"WASHINGTON—A simulator session flown by a U.S.-based Boeing 737 MAX crew that mimicked a key portion of the Ethiopian Airlines Flight 302 (ET302) accident sequence suggests that the Ethiopian crew faced a near-impossible task of getting their 737 MAX 8 back under control, and underscores the importance of pilots understanding severe runaway trim recovery procedures.
Details of the session, shared with Aviation Week, were flown voluntarily as part of routine, recurrent training. Its purpose: practice recovering from a scenario in which the aircraft was out of trim and wanting to descend while flying at a high rate of speed. This is what the ET302 crew faced when it toggled cutout switches to de-power the MAX’s automatic stabilizer trim motor, disabling the maneuvering characteristics augmentation system (MCAS) that was erroneously trimming the horizontal stabilizer nose-down.
In such a scenario, once the trim motor is de-powered, the pilots must use the hand-operated manual trim wheels to adjust the stabilizers. But they also must keep the aircraft from descending by pulling back on the control columns to deflect the elevator portions of the stabilizer upward. Aerodynamic forces from the nose-up elevator deflection make the entire stabilizer more difficult to move, and higher airspeed exacerbates the issue.
The U.S. crew tested this by setting up a 737-Next Generation simulator at 10,000 ft., 250 kt. and 2 deg. nose up stabilizer trim. This is slightly higher altitude but otherwise similar to what the ET302 crew faced as it de-powered the trim motors 3 min. into the 6 min. flight, and about 1 min. after the first uncommanded MCAS input. Leading up to the scenario, the Ethiopian crew used column-mounted manual electric trim to counter some of the MCAS inputs, but did not get the aircraft back to level trim, as the 737 manual instructs before de-powering the stabilizer trim motor. The crew also did not reduce their unusually high speed.
What the U.S. crew found was eye-opening. Keeping the aircraft level required significant aft-column pressure by the captain, and aerodynamic forces prevented the first officer from moving the trim wheel a full turn. They resorted to a little-known procedure to regain control.
The crew repeatedly executed a three-step process known as the roller coaster. First, let the aircraft’s nose drop, removing elevator nose-down force. Second, crank the trim wheel, inputting nose-up stabilizer, as the aircraft descends. Third, pull back on the yokes to raise the nose and slow the descent. The excessive descent rates during the first two steps meant the crew got as low as 2,000 ft. during the recovery.
The Ethiopian Ministry of Transport preliminary report on the Mar. 10 ET302 accident suggests the crew attempted to use manual trim after de-powering the stabilizer motors, but determined it “was not working,” the report said. A constant trust setting at 94% N1 meant ET302’s airspeed increased to the 737 MAX’s maximum (Vmo), 340 kt., soon after the stabilizer trim motors were cut off, and did not drop below that level for the remainder of the flight. The pilots, struggling to keep the aircraft from descending, also maintained steady to strong aft control-column inputs from the time MCAS first fired through the end of the flight.
The U.S. crew’s session and a video posted recently by YouTube’s Mentour Pilot that shows a similar scenario inside a simulator suggest that the resulting forces on ET302’s stabilizer would have made it nearly impossible to move by hand.
Neither the current 737 flight manual nor any MCAS-related guidance issued by Boeing in the wake of the October 2018 crash of Lion Air Flight 610 (JT610), when MCAS first came to light for most pilots, discuss the roller-coaster procedure for recovering from severe out-of-trim conditions. The 737 manual explains that “effort required to manually rotate the stabilizer trim wheels may be higher under certain flight conditions,” but does not provide details.
The pilot who shared the scenario said he learned the roller coaster procedure from excerpts of a 737-200 manual posted in an online pilot forum in the wake of the MAX accidents. It is not taught at his airline.
