as @Peter mentioned, the trim wheel on the B737 can spin at an alarming rate. I think most pilots would balk at trying to stop it with their hand.
The following video shows how fast the trim wheels can move when trimming at their highest speed. The trim rate is slower when the flaps are retracted:
https://youtu.be/MfhtGhsg-hY?t=92
The ‘grasp and hold’ technique that must be used to stop the trim wheel:
https://www.youtube.com/watch?v=cQirIH_DuAs
That’s not easy to do, and makes it very difficult for the other pilot to use the thrust levers. It would be far easier to disable the trim via the CUTOUT switches on the centre pedestal and it’s a mystery why the crew of JT610 did not do so.
@Andrew: thanks for confirming. Exactly what I had suspected.
> However, they need to know there’s a problem! The crew of JT610 did mention a
> flight control problem and later advised ATC they could not determine their
> altitude, but I don’t think that was enough for ATC to appreciate the severity of
> the problems the crew encountered.
Well, the ATC controller marked LNI610 as having “FLIGHT CONT TROB”, but I agree with you nevertheless. That was at 23:24:51 UTC, exactly 2 min after the flaps were extended. There was still a whole lot of dispensable radio coms after “FLIGHT CONT TROB”.
Something else that I find very intriguing:
The SIC had the presence of mind to “ask the TE controller the speed as indicated on the radar display.” (p.2). This was 8 sec after extending the flaps (for the second time), so panic could have still developed later on. But (at least until this point) it seems that the crew (or at least the SIC) didn’t suffer from tunnel vision but – quite to the contrary – was looking for solution slightly out of the box. The curious part is that they came up with (slightly) creative ideas, but didn’t think of the trim runaway memory items, which wouldn’t have required them to be creative at all. It doesn’t compute for me.
@David: Likely there were several contributing factors that led to the assassination of the diplomat in Madagascar. In light of the coincidental timing, we have to consider the link to the debris as one factor. I don’t think many people disagree with that possibility. I don’t know how aggressively Malaysia has asked for the debris. Based on their past behavior, my guess is they are not trying very hard to collect it. Luckily, the vortex generator part was identified by independent investigators outside of the official investigation.
@Peter Norton
RE: “The curious part is that they came up with (slightly) creative ideas, but didn’t think of the trim runaway memory items, which wouldn’t have required them to be creative at all.”
I suspect they were focused on the UAS problem and were concerned about over-speeding the flaps. We’ll have to wait for the CVR to be found to understand the crew’s thought process (or lack thereof).
@Andrew: Ah yes, that’s a good idea, as the question was asked right after extending the flaps.
Speaking of the trim runaway memory items … I just read this:
“If I’m in an aircraft and I’m getting stall warnings, stick shakers […] I’m definitely not going to run runaway trim memory items.” (source)
How to decide on the correct procedure in this situation, given that they were not sure, which information they could trust (false stall warning) ?
Memory items for unreliable airspeed, then for runaway trim ? Or vice versa ?
Or something else ?
The crew had 5200+4300 hours on type, so quite experienced.They would have known how to react to an identified problem, but I guess they were unable to find out in which situation they were in.
It would be interesting to sim-verify how many different alarms are triggered in this situation. I can imagine that it’s quite overwhelming and when you don’t know what alarms/information/instruments to trust, I think there is a possibility to lose situational awareness (as happened on AF447, where they also didn’t believe certain indications):
« the sensor may feed false information about this angle to the flight computer, which in turn triggers other errors. With the sensor falsely indicating that the nose is too high, when it isn’t, it causes [hstab nose-down trim]. At the same time, it causes an indicator of the minimum speed to tell the pilot that the plane is near a stall, which also causes the pilot’s control column to shake as a warning. And the airspeed indicators on both sides of the flight deck disagree. The pilots can use extra force to correct the nose down trim, but the failure condition repeats itself, so that the nose-down push begins again 10 seconds after correcting. “If the nose is trimmed down on an aircraft, it becomes difficult for the crew to hold it,” said the person briefed on Boeing’s bulletin. “The nose is turning itself down and they are having to fight it. It takes a lot of effort to keep it from diving. Especially if you have a crew that’s confused and doesn’t know what’s going on.” […] Pilots are typically trained on how to handle a runaway trim situation, said the person briefed on the Boeing bulletin, but that’s with everything else working as it should. In this case, the control-column shaking, the stall warning and the air-speed-indicator disagreement all combine to create confusion. » (source)
I agree with your previous comments that the crew should have known to fly pitch&power and handle the runaway trim by using the stab trim cutout switches, but I think the system shouldn’t put the crew in a situation where it has to decide which alarms to ignore.
