Lion Air crash: Jakarta Boeing 737 'had prior instrument error'

[Nusa]

Ever since you said you were deliberately presenting media-filtered layman explanations as fact, while also admitting you knew they were technically wrong, I've lost all urge to correct you. You're basically trolling us at this point.

[KL27x]

^I don't know if that was directed at me, or not. But that's the point. I'm not an authority, and I don't want anyone should take me seriously. It's just a discussion.



But... I came here because I had a revelation.

I'm an idiot.

4b. You say that you can't negate a positive feedback loop with a static change, but this isn't correct. If the position of the engines is what causes the problem in the first place, that's a static change to the aerodynamics of the aircraft. If a static change can cause a feedback loop, it can stop it.

Of course you are right.

Where I went wrong:
I stumbled over the problem that the stabilizer can't possibly have the same (but opposite) effect that the engines aerodynamics have by mere coincidence. But it doesn't matter. As long as the stabilizer is set to the point where its own counter effect is at least as much, if not a bit greater, than the lifting effect of the engines @ something slightly under the maximum effective AOA for the plane (for safe margin), then the plane would at least be stable near maximum AOA.

The MCAS is a dynamically acting solution. It's just that it has a 1 bit resolution response. We have "hi AOA" and "lo AOA". And then analog takes over from there.

Depending on the shape of the function graphs of the two opposing forces, this might result in the plane acting differently at this transition point. From lo to hi, there would be a change, like the step in a "logarithmic pot" [which is typically just 3 linear sections, connected, to approximate the actual curve]. There would be the distinct and steep downward net, here, over 10 or so seconds with a binary outcome.

1.But after powering through it with the elevators, it could drop and get closer to balanced. Even if the sum force temporarily trended to net positive at some point, it would be ok, as long as it is net negative, again, by the time the max AOA is reached.

2.converserly, if the pilot didn't fight through it, and just let the AOA drop, the MCAS would "deactivate" and put the trim back up for 10 seconds. (Not in the news, but now I simply assume Boeing is doing it like this.... hm, which means software bug is definite possibility to explain refiring. So this is good... and bad.... I mean the thing only has to remember two states in this hypothetical model).

But yeah, I'm an idiot. This is how buggy software gets written and later fixed after non-idiots test it.

[Brumby]

My point isn't that this isn't true, but that it's not important.

This is a point that the Media abuse incessantly.

The fact that they can write about a particular subject is all that matters - its relevance to the overall situation is ignored.  The reason is quite obvious - sensationalism sells.  It's all about the money.

In fact, how the Media operate has a lot in common with some of the scams we have seen... A kernel of truth that is embellished and perverted, all for the sake of a buck.

[Obi_Kwiet]

I'd argue that it's more a systems level bug than a software bug. The unlimited reactivation of MCAS was to spec, AFIK, it's just that that's dangerous and pointless.

I'm really annoyed at what a sloppy job of reporting the media does. You really have to dig to understand what is going on, and the media are just straight up wrong so much of the time. Unfortunately, that means that the public's takeaway is inaccurate and misdirected. I think we may have some interesting takeaways from all this on a system design and review process perspective, but everyone is shouting for some big shot to take the fall. Honestly, I doubt if anyone high up in the company knew about MCAS in any real detail.

[MT]

As previous mentioned from Nasa database( pages ago) pilots had little or no training on MCAS now its reported flight crew manual mention MCAS only once in glossary, Boing refuses to reply! Oh the drama, death, pain, money, fat lady not even dress'ed up yet while Boing apologists and top management on front row starts to sweat !




Senator Blumenthal refers to NASA pilot database and request personal parachute if to fly with 737MAX.

[MT]

Pilot talks about the 737Max problems in its DNA and how pilots dont understand MCAS and the planes behavior.

[GeorgeOfTheJungle]

-Yes, there's a flaw in the software.
-Yes, the AoA sensor may have failed.
-Yes, Boeing had not provided yet any fix for these known defects, five months after the Lion Air 610 crash.

But,

That said, this pdf put out by Boeing on 6th November 2018 explains very clearly the problem and a fix/workaround:
TBC-19-Uncommanded-Nose-Down-Stab-Trim-Due-to-AOA
http://www.avioesemusicas.com/wp-content/uploads/2018/10/TBC-19-Uncommanded-Nose-Down-Stab-Trim-Due-to-AOA.pdf

In the event of erroneous AOA data, the pitch trim system can trim the stabilizer nose down in increments lasting up to 10 seconds. The nose down stabilizer trim movement can be stopped and reversed with the use of the electric stabilizer trim switches but may restart 5 seconds after the electric stabilizer trim switches are released. Repetitive cycles of uncommanded nose down stabilizer continue to occur unless the stabilizer trim system is deactivated through use of both STAB TRIM CUTOUT switches in accordance with the existing procedures in the Runaway Stabilizer NNC.

