China Eastern 737-800 MU5735 accident March 2022

[etrang]

Originally Posted by finestkind

Tend to agree especially if someone states this is the reason. Maybe I missed that but most appear to have indicated along the lines of this "may have happened".
.
In respect to offering an opinion, "Causative Factors:
Absolutely no idea. How could I offer an opinion before the data is in?.

An opinion is "a view or judgement formed about something not necessarily based on fact or knowledge". So by all means offer your opinion and if outside the bounds of reasonability standby to receive.

No, there's no reason to preclude it, but no reason to include it either. There are countless possible causes, so its not helpful to raise them all without some sort of supporting evidence. This is probably why your earlier post was deleted.

[PJ2]

Originally Posted by henra

While I do agree from a technical perspective, historically in past accidents the ADS-B data rarely if ever painted a really different picture regarding trajectory/flight path and velocities than were leater determined by the proper tools.

Thank you for your response.

FR24 began in 2009, so there was nothing available from AF447 at the time, but we did use the ACARS messages to read the tea leaves in the AF447 accident and got quite a bit from that source until the recorders were located two years later. But until we had the side-stick parameters, the stall-warning parameter plus pitch/roll, it was not possible to say where the problem originated and why; from the AF447 thread in Tech Log on ACARS: AF447

Data is not opinion, nor is validated data stochastic. That's really my only argument. There can be a form of equivalence between FR24 information and the eventual results from the recorders but the former cannot be turned into action nor legitmately be included as part of any final report / recommendations.

[DaveReidUK]

Originally Posted by henra

While I do agree from a technical perspective, historically in past accidents the ADS-B data rarely if ever painted a really different picture regarding trajectory/flight path and velocities than were later determined by the proper tools.

That has certainly been true - up to a point - for the dozens of accidents/incidents that have featured in these columns over the years and where FDR traces have subsequently been published in the investigation report (including the two Max accidents).

But there is an important caveat: ADS-B data (as transmitted by the aircraft) and an FR24 download (however granular) are not the same thing. The latter will be subject to asynchronicity, hysteresis and latency - though many (but not all) of those issues can be mitigated with some work and patience.

So, while I have no sympathy for the knee-jerk "ADS-B is rubbish" view sometimes expressed in these columns, nor do I agree with the "ADS-B is gospel" view when applied to FR24 data.

[henra]

Originally Posted by PJ2

Data is not opinion, nor is validated data stochastic. That's really my only argument. There can be a form of equivalence between FR24 information and the eventual results from the recorders but the former cannot be turned into action nor legitmately be included as part of any final report / recommendations.

Fair and valid comment! I do agree.

[Ka6crpe]

I know this is pure speculation but one report says "One possibility is that this 737 may have previously been involved in a tail strike incident on landing or take-off and the damage done to the rear pressure bulkhead was not noticed or not repaired properly causing it to fail, damaging the tail."

Has there been any evidence to support this or are they just making noise?

https://www.airlineratings.com/news/...und-10km-away/

[DingerX]

Originally Posted by PJ2

The reliance upon FR24 data as an investigative, diagnostic or even theorizing tool is significantly beyond its design, conception and utility.

While "realistic" results may obtain on occasion, its equivalence to reality is largely stochastic.

The process of building a theory and then going to data to find support is essentially what is occurring through the use of FR24 data. The method is faulty simply because the data isn't good enough to do so. FR24 data is not validated through testing to the same precise standards & legal requirements that DFDR data is. It has its uses, but not in the manner being sought here.

The only tools registered and recognized for such work are the SSCVR & SSFDR, and even then, as we have seen, from such a point there can still be a number of valid theories & interpretations, some ultimately critical to understanding what actually occurred.

If scientists had gone about validating MRNA vaccines in the same manner, we would not be as eager to use them and the FDA would have trouble justifying their wide public use. The principle here and elsewhere in other investigative endeavours is the same.

Respect for the data and for accuracy is paramount as many lives have been affected.

PJ2

Of course, that invites its own study. Take the last 10-20 incidents discussed on PPRUNE for which a final report is available. How do the publicly available ADS-B data match the DFDR readout and the reconstruction? What percentage of posts mention technical details or theories considered by the final report, compared to issues not mentioned (and thus probably irrelevant noise)?
How much should we respect the data?
​data?
Since Traditional and Internet news outlets routinely raid forums like this one, maybe we should scientifically assess the information that comes out here. I won't even ask to be principal author.