@Viking
Perhaps the best visualization of commercial air traffic is recently shown by Jean Luc at
http://mh370-captio.net/wp-content/uploads/Traffic-8-March-2014-uploadee.mp4
There could be military and other air traffic not shown.
@all: I’m certainly missing something, but why were there several automatic nose down trim commands during JT043’s decent (orange line on p.16/25) although the crew had moved the stab trim switches to CUTOUT ?
« The PIC moved the STAB TRIM switches to CUT OUT and the SIC continued the
flight with manual trim without auto-pilot until the end of the flight. » (p. vii/9)
@Peter Norton
Regarding Something else that I find very intriguing:
The SIC had the presence of mind to “ask the TE controller the speed as indicated on the radar display.” (p.2). This was 8 sec after extending the flaps (for the second time), so panic could have still developed later on. But (at least until this point) it seems that the crew (or at least the SIC) didn’t suffer from tunnel vision but – quite to the contrary – was looking for solution slightly out of the box.
If I can dive in on that, I think that it’s actually the other way around; the request to ATC for speed was evidence that they were working to a pretty well prepared pre-briefed plan for Airspeed Unreliable and ALT DISAGREE on take-off. The FCTM actually mentions ATC radar as a potential cross check for managing Airspeed Unreliable;
‘Ground speed information is available from the FMC and on the instrument displays. These indications can be used as a crosscheck. Many air traffic control radars can also measure ground speed.
It also gets a run in an article called Erroneous Flight Instruments in Boeing Aero, magazine 8. Under ‘Getting assistance from others.’ the article states;
‘After maintaining control with pitch and power and having a plan for daylight visual conditions, the crew should seek help from air traffic control (ATC). Most ATC facilities have groundspeed readouts and general knowledge of the winds.
I think there’s pretty good evidence that the accident crew did what you’d expect a professional crew to do with the limited information that they had prior to the flight; they briefed for Airspeed Unreliable and ALT DISAGREE on take-off. And that is exactly what they were initially presented with. From there the ‘priming’ effect and confirmation bias would have shaped the crew’s perceptions of and decisions about everything that happened subsequently. Rather than thinking out of the box, I think that they were constrained to working inside a well prepared box, regrettably it was the wrong one.
@Victor. “Luckily, the vortex generator part was identified by independent investigators outside of the official investigation”.
True, though there are some features about the recovered item which I find puzzling. Naturally enough the Malaysians and the ATSB did MH370 origin assessments, the ATSB adding some geology and biology to establish consistency with nominal recovery site.
As to damage assessments the French did best with the flaperon though its attachments were too badly corroded for microscopic examination. Then there was a conclusion drawn by the ATSB from some flaperon and flap damage. That was about it.
As to other recovered items, take item 10, the left flap part: what was the direction of failure? Item 11, the IFE monitor frame. What forces and source of those could have caused that distortion, e.g. had the seat been ripped out? Item 26, the right aileron part, its hinges torn off. What would cause fracture like that?
Please forgive my harping.
@Mick Gilbert:
re: your takeaways from the report
Good points, I agree with all of them.
I just saw you had the same reaction to the ATC communication.
Mick Gilbert: “One of the big issues for JT43 when they experienced their MCAS event was control column forces. The FO ‘commented that the control column was too heavy to hold back‘. At the onset of JT43’s MCAS event the control column forces rise rapidly to around 85 (out of 100) of whatever the units of measure are. For JT610, until the terminal 90 seconds or so, the forces never get above 50 units and tend to bounce around the 25 mark.”
Sorry, which FDR line is that ?
Mick Gilbert: “When the Captain of JT610 accepted that aircraft he was aware of […] the fact that the [STS] was ‘… running to the wrong direction’.”
How would that manifest itself (symptoms, consequences) ?
@Mick Gilbert:
Thank you for diving in. I didn’t assume ATC radar cross-check would be SOP simply because it is not always available (i.e. mid-ocean). But yes, your assessment reads very plausibly and I completely agree with you on the priming effect and shaped perceptions. That’s a very likely scenario.
> After maintaining control with pitch and power and having a plan for daylight
> visual conditions, the crew should seek help from air traffic control (ATC).
> Most ATC facilities have groundspeed readouts and general knowledge of the winds.
Why would wind be important ?
I thought wind is irrelevant for flying pitch&power ?