Why did the ET flight 302 pilots not do that?
Is it possible that they had not seen/read/known about that pdf?

[MT]

-Yes, there's a flaw in the software.
-Yes, the AoA sensor may have failed.
-Yes, Boeing had not provided yet any fix for these known defects, five months after the Lion Air 610 crash.

But, That said, this pdf put out by Boeing on 6th November 2018 explains very clearly the problem and a fix/workaround: Why did the ET flight 302 pilots not do that? Is it possible that they had not seen/read/known about that pdf?

You forgot to mention the certification process sidestepping and now the added safety features "ala carte" debacle!

Precisely this bulletin note did "not" go into the flight crew manual as explained/excused in one of above videos, combine that with no training,and there is a result.

1: WHERE is this bulletin supposed to be if not in the flight crew manual?
2: Are a pilot supposed to memorize everything and if do but later forgets or not not able to, the license withdrawn?
3: How many armchair pilots here remembers every fix and workarounds for STM32 devices published in errata sheets
    and what about all those undocumented bugs STM32 have who popp's up when you least expect it?

This bulletin says only during manual flight, Juan above says only during autopilot engagement, cant even be turned off
during take off. This mess clearly tells why Ethiopians went to the French. All about the faff FFA and Boing is doing and not doing.

This whole story have streaks of not pilot failure, nor engineering failure(-softbugs) rather middle and top management failure.

[tooki]

The thing is, the checklist for runaway stabilizer trim is something that all 737 pilots are trained for, and is required to be known by memory. So while MCAS adds another point of failure, going through the standard 737 runaway stabilizer trim checklist would still disable MCAS and return stabilizer trim authority to the pilots. Even without knowing about MCAS, the pilots should still have been able to recover, if they acted quickly.

Note that the first item of this checklist is to disable autopilot. But this would already be the case, since MCAS only applies to manual flight. When autopilot is on, the autopilot, and not MCAS, has authority over the stabilizer trim.

[Nusa]

This bulletin says only during manual flight, Juan above says only during autopilot engagement, cant even be turned off during take off.

Manual flight is any time the autopilot is disengaged while flying.

When you're already on the ground, and something that should be working is not, then one should NOT take off and have the aircraft serviced instead. The automatic takeoff procedure merely enforces that policy, and Juan was reflecting that reality. I'm sure that he COULD get the aircraft off the ground manually if the alternative was worse (tidal wave incoming or something), but otherwise the actual policy is the same...don't fly with a known problem. (The exception would be pre-authorized ferry flights for maintenance with no passengers.)

Imagine how loud YOU would be screaming if someone knowingly took off without EVERYTHING working and then crashed?  You do realize that's exactly what you just demanded, right?

[tooki]

Aircraft routinely begin flights with things out of order. Some faults automatically ground an aircraft, others can be deferred.

[Nusa]

Aircraft routinely begin flights with things out of order. Some faults automatically ground an aircraft, others can be deferred.

I'm aware of that, but it was sorta besides the point, since the automatic trim system has never been one of those exceptions to my knowledge. Nor would that fact have anything to do with how loud MT would rant when a crash is involved.

Even the authorized exceptions have to be documented and signed off before the aircraft leaves the ground if it has anything to do with the flight or power systems. Nonworking passenger wifi or a broken coffee pot won't matter much, however.

[KL27x]

Note that the first item of this checklist is to disable autopilot. But this would already be the case

You would still go through the checklist from the top. If you saw Mentour Pilots video demonstration during simulation, it's a bit comical how formal it is. So in their 40 seconds, they would potentially be verbalizing each item, starting from the beginning.

So
1. "Memory items for runaway stabilizer"
2. "Memory item 1:Turn off the autopilot." Memory item 1 is not "If the autopilot is already off, skip to item 2"
3. The pilots would turn off autopilot
4. Then they would monitor for the result. And since MCAS is intermittent, "the result" may initially appear to be perfectly satisfactory.

You are supposed to monitor the result. You are NOT supposed to proceed to memory item 2, blindly. And if the result is satisfactory, you are supposed to stop.

That can waste a lot of time out of the 40 seconds the pilots had.

In the first incident with Indonesian Air/Jakarta, the pilots did the dance for 10 minutes, or something. I have not seen any detailed info for the Ethiopian Air crash. Just that the news said they had 40 seconds. And Ethiopian Air stated all their MAX pilots received MAX specific updated training, including the hiterto optional simulation in the interim between the Jakarta crash and this one.