[B2N2]

Second black box found

https://www.channelnewsasia.com/asia...-found-2589126

[WideScreen]

Originally Posted by fdr

....
Planes fly nicely unless they are either -

1. commanded to do something else; or
2. upset by external factors; or
3. the fundamental stability of the toy gets upset.

......

I like your writing style, full of subtle references to past happenings, as discussed on PP.

Your list of conditions does suggest something "permanent", IE mechanics, though the issue with fundamental toy stability can also be something intermittent. Where mechanical items either work or break, electrical/electronic/software items can easily show intermittent behavior. IE, upset, working again, upset, working again, etc. For MU5735, the FR24 data does show a severe upset, a temporary recovery and again a severe upset, suggesting an intermittent issue with the aircraft.

Of course, FR24 data is not perfect, though it was good enough to determine on PP, what potentially happened with SJ182.

[Sailvi767]

I posted a paragraph from a Chinese site discussing the crew. Everything was deleted for reasons I don’t know.






It was deleted because you started your post with:

I don’t place much validity to the following info but it was translated from a Chinese site. I suspect it can be debunked quite quickly by the China hands on here

Not only a negative post, but already discussed in earlier contributions.

Senior Pilot

[Magplug]

With the report that wreckage has been found 8km away reinforces the observation that the airframe breakup was at medium to high level. The multiple negative results of explosives residue negates any suggestion of a bomb.

On the one hand the Chinese are being uncharacteristically open with their press conferences but may still be guarded in their admissions of the possibility of a mid-air collision with an errant mil-jet. Everything outside of airways in China is mil-controlled. Consequently getting co-ordination for wx avoidance is downright impossible. Equally an errant military jet may have infringed civil airspace resulting in a mid-air, survivable for one party but not the other. Without some force-majeure intervening towards the end of a perfectly stable cruise, the likelihood of the tail simply falling off seems very unlikely indeed.

[AAKEE]

Originally Posted by DaveReidUK

That has certainly been true - up to a point - for the dozens of accidents/incidents that have featured in these columns over the years and where FDR traces have subsequently been published in the investigation report (including the two Max accidents).

But there is an important caveat: ADS-B data (as transmitted by the aircraft) and an FR24 download (however granular) are not the same thing. The latter will be subject to asynchronicity, hysteresis and latency - though many (but not all) of those issues can be mitigated with some work and patience.

So, while I have no sympathy for the knee-jerk "ADS-B is rubbish" view sometimes expressed in these columns, nor do I agree with the "ADS-B is gospel" view when applied to FR24 data.

Using the granular data from this accident in a software that allows for a 3D view of the data points in position and altitude we can se things that we can not when just looking at the table.

This aircraft started the departure from stable flight with a roll to the left, and this roll got more and more like a barrel roll from about inverted. Probably due to a about 1G postitive load during the roll.
The aircraft seem to have performed at least one complete roll, completed about when the aircraft briefly gained altitude again. There is a possibility that the aircraft continued to roll after the first barrel roll.

Sriwijaya 182 had a A/T failure that put #1 Engine to idle. It made a similar left barrel roll.

Maybe something similar A/T or engine failure or a rudder hardover( I know they aren't supposed to happen these days but…)

[Sailvi767]

I have had a catastrophic engine failure in cruise flight in a 767-300 ER. Instant loss of all thrust. In cruise flight it’s actually a very benign maneuver. You would have to be dead asleep to let it rollover. It’s also unlikely a rudder hard over would cause a upset as rudder travel is limited by dynamic pressure at cruise speeds. Rudder hardcovers normally are a problem at low speeds where full throw is available.

[EDML]

Originally Posted by Sailvi767

It’s also unlikely a rudder hard over would cause a upset as rudder travel is limited by dynamic pressure at cruise speeds. Rudder hardcovers normally are a problem at low speeds where full throw is available.

Correct. The accidents after the rudder hard over with the 737 happened at rather low speed during the approach. At higher speeds there is enough aileron authority to overcome the rolling moment caused by the rudder. One of the "fixes" was to apply higher speeds for flaps 1-10 to avoid the speed envelope where a hard over could not be counteracted by the ailerons (called crossover speed).

There were also some incidents with the rudder PCU that happened in cruise flight. They were recoverable.

[Magplug]

Originally Posted by Sailvi767

It’s also unlikely a rudder hard over would cause a upset as rudder travel is limited by dynamic pressure at cruise speeds.