@Mick Gilbert:
re: which FDR line is that?
Found it:
https://4.bp.blogspot.com/-lAZwDDq_TwY/W_fsqogDKMI/AAAAAAAAFFU/PqHYsDuS5cQdGf3EZAfltPV9Qhfh3nVuACLcBGAs/s1600/stabilizer+-+1.png
It’s no in the preliminary report. Why?
Where can I find an as complete as possible set of the FDR diagrams ?
ok, found it: https://ngamotu.nz/images/20181122-jt610-knkt.pdf
Or do we have more data than this ?
Is it not curious some of this FDR data was left out from the report ?
@Peter Norton
Regarding the FDR, specifically the inclusion of the CFFORCE_PITCHCWS FOREIGN and LOCAL parameters in the report to Parliament and their absence from the Preliminary Report, yes, I thought that was curious too.
I guess at the end of the day, the Preliminary is just that. From the media reporting of the report to Parliament it sounds as though they wanted to stress the increasing control forces towards the end-of-flight and that’s why CFFORCE_PITCHCWS was included. There are obviously a lot of FDR parameters that can be displayed in a report; like any graphical representation picking the right data and the right format is key to clarity of message.
I’m still trying to hunt down exactly what CFFORCE_PITCHCWS FOREIGN and LOCAL actually represent and what the unit of measurement is. What I’ve inferred is that 100 = Herculean, 0 = easy peasy and that FOREIGN and LOCAL probably refer to pilot applied forces and control column back pressure although that’s all conjecture, right down to which is which.
Regarding why the wind would be important, Andrew’s the go-to guy but my understanding is that if you’re trying to determine which IAS display is correct and you’ve got ground speed and your heading, below around 5,000 feet, all you need is wind speed and direction to determine airspeed.
Regarding the speed trim system ‘running the wrong direction’, again, not an expert but from what I’ve been able to find and read the STS is a stability augmentation system that provides ‘artificial’ speed stability by trimming against changes to the target speed. My understanding is that if there’s an input that slows the aircraft away from the target speed the STS will trim slightly nose-down to resist that, vice versa for an input that accelerates the aircraft away from the target speed. The STS trim inputs tend to be fairly ‘refined’.
My reading of the way the JT43 crew interpreted the MCAS induced trim commands is that their expectation was that as they accelerated through take-off and climb the STS would normally wind in small increments of nose up trim to counteract the acceleration. The nose down MCAS trim was interpreted as the STS running to the wrong direction (on steroids!).
Fun fact regarding STS (and its high speed cousin, the Mach Trim System) is that they are deemed to be Stability augmentation and automatic and power-operated systems for the purpose of certification. As such, they must comply with 14 CFR §25.672 which is as follows:
If the functioning of stability augmentation or other automatic or power-operated systems is necessary to show compliance with the flight characteristics requirements of this part, such systems must comply with §25.671 and the following:
(a) A warning which is clearly distinguishable to the pilot under expected flight conditions without requiring his attention must be provided for any failure in the stability augmentation system or in any other automatic or power-operated system which could result in an unsafe condition if the pilot were not aware of the failure. Warning systems must not activate the control systems.
(b) The design of the stability augmentation system or of any other automatic or power-operated system must permit initial counteraction of failures of the type specified in §25.671(c) without requiring exceptional pilot skill or strength, by either the deactivation of the system, or a failed portion thereof, or by overriding the failure by movement of the flight controls in the normal sense.
(c) It must be shown that after any single failure of the stability augmentation system or any other automatic or power-operated system—
(1) The airplane is safely controllable when the failure or malfunction occurs at any speed or altitude within the approved operating limitations that is critical for the type of failure being considered;
(2) The controllability and maneuverability requirements of this part are met within a practical operational flight envelope (for example, speed, altitude, normal acceleration, and airplane configurations) which is described in the Airplane Flight Manual; and
(3) The trim, stability, and stall characteristics are not impaired below a level needed to permit continued safe flight and landing.
[Amdt. 25-23, 35 FR 5675 Apr. 8, 1970]
In compliance with (a) failure of either STS and Mach Trim is annunciation via SPEED TRIM FAIL and MACH TRIM FAIL warnings.
The $64 question; is MCAS a stability augmentation and automatic and power-operated system?
@Peter Norton
RE: “How to decide on the correct procedure in this situation, given that they were not sure, which information they could trust (false stall warning) ?
Memory items for unreliable airspeed, then for runaway trim ? Or vice versa ?
Or something else ?”