Boeing’s assumption was that erroneous stabilizer nose-down inputs by MCAS, such as those experienced by both the JT610 and ET302 crews, would be diagnosed as runaway stabilizer. The checklist to counter runaway stabilizer includes using the cutout switches to de-power the stabilizer trim motor. The ET302 crew followed this, but not until the aircraft was severely out of trim following the MCAS inputs triggered by faulty angle-of-attack (AOA) data that told the system the aircraft’s nose was too high.
Unable to move the stabilizer manually, the ET302 crew moved the cutout switches to power the stabilizer trim motors—something the runaway stabilizer checklist states should not be done. While this enabled their column-mounted electric trim input switches, it also re-activated MCAS, which again received the faulty AOA data and trimmed the stabilizer nose down, leading to a fatal dive.
The simulator session underscored the importance of reacting quickly to uncommanded stabilizer movements and avoiding a severe out-of-trim condition, one of the pilots involved said. “I don’t think the situation would be survivable at 350 kt. and below 5,000 ft,” this pilot noted.
The ET302 crew climbed through 5,000 ft. shortly after de-powering the trim motors, and got to about 8,000 ft.—the same amount of altitude the U.S. crew used up during the roller-coaster maneuvers—before the final dive. A second pilot not involved in the session but who reviewed the scenario’s details said it highlighted several training opportunities.
“This is the sort of simulator experience airline crews need to gain an understanding of how runaway trim can make the aircraft very difficult to control, and how important it is to rehearse use of manual trim inputs,” this pilot said.
While Boeing’s runaway stabilizer checklist does not specify it, the second pilot recommended a maximum thrust of 75% N1 and a 4 deg. nose-up pitch to keep airspeed under control.
Boeing is developing modifications to MCAS, as well as additional training. Simulator sessions are expected to be integrated into recurrent training, and may be required by some regulators, and opted for by some airlines, before pilots are cleared to fly MAXs again. The MAX fleet has been grounded since mid-March, a direct result of the two accidents."
I made the prediction a month ago and I’ll make it again. MCAS won’t ever fly again.
I think the plane will fly without MCAS with a new type rating, and where pilots specifically train on the stick feel during a power on stall.
They needed the 737 with new engines to fly like the old 737 in order to maintain the type rating and avoid training. Now that they obviously have lost that and require substantial training, I don’t see any reason to keep the system.
Some people think the plane is unflyable / inherently unstable without MCAS. I’ve only flown about 15 hours in my life, and a Cessna at that so I’m not qualified to offer an expert opinion.
But my intuition is that if that were true, if MCAS was not just a bandaid to maintain a type rating, but rather a safety critical system essential for flightworthiness, it would never have shipped in its current form in the first place. Or rather, the 737 MAX would not have been viable to begin with.
If it is the case that 737 MAX is not actually airworthy without a functioning MCAS then Boeing has fucked up an order of magnitude worse than even they appear to have at this point.
MCAS is not there because of maintaining pilot type rating.
The plane's airworthiness certificate depends on certain behavior while approaching stall and the new larger and more forward engines on the MAX changed the behavior of the machine in such way that would make it very difficult for the crew to bring the nose down at high angles of attack and high thrust.
The engines on the MAX are forward of the CoG, so they are pushing the nose up as you add thrust - exactly what you do not want when approach a stall. MCAS has been added to help with this - to help the crew push the nose down in such situation by using the stabilizer trim.
The older NG doesn't have this issue because the engines are both smaller and more to the back, so the pitch-up moment is smaller, even though it is still present. This is an inherent issue with underslung engines, all planes with engines mounted under and ahead of the CoG have this issue.
>If it is the case that 737 MAX is not actually airworthy without a functioning MCAS then Boeing has fucked up an order of magnitude worse than even they appear to have at this point.
That's BS. By that logic no Airbus or any fly-by-wire plane could ever be certified - they actually require the computer augmentations to fly. Even the 737 NG has a similar trim system that automatically trims the plane depending on speed, angle of attack and what not.
The engines on the MAX are forward of the CoG, so they are pushing the nose up as you add thrust - exactly what you do not want when approach a stall. MCAS has been added to help with this - to help the crew push the nose down in such situation by using the stabilizer trim.