Now, how pilots actually do this may be totally different in the plane. But even with the updated training, the solution may be insufficient? I'll bet if the pilot skips the first step, it's an automatic fail, in training. If you were to go "by the book," the book might kill you? I think that could be considered an insufficient solution. I imagine that strictly following procedure/protocol is supposed to be sufficient.

The protocol is there, presumably to prevent pilots from jumping to incorrect conclusions. There could be good safety reasons, statistically, that we want pilots that blindly follow protocol in this kind of situation. There could be financial pressures, as well. If the pilot cuts stab trim out of order, then he may not be allowed to later say "oops, my bad" and turn it back on. And he may have to divert the plane? Just speculating out loud. In this case, the pilot had already requested to return to the airport of departure, so he was already past that.

[Mechatrommer]

well you can talk all technical all protocol you like, they can do any cover up they like. for me the simple fact is, at this age, boeing is stupid enough not to put AI, ie just to let a commercial airplane to dive at military fighter jet grade of dive down, or overwhelming pilots with pages of manual and plethora of gauges and controls of the past centuries while the plane is tumbling down. not sure if its mentioned in this 29 pages thread, another 737 mishap on the best flight brand... https://edition.cnn.com/2019/03/10/africa/ethiopia-airline-crash-nairobi-intl/index.html imho all 737 shoud be called back, it hasnt happened before with older models, just think sensibly no need super complicated discussions just to find an answer/solution. pitot tubes failed? give me $5 gyro an accelerometer and an arduino and i'll fix that for you. call me ignorant or a fool if you like.

[langwadt]

well you can talk all technical all protocol you like, they can do any cover up they like. for me the simple fact is, at this age, boeing is stupid enough not to put AI, ie just to let a commercial airplane to dive at military fighter jet grade of dive down, or overwhelming pilots with pages of manual and plethora of gauges and controls of the past centuries while the plane is tumbling down. not sure if its mentioned in this 29 pages thread, another 737 mishap on the best flight brand... https://edition.cnn.com/2019/03/10/africa/ethiopia-airline-crash-nairobi-intl/index.html imho all 737 shoud be called back, it hasnt happened before with older models, just think sensibly no need super complicated discussions just to find an answer/solution. pitot tubes failed? give me $5 gyro an accelerometer and an arduino and i'll fix that for you. call me ignorant or a fool if you like.

hm.., how are you going to measure air speed with a gyro/accelerometer?

[james_s]

Any sort of automation falls apart as soon as you have one or more bad sensors feeding it the data. This is why you have human pilots in the cockpit ready to take over manually at any point. Jumping to conclusions and saying "oh this is easy, just have the software automatically do X, Y and Z!" is how you end up having discussions like this thread.

Personally I think pilots should be encouraged to hand fly much more often than they do. Too much reliance on automation takes people out of the loop and encourages the mind to wander.

[KL27x]

Personally I think pilots should be encouraged to hand fly much more often than they do. Too much reliance on automation takes people out of the loop and encourages the mind to wander.

I agree, but in today's world...

Management: You want employees to actually have fun on the job? We pay them to be miserable.

Seriously, though, as Boeing CEO has stated, and which I'm sure is true, Boeing, and probably the industry as a whole, is driven by data. Sure, these crashes might have been avoided if X, Y, or Z. But overall, things are done the way they are because it's the statistically safest way we currently know about... And venturing into the unknown on a hunch can produce much worse results in a hurry.

There are probably a lot of things about the planes we get on that we have no clue about, nor do we care. It's interesting how suddenly we all want to know about MCAS. I wonder if there are trade secrets (or even matters of national security) why Boeing can't spell out MCAS, completely. Of course, now, there are also legal reasons.

[Mechatrommer]

hm.., how are you going to measure air speed with a gyro/accelerometer?

given on condition of pitot tube failure (air speed measurement) and autopilot activated.
1) make sure airplane level (gyro and magnetometer)
2) altitude decreasing? or stall detection from accelerometer? increase engine speed and flap, aileron etc to appropriate position.
3) warning led to let pilot approve turning back or emergency landing to nearest site.

measure something with redundancy backup unit is better than measure nothing. otoh on manual mode, do not let pilot do stupid thing like miitary jet fighter acrobatic, (i saw youtube investigation that the pilot let his kid playing with the wheel ) AI will take over if pilot proved incapable or imminent crash is detected. any other questions?

[Brumby]

hm.., how are you going to measure air speed with a gyro/accelerometer?

given on condition of pitot tube failure (air speed measurement) and autopilot activated.
1) make sure airplane level (gyro and magnetometer)
2) altitude decreasing? increase engine speed and flap, aileron etc to appropriate position.
3) warning led to let pilot approve turning back or emergency landing to nearest site.
on manual mode, do not let pilot do stupid thing like miitary jet fighter acrobatic, (i saw youtube investigation that the pilot let his kid playing with the wheel ) AI will take over if pilot proved incapable. any other questions?