The stall warning began as the aircraft rotated during take-off. Given that the aircraft got airborne and otherwise climbed away normally (initially), the crew should have recognised the stall warning as spurious. The stick shaker would certainly have been distracting, but it should have been ignored.
It’s not clear if the crew of the accident aircraft had the same IAS/ALT DISAGREE alerts as the crew of JT043. However, assuming they did and that they occurred at about the same point during the initial climb (400 ft), the crew should then have completed the memory items for unreliable airspeed, the first of which is ‘Adjust the airplane attitude and thrust. Maintain airplane control’. That would have entailed making sure the aircraft had the correct climb attitude with take-off thrust set, which should have already been the case (ie no change should have been required).
The next thing would have been the acceleration and flap retraction, which is when the situation really deteriorated. MCAS kicked in as soon as the flaps retracted, which obviously caused pitch control problems. In this case, the crew reselected the flaps within about 10 seconds of their retraction, which inhibited the MCAS. Had they left the aircraft in that configuration we probably wouldn’t be having this discussion. However, for whatever reason, they retracted the flaps again after levelling off at 5,000 ft and the MCAS again started trimming nose down. At that point, the crew should have recognised that the trim was causing pitch control problems and should have called for the Runaway Stabiliser checklist memory items.
It’s obviously more difficult to analyse and make those decisions on the spot, but I don’t think the problems were insurmountable. As I said previously, the priority is to FLY THE AIRCRAFT. If it’s not doing what you want it to do, then DO SOMETHING to make it happen!
RE: “I think the system shouldn’t put the crew in a situation where it has to decide which alarms to ignore.”
That’s life, unfortunately. The warning systems we have today aren’t sophisticated enough to decide what’s relevant and what’s not. That’s where pilots need to step in and quickly prioritise what actions they need to take, which is sometimes easier said than done.
Seabed Constructor arrived in Cape Town at 03:47 UTC this morning 29th November 2018.
@Viking
The Inmarsat paper by Ashton et al. that you quote states “The kink in the measured Pilot frequency characteristic was unexplained until it was noticed that it coincided with the satellite eclipse which took place between 19:19 and 20:26 UTC, and so was probably due to satellite oscillator errors due to thermal variations caused by the eclipse.”
Models built by a number of us here include the effect of the eclipse, which appears to be missing in your paper.
I did wonder whether the BFO errors you explain with the thunderstorm were actually the eclipse, but you see the thunderstorm effect at 21:41 UTC and the eclipse was in an earlier timeframe. The eclipse took place between 19:19:42 UTC and 20:26:02 UTC. The minimum oscillator external temperature was at 20:33 UTC. The maximum satellite frequency translation variation was at 20:38 UTC.
@Viking: I had a previous comment about your BFO model which you did not comment on. Until you use the proper BFO model that incorporates the Doppler pre-compensation algorithm of the AES, any conclusions you draw that uses the BFO should be viewed with suspicion. The Doppler shift and the BFO do NOT have the same sensitivities to the individual components of the velocity vector.
@Victor
A picture is worth 1000 words? In the link below D3 is the satellite to ground Doppler compensation. SK999 and I have had numerous discussions both public and private about how this value evolved over time. Our final conclusion is that the data presented by Inmarsat includes the effect of the eclipse.
D2 is the Doppler produced by the change in range from the aircraft to the satellite (includes both satellite and aircraft motion).
D1 is the AES Doppler compensation based only on the motion of the aircraft in the local tangent plane.
https://photos.app.goo.gl/MTqzLwDZwjNrmLyx5
Aside1: I can never understand why people use dropbox and other file sharing programs. Google sheets and docs are perfect for creating information as well as sharing information. Google photos is great for sharing graphics as well.
Aside2: I am really into a new book by George Gilder, “Life After Google”. Gilder is perhaps the undisputed oracle of the future of technology despite living in Massachusetts.
@Dennis
Nice picture from the archives!
@DennisW: I don’t know what information you wish to show me. I remember that graphic well, and I agree it is good.
@Viking assumes that the AES pre-compensation is constant because he could find no details about the algorithm. In fact, the BFO has been thoroughly understood (including the effect of the eclipse) since June 2014. It would be better to ignore the BFO completely than to use an incorrect model to discriminate between final locations on the 7th arc.
As for Dropbox, I use it in preference to Google Sheets and Docs because I log into Google as little as possible for privacy reasons.
@Victor
The graphic was intended for Viking.
@Viking
Please find below a link to my MH370 Flight Path Model V17.0 with your route.