The forward-backward position of the engines have nothing to do with their resultand pitch torque. What matters is that they are below the CG.
The engines being forward matters because it shifts the center of pressure forward, giving an inherent pitch-up tendency that becomes more pronounced at high angles of attack.
And like you say, all planes with underslung engines have this issue to some degree. Maybe the issue with the MAX is severe enough to cause certification problems but the stall recovery procedures for the NG apparently already call for reducing pitch angle before going to full power or you may not be able to get it down.
The fore/aft position has a lot to do with the AofA-dependent effect, as it results from the approximately cylindrical engine nacelles producing some lift when their axis is at a positive angle of attack (fuselages have a similar effect [1].) On airplanes with rear-mounted engines, this effect on their nacelles increases the longitudinal stability [2].
The high/low position matters for the power-dependent effect, and furthermore, what matters in steady-state is the pitch moment created by misalignment of thrust and drag (when accelerating, the moment around the CofG also matters.) A modest tendency to pitch down when the power is reduced is preferable to the opposite.
The issue MCAS was supposed to solve was due to the AofA-dependent effect. That's why it has AofA as an input, and not any measurement or proxy for thrust.
"The forward-backward position of the engines have nothing to do with their resultand pitch torque. What matters is that they are below the CG.
"The engines being forward matters because..."
Janoc repeated a mistake that has been made several times, attributing the problematic pitch-up to the thrust of low-slung engines, and also made some mistakes about how that effect works. I am sure you did not intend it, but by combining corrections to both the former and the latter, and not differentiating between the two, you ended up with something quite confusing, so I felt some clarification would help. The key points are that the engines do produce lift at high AofA, and it is this, not thrust, that creates the problem MCAS was intended to fix.
The key points are that the engines do produce lift at high AofA, and it is this, not thrust, that creates the problem MCAS was intended to fix.
Is that really true? I agree that the pitch-up torque from engine thrust does not vary based on AoA, but the elevator deflection needed to counteract said torque definitely depends on airspeed and hence on AoA. Even if there was no aerodynamic effect of the engines at all, the pilot would be required to trim nose-down as airspeed goes down relative to an airplane with centerline thrust.
If the thrust gets high enough, at some point this pitch torque would overwhelm the natural tendency of the airplane to pitch down as airspeed goes down and result in negative longitudinal stability.
My impression is that MCAS was intended to counteract the decreasing stick force with decreasing airspeed. Both effects, offcenter thrust and aerodynamics, combine to produce the same effect.
First you said I was repeating what you said, now you are saying the opposite?
Various good sources [1],[2] state that the issue is lift, not thrust. The article I linked to from my earlier post shows that this lift is not unique to 737 Maxes, and that it contributes to the stability of the DC-9, on account of being behind the CofG on that airplane.
The LEAP-1B engines of the Max variants are not significantly more powerful than those on NGs (they are more efficient and somewhat quieter.) As the whole problem arises from the fact that they cannot be mounted lower, it seems unlikely that an increased thrust moment is what is making the Max handling unacceptable.
> If the thrust gets high enough, at some point this pitch torque would overwhelm the natural tendency of the airplane to pitch down as airspeed goes down and result in negative longitudinal stability.
There seems to be some confusion here over what longitudinal static stability is: it a matter of the rate of change of pitching moment with respect to AofA (take a look at the links in my first post for data from real airliners, where a negative slope indicates stability.) Thrust is not a function of AofA, and consequently is not factor in the relationship that gives longitudinal stability of ordinary airplanes - and, as I said before, neither thrust nor any proxy for it is an input to MCAS (while AofA is.)
Sorry for making unclear references. My question was referring to "it is [lift], not thrust, that creates the problem MCAS was intended to fix", which I never addressed before.