Remind me to never be anywhere an aircraft you design is in the air.  Your simplistic view is deadly.

Let's just take ONE aspect of this statement:

2) altitude decreasing? increase engine speed and flap, aileron etc to appropriate position.

With no indication of air speed, you could find the aircraft in an overspeed situation where the wings could literally get ripped off the fuselage.

[Mechatrommer]

Let's just take ONE aspect of this statement:

2) altitude decreasing? increase engine speed and flap, aileron etc to appropriate position.

With no indication of air speed, you could find the aircraft in an overspeed situation where the wings could literally get ripped off the fuselage.

so just let the plane do military dive, fine.

[tooki]

Note that the first item of this checklist is to disable autopilot. But this would already be the case

You would still go through the checklist from the top. If you saw Mentour Pilots video demonstration during simulation, it's a bit comical how formal it is. So in their 40 seconds, they would potentially be verbalizing each item, starting from the beginning.

So
1. "Memory item runaway stabilizer turn off autopilot"
2. "Memory item 1:Turn off the autopilot." Memory item 1 is not "If the autopilot is already off, skip to item 2"
3. The pilots would turn off autopilot
4. Then they would monitor for the result. And since MCAS is intermittent, "the result" may initially appear to be perfectly satisfactory.

You are supposed to monitor the result. You are NOT supposed to proceed to memory item 2, blindly. And if the result is satisfactory, you are supposed to stop.

That can waste a lot of time out of the 40 seconds the pilots had.

In the first incident with Indonesian Air/Jakarta, the pilots did the dance for 10 minutes, or something. I have not seen any detailed info for the Ethiopian Air crash. Just that the news said they had 40 seconds. And Ethiopian Air stated all their MAX pilots received MAX specific updated training, including the hiterto optional simulation in the interim between the Jakarta crash and this one.

Now, how pilots actually do this may be totally different in the plane. But even with the updated training, the solution may be insufficient? I'll bet if the pilot skips the first step, it's an automatic fail, in training. If you were to go "by the book," the book might kill you? I think that could be considered an insufficient solution. I imagine that strictly following procedure/protocol is supposed to be sufficient.

The protocol is there, presumably to prevent pilots from jumping to incorrect conclusions. There could be good safety reasons, statistically, that we want pilots that blindly follow protocol in this kind of situation. There could be financial pressures, as well. If the pilot cuts stab trim out of order, then he may not be allowed to later say "oops, my bad" and turn it back on. And he may have to divert the plane? Just speculating out loud. In this case, the pilot had already requested to return to the airport of departure, so he was already past that.

I never said anything about skipping checklist steps. Literally the point of a checklist is to GO THROUGH EACH ITEM.

I saw the Mentour demo long before my reply above, and yeah, it’s formal, but it would still get you through it fast.

[tooki]

Personally I think pilots should be encouraged to hand fly much more often than they do. Too much reliance on automation takes people out of the loop and encourages the mind to wander.

I agree, but in today's world...

Management: You want employees to actually have fun on the job? We pay them to be miserable.

It is my understanding that airlines do, in fact, encourage routine manual flying for precisely this reason. Many potential autopilot features (like auto-land) exist but are essentially never used.

The media perpetuates the myth that pilots don’t fly any more and just baby the autopilot in case it ever goes wrong, but this plain and simply isn’t true.

[KL27x]

I never said anything about skipping checklist steps.

Sorry, I didn't say you did. I was just making a point that even though they might think autopilot is off, that's where they might start.
 
I think 40 seconds is cutting it pretty close if you watch how long that takes.

The airline stated their faith in Boeing planes; they also stated their pilots received all updated training. This certainly suggests there is still a problem that is beyond training/knowledge.

Boeing must know more about the crash, else I don't see how they can properly fix a problem that "isn't there." They've apparently already fixed the "non-problem" and have "tentative" FAA approval.

With 300 wrongful death suits on the line and billions in business, who knows what version of the truth we well ever get.

[Brumby]

Let's just take ONE aspect of this statement:

2) altitude decreasing? increase engine speed and flap, aileron etc to appropriate position.

With no indication of air speed, you could find the aircraft in an overspeed situation where the wings could literally get ripped off the fuselage.

so just let the plane do military dive, fine.

Without an indication of air speed, you are just trading one disaster for another.  Not smart.

[Mechatrommer]

you dont get me. dont trade it if you like, just let the disaster happens, fine with me. drones flew, no pitot tubes. oh i forgot, drone is different.