I have included the wind speed and direction as well as air temperature en route.
In Column T you will see the “”Error Calibrated BFO (Hz)” increases steadily to 26.1 Hz as the flight progresses, despite including the “Aircraft Compensation Doppler (Hz)” in column DN and the “EAFC Effect” and “Eclipse Effect” in columns DQ and DR.
https://www.dropbox.com/s/5oghhdzpkj2bkng/MH370%20Flight%20Path%20Model%20V17.0%20MK.xlsx?dl=0
@Viking
Please find below a link to my drift analysis from your end point at 13.279°S 106.964°E.
The simulated debris path ends after 266 days near Mombasa in Southern Kenya.
The track passes Reunion after 208 days, around 1,256 km north of St. Andre, where the flaperon was found.
https://www.dropbox.com/s/7odxz0b64wiflas/Drift%20Map%2013.2790S%20106.9640E%204.8228S%2039.8585E%20265d.png?dl=0
@Victor
fyi Boeing Stock ended up 2.7% today to 342.5 which pretty much back to normal it would seem
@Richard Godfrey
I included the effect of the kink (in the BFO) the same way as in the Inmarsat paper, so in light of @DennisW comments above it is included in my calculation.
@Victor
Nice that you have run my data through the BFO compensation model so quickly. Unfortunately I had overlooked that there was a model available, so I developed a crude one myself (as presented in the Note added in proof). I am sure yours in much better.
@Richard Godfrey
Could you tell me if the 26.1 Hz error you find refers to my table 3 or to the numbers in the Note added in proof?
@DennisW
May I ask if you got the recipe presented in the figure from Inmarsat or worked it out yourself (or as part if the independent group)?
The reason I ask is that if you got it from Inmarsat it answers my question concerning the inertial system. Then they did indeed use the SAT ground projection (as I guessed, since it seemed to agree best with their results).
@Richard Godfrey
Concerning your drift analysis with my starting point:
Did you include the effect of hurricane Gillian? As far as I can see this hurricane would have pushed the debris significantly south (and moderately east), and dispersed them (critically dependent on their detailed position when it hit).
@Viking
I pieced the model together from basic physics and the discription of the AES compensation provided by Inmarsat. D3 and D2 in my model are simple physics. D1 is based on the Inmarsat description. D1 differs from D2 by using the nominal satellite position over the equator (no satellite motion), and the motion of the aircraft in a plain tangent to the earth.
@Richard Godfrey
A brief follow up on my other citations (from yesterday):
1) I only cite Chillit for his back-tracing of temporarily beached debris and his observation of the impact of hurricane Gillian. As far as I can see he got those right, so he deserves citation. I have absolutely no intension of citing him for anything else.
2) I cite Wise for a couple of things. Most of these are actually citations of other bloggers writing in his blog (I should have been more clear on that, sorry). I think some of these contributions were actually written by you and a couple of others from here. As far as I remember, I only cited him personally for one or two things, including his proposed northern route and something concerning the clamps. His proposed route fits poorly, and actually it is more than 500km off target, but I still think he is right pointing out that it deserves more thorough analysis than a clean rejection of everything going north (as eg. found in the Inmarsat paper).
@DennisW
Thanks for your additional information on the figure, – I agree completely. However, I think Inmarsat has hidden some additional details in there. One of them is what they did to the satellite height change. Another one is if they made some numerical approximations to speed up calculations. One more concerns correction for weather.
I do not believe they are able to handle the most important weather contribution from thunderstorms. I have some practical experience with that from my home-built satellite TV receiver. My girlfriend always complains that her preferred channels are the first to drop out when a thunderstorm approaches. When a strong thunderstorm comes really close all channels disappear. The damping is 20dB or more.
@Viking
The weather has very little effect on an L-band signal. That is why it was selected for GPS.
@Viking
The Inmarsat 3F1 ephemeris is well known, and of course included altitude. It was also parameterised by Henrik, in about Dec 2015. I’ll send you the data by email.
@DennisW
Indeed the L-band is much less affected, but powerful tropical thunderstorms have some broadband contributions, and I only need 2dB (not 20dB) to explain things.
@Viking
Link below to Python 2.7 script for 3F1 position and velocity.
https://docs.google.com/document/d/1sSabQESr0YMaeOggJMVzjR-Ux38nuFXLNfoI-P17pMA/edit?usp=sharing
@Viking
and I only need 2dB (not 20dB) to explain things.