The article you linked has some nice cross-sectional illustrations that makes clear that the thrust line of the MAX engine actually is higher than on the NG. I hadn't realized this but it makes sense given their larger diameter and needed ground clearance. That would indicate that the thrust dependent pitch up is likely smaller on the MAX than the NG so, if the MAX behaves worse, it stands to reason that the aerodynamic effect must dominate.
I guess the issue is that lift overall is just so much larger that a small shift in the center of lift has a large effect on pitch moment?
As for your comment about longitudinal stability, I agree that
the pitch moment due to thrust is independent of AoA and hence can't change the shape of the pitch moment curve. I gotta think about this some more.
This is completely wrong. An Airbus is not made conformant with FAR 25 airworthiness standard by software protections. It conforms to all of FAR 25 requirements without any of the safeguards in place, whereby stick inputs are directly translated to control surfaces. The airworthiness standards are predicated on aerodynamics, not software abstraction of an otherwise unairworthy design.
The idea there are switches in the airplane I can use to make the airplane unairworthy is absurd. And it is equally absurd that the aircraft flight manual and an emergency airworthiness directive would so direct a pilot to make the plane less airworthy than it already is, let alone unairworthy, as a corrective action. It's just absurd. You're welcome to provide a citation, however.
I submit FAR 25. I've read the entire thing and there's nothing in there that permits software to paper over an otherwise unairthworthy aircraft design.
It doesn't make much difference, most of what you've said is accurate to the best of my knowledge, however the thrust can be as far forwards of the CoG as it likes, that doesn't increase the pitch up moment. Having them lower may increase that moment and them having more power too. But the main issue on the Max is airodynamic lift generated by the nacelles at high AoA.
>But the main issue on the Max is airodynamic lift generated by the nacelles at high AoA.
I agree.
I did not find the original article I read about it but e.g. this article ( https://moneymaven.io/mishtalk/economics/boeing-737-max-unsa... ) contains a statement similar to what I read - a bit more aggressive and I did not actually read the article and I do not know the website (so yeah, it might be a bad and unconfirmed source). Anyway, just focusing only on the nacelles and AoA here is the statement:
>>because the engine nacelles were so far in front of the wing and so large, a power increase will cause them to actually produce lift, particularly at high angles of attack. So the nacelles make a bad problem worse.
>>I’ll say it again: In the 737 Max, the engine nacelles themselves can, at high angles of attack, work as a wing and produce lift. And the lift they produce is well ahead of the wing’s center of lift, meaning the nacelles will cause the 737 Max at a high angle of attack to go to a higher angle of attack.
Summarized, I understand this as a looping chain reaction: the higher the AoA the higher the lift produced by the nacelles, which in turn (without auto-trim by MCAS or active pilot correction) will increase the AoA which in turn will increase the lift, etc... .
>>>because the engine nacelles were so far in front of the wing and so large, a power increase will cause them to actually produce lift,
I thought that it would make the plane "tilt" upwards, but the quote mentions "lift".
Does this happen because the engines, being in front of the wing and as well very near to it are "sucking" air that would otherwise flow as well >>over<< the wing, therefore creating a lower-than-usual pressure above the wing, therefore making the airplane climb/generate lift?
It has nothing to do with power. Even cylindrical bodies produce some lift when their axis is not aligned with the airflow, and all engine nacelles do this. If they are forward of the CofG, this produces a pitch-up moment that increases with increasing angle of attack, and therefore reduces stability. The further forward they are, the more the leverage and the stronger the effect. With rear mounted engines, the effect produces a nose-down moment, which contributes to the longitudinal stability. See the links in my other post: https://news.ycombinator.com/item?id=19904685
A good way to visualize it, I've found, is imagine the low pressure area created by the wing as a bubble attached to the aircraft, "sucking" the aircraft up.
Imagine the air going under the wing as analogous to the surface of a lake.
The more the plane diverges from parallel to the airstream, the stronger the tendency for the bubble to suck through the "shadow" created in the airstream, and the plane to "skip" on the surface (the air getting pushed under the airfoil).
In reality, these are two different ways of illustrating the same force. This visualization though, let's you pick out two different aspects.
Air is a fluid. Just like water will follow the edge of a pan when you pour it, the curvature of an airfoil does the same thing with air flowing by.
Those nacelles basically enhance the surface area across which that pressure differential acts through. In short, generating extra lift, "pulling" the plane up more, (or pushing, if you prefer looking at it from the skipping stone aspect) creating more lift, all the way into a stall condition.
> the main issue on the Max is airodynamic lift generated by the nacelles at high AoA.
This simplifies the physics so much that it's wrong. It is true that there are texts attributing everything just to the "nacelles", but the engineering doesn't work the way you argue:
First, you can't put the engines on the plane without some casing for them (you have to have nacelles).
Second, if you'd try to modify the casing of the engines (the said nacelles) to have other aerodynamic properties, you'd interfere with other design goals (e.g. you can increase the fuel consumption which was the most important design goal for the MAX).
Third, the position of the engines definitely does matter for the behavior like "nose up" as long as the rest of the plane has some already determined form, which in this case is what has to be kept to have the plane being 737 and not something completely different.
In short: the 737 MAX is Boeing's equivalent of VW's diesel cars. Its behavior was determined by the prescribed design goals (by those who sold the plane): Boeing simply directly lied, selling something that can't be achieved given the "sold" design characteristics.
Just like the diesel cars can reduce nitrogen oxides production, but then they are less "efficient" than as they were sold (they need more refills, or use more fuel, i.e. cost more than advertised). Likewise, the 737 MAX-like plane could have been safer, but then Boeing could have not sold it as a plane for which there's "no need for pilot's re-certification and retraining."
At the end, Feynman conclusion of his Shuttle investigation report holds:
"For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled."
("Former head of Volkswagen could face 10 years in prison -- Martin Winterkorn and other executives charged with fraud over emissions test cheating")
I'm quite sure Boeing's "heads" won't even be charged. And no, the "nacelles" aren't the main issue. It's the fraudulent selling of the planes which can't deliver what they were sold for. Fraud.
I wonder if thrust vectoring could be of use here? I know it's primarily a military technology now, but maybe the MAX can be saved by having engines that slightly divert their thrust in very specific situations to offset this undesirable effect?
For this to be stable it would seem very elegant if you could design this in such a way that it occurred in equilibrium with any undesirable lift generated by the nascelles.
As supposed to an active measure controlled by electronics based on a separate AoA sensor, if somehow the lift caused the engine the thrust vector in opposition to the lift as a matter of physics.
I mean, in theory there’s nothing to say inherently such a design would be more reliable than triple redundant sensors and a bunch of code, but that would be pretty neat.
The Boeing whistleblower in the recent 60 Minutes Australia exposé said that's exactly why it was designed. Not only was a new type rating unacceptable, engineers were not permitted to make any design changes which would require simulator training.
> That's BS. By that logic no Airbus or any fly-by-wire plane could ever be certified - they actually require the computer augmentations to fly.
I would imagine that, to be certified, the Airbus planes had to demonstrate that:
a) The computer augmentation was unlikely to fail,
b) when failed, the plane was still flyable enough, and
c) the pilots were trained to fly the plane when the augmentation failed. (I believe this is called “direct law”, and I would be utterly shocked if pilots are not trained to fly the plane in direct law mode.)
They are so trained, and the airplane is still considered airworthy in that mode, including in particular as it related to this conversation, the FAR 25 positive static stability and stall behavior requirements.
Fly by wire systems are classified as safety critical and require much more redundancy. The Boeing 777 has three flight control computers, four sets of low-level controllers, and a fault-tolerant air data sensor system.
Bad air data (airspeed, altitude, angle of attack) has been a big problem in the past. Each of those involves a sensor that sticks out through the skin, and can potentially be damaged or plugged. Remember Air France 447, where the airspeed sensors iced up. Those things need redundancy.
>The engines on the MAX are forward of the CoG, so they are pushing the nose up as you add thrust - exactly what you do not want when approach a stall. MCAS has been added to help with this - to help the crew push the nose down in such situation by using the stabilizer trim.
And the pilots canot push the nose down in such a situation by actuating the stabilizers?
I've never flown an airplane, but as far as I know, MCAS doesn't have unique access to controls that the pilot does not. (eg, it's different from ABS in a car where the driver does NOT have 4 individual brake levers).
It seems to me if the pilots are trained and certified on the Max, they'll do a better job flying it than a poorly designed system.
OTOH, the system is also fixable, but it is becoming a matter of trust.
>I think the plane will fly without MCAS with a new type rating, and where pilots specifically train on the stick feel during a power on stall.
Maybe. Keep in mind MCAS was a system put in place to mitigate a catastrophic risk. If you remove it, you no longer have that mitigation - meaning you either have to argue that increased training can be a substitute (I would be skeptical of that if I was a regulator) or come up with something else. It's a shitshow.
I think it all comes down to certifying bodies other than the FAA. FAA seems happy to go along with whatever Boeing wants (as evidenced by the 'self certification' efforts of the agency); it remains to be seen if the EU or another major market airspace will capitulate.
>FAA seems happy to go along with whatever Boeing wants (as evidenced by the 'self certification' efforts of the agency)
I strongly suspect this is way overstated.
I think there's a misinformed view of what a regulator is and the role they play. Regulations and regulators don't work within a system where there's a preponderance of cheating. This is why you can't go do a corrupt country and simply legislate corruption away (all corrupt countries have very good laws on the books). Regulators necessarily exist within a largely honest system. The FAA, and FDA and other regulatory agencies do not go around testing every claim made by every manufacturer - though they reserve the right to. If that was the case the economy would grind to a halt and no product would be released. This would be akin to the IRS auditing every tax return every year.
The second point about regulations is that like security or marketing, there is no end-game. You can always have more regulations, just as you can invest in more security and more marketing - but at some point, you will have diminishing resulting leading to negative outcomes. In the specific case, Boeing put themselves under a certification process meant for certifying non-trivial but also non-major incremental upgrades (i.e. no fuselage and wing redesigns). The process is less onerous because the aircraft upgrades don't introduce new significant risks. This process does not apply to the 737Max8. Boeing was wrong to put the plane in that category and they will pay dearly for this and will serve as a cautionary tale moving forward, but let's not throw out the regulatory regime that resulted in making the airline industry incredibly safe, and the safest it has ever been in history.
> This process does not apply to the 737Max8. Boeing was wrong to put the plane in that category...
If you, the layperson, can make this statement, then why wouldn't the regulators make this obvious determination? Regulators should have said "No, Boeing, you're not eligble for this process because of XYZ"
So, if they can't make such a high-level determination, what exactly are they doing? Not much, it seems.
>If you, the layperson, can make this statement, then why wouldn't the regulators make this obvious determination?
No doubt FAA will re-evaluate their process moving forward, but it depends on what Boeing wrote in their submission.
If they didn't put in pertinent details for the FAA to make this determination, then it would only get caught if the FAA did a deeper audit (which they reserve the right to but it may not be something that happens with every submission of this type).
On the other hand, it may have been the case that Boeing did put in the pertinent details and FAA was negligent.
So neither FAA nor Boeing are in the clear. Boeing is in trouble no matter what - they fucked up real good and they are ultimately responsible for the safety of their airplanes - but it remains to be seen how negligent the FAA is.
I don't have experience with aviation and the FAA, but I do work on a regulated product and deal with the FDA quite a bit.
It is a critical system. Part of the problem is that AOA sensors and MCAS were never considered critical before the software changes made for the 737 MAX and uprating them to critical was overlooked (which would have forced a lot more testing of them).
There is no established requirement that says software can't be used to correct for pitch up on acceleration. The software just has to work properly.