That's very true... Under normal conditions and presuming the RTL performs as expected. Look at the the 747 lower rudder valves malfunctions that have occurred and you will see that is not always the case. The rudder may travel beyond the RTL limits resulting in LOC.

[Feathered]

Originally Posted by Sailvi767

The trim system in the 737 is very obvious if not annoying when running. That includes when the autopilot trims. If in fact they ignored the spinning trim wheels and noise as soon as the autopilot moved the elevator to counter the incorrect trim input the trim brake would engage and stop the runaway.

The 737's trim does make a very noticeable clatter when running. But if there is also cacophony of alerts, plus a stick shaker simultaneously making noises and distracting the pilots from listening to trim wheels, I'm not sure we can count on trim system noise to provide sufficient a human factors warning when things are off nominal. The trim brake is another issue.... So many unknowns at this time. Anxiously awaiting for actionable FDR data....

[Mr Optimistic]

[pax] I think any proposed cause has to account for the timing of the event ie at or about start of descent. A random occurrence, eg collision or mechanical failure could otherwise have occurred with equal probability at any earlier time in the cruise. More likely to be associated with a circumstance specific to that phase of flight surely.

[Stick Flying]

Originally Posted by Mr Optimistic

[pax] I think any proposed cause has to account for the timing of the event ie at or about start of descent. A random occurrence, eg collision or mechanical failure could otherwise have occurred with equal probability at any earlier time in the cruise. More likely to be associated with a circumstance specific to that phase of flight surely.

I don't agree. That is making an assumption it was changing from cruise to descent at that point and something in that process went catastrophically. It could be a multitude of reasons this flight went horribly wrong. Probabilities are fortunately something the investigation team will ignore unless it backs up FACT. If the recorders are readable, they will give a far better insight than any assumptions can.

[N600JJ]

Could this be similar to Southwest B737-700 flight 1380 in 2018 whose incident report can be found at https://data.ntsb.gov/carol-repgen/a...ort/93897/pdf? https://en.wikipedia.org/wiki/Southw...es_Flight_1380

Uncontained engine failure with a sudden left roll of a 41° bank angle leading to decompression of the aircraft, emergency descent and damage to, amongst other parts of the aircraft, wing tip and horizontal stabilizer?

[SteinarN]

Originally Posted by N600JJ

Could this be similar to Southwest B737-700 flight 1380 in 2016 whose incident report can be found at https://data.ntsb.gov/carol-repgen/a...ort/93897/pdf?

Uncontained engine failure with a sudden left roll of a 41° bank angle leading to decompression of the aircraft, emergency descent and damage to, amongst other parts of the aircraft, wing tip and horizontal stabilizer?

I made an argument the left roll (there can be no doubt that a left roll was the start of the flight path deviation) could most likely be caused by either an uncommanded rudder movement or an engine malfunction causing the left engine to lose thrust and taking the crew by surprice. ie crew not taking action before AP disconnect in an out of trim state. Alternatively an uncontained engine failure also damaging rudder cables/controls and/or wing surfaces.
However that post have since been deleted by the mods.

I was surpriced by the number of cracked fan blades found acording the the report you are linking. That another fan blade could depart an 737NG engine can hardly come as a surprise, however I do not say with any certainty that an engine malfunction did actually happen in this accident, only that such a failure is one of the more likely reasons for this accident.

[A0283]

Originally Posted by Dave Gittins

Looking at posts 250 and 255 and wholly speculatively, how could a reduction in thrust at top of descent instead have activated a thrust reverser ? Did somebody say they have only found one engine ? Post 214 and nobody has rushed it to say it's wrong.

Dave it means that on the photo and video material that I have seen, there was only one engine visible (suprisingly - with the fan side up) next to a pool of water that they called the 'first pond', the circumference appeared quite intact, but the engine somewhat flattened. I did not observe the removal of that engine from the mud. And did not see the other engine. Indeed nobody commented on this then or later. So we dont know if it is the left or right hand engine that was found.

I did observe for example the removal of the outerwing/winglet from the deep mud, and the removal of a top of one big MLG fragment including trunnions. And observed the handling of the FDR 'can' just after it was found. The CVR can was a Honeywell.

The investigation reports things like 24,000 earlier and yesterday 33,777 pieces found in their press conferences, but does not make statements on the four corners or main components that have been found, apart from the CVR and FDR. They appear to have established a pretty thorough process chain for filtering, handling and identifying parts, so that information should be available by now. That they dont make that public may or may not be surprising.