Richard’s drift analytics for 13S are difficult to reconcile with a late July 2015 beaching of the flaperon at ReUnion. I think your preferred terminus is a very hard sell in this group.
Viking: As I noted above (you did not respond), and Dennis also noted, rain attenuation at 1600 MHz is <0.1 dB…even in hard rain. And the assumption the link margin improves as range increases is also wrong, as previously noted.
@Viking: If you are going to propose a route based on the BFO, then you have to show that your route satisfies the true BFO model. I am not able to find a great circle route that ends near 13S that does.
Also, the gain of the satellite’s “bent antenna” is not constant, but in fact reduces as the volume of traffic increases, as Mike has previously explained.
@Andrew: thank you for outlining the correct procedure
@Andrew / Mick Gilbert:
Why are there several automatic nose down trim commands during LIN043’s descent (orange line on p.16/25) although the crew had already moved the stab trim switches to CUTOUT ?
re: “The nose down MCAS trim was interpreted as the STS running to the wrong direction (on steroids!).”
I’m wondering if the pilot accepted this aircraft due to the overall pressure and if he deemed it safe, considering that they apparently had planned for and probably were even over-fixated on Airspeed Unreliable and ALT DISAGREE on take-off.
Would the LIN610 crew automatically know what the LIN043 crew “reported through the electronic reporting system of the company A-SHOR” (notably the “STS running to the wrong direction” which was not mentioned in the maintenance log) ?
re: not including the control forces graph from the FDR
I would have understood if they had left out something unrelated, but that feels weird given this is right at the core of the problem.
@Viking
Hey, I’ve been here many times over the course of this investigation. Know that we all appreciate sincere effort.
https://media.giphy.com/media/O8cQ3BUcmhX56/giphy.gif
Victor Iannello: “@Andrew: Thanks for that link to the article. That’s the best one I’ve seen written so far.”
This article also highlights the lack of redundancy I mentioned, which you took issue with, Victor:
« a potential design flaw in Boeing’s MCAS system: The sequence was triggered by a single faulty AOA sensor. A so-called “single point of failure” that could bring down an airplane is absolutely anathema in aviation safety protocols. Lemme, who designed flight controls at Boeing, said that […] the AOA malfunction is a single point of failure of the equipment — something airplanes are rigorously designed to avoid »
Now Boeing considers a software update:
Reuters: Boeing eyes Lion Air crash software upgrade in 6-8 weeks
« While each 737 has 2 of these blade-shaped vanes, the plane’s anti-stall system relies on data pulled from just a single vane during each flight, compared with a 3-sensor “voting” system on rival Airbus (AIR.PA) jetliners.
Boeing’s software update would come as an emergency measure from Boeing and FAA, the sources said.
The specific system for preventing stalls was not originally designed to monitor both probes because regulators assumed risks of a mishap were small and would be further reduced by the presence of a trained crew and power switches on errant systems.
Now, however, Boeing is examining whether the anti-stall system should also check data from the second probe before engaging, according to a person briefed on the matter. »
This is in line with the EAD stating:
« This emergency AD was prompted by analysis performed by the manufacturer showing that if an erroneously high single AOA sensor input is received by the flight control system, there is a potential for repeated nose-down trim commands of the horizontal stabilizer. This condition, if not addressed, could cause [a crash]. »
@Peter Norton
RE: “Why are there several automatic nose down trim commands during LIN043’s descent (orange line on p.16/25) although the crew had already moved the stab trim switches to CUTOUT ?”
I suspect they’re spikes in the data and are spurious. There does not appear to be any associated change in the pitch trim position.
RE: “Would the LIN610 crew automatically know what the LIN043 crew “reported through the electronic reporting system of the company A-SHOR” (notably the “STS running to the wrong direction” which was not mentioned in the maintenance log) ?”
I assume ‘A-SHOR’ is some kind of post-flight reporting system or a safety occurrence reporting system. I don’t know about Lion Air, but in my experience such reports take time to filter through airline management and they are not normally available to line crew unless management chooses to publish them.
@Peter
Boeing’s responses are crafted by their lawyers. The subject aircraft should have been taken out of service until a completed investigation. To me it shows a weak kneed FAA and NTSB response, and a good reason not to get on board an aircraft (any aircraft).
@Peter Norton: Nobody disagrees that the failure of a single sensor could cause a crash. It did. The sensors were redundant but not fault tolerant. Active intervention on the part of the pilots was required to detect and isolate the fault.
[Comments here are closed. Please continue the discussion under the new article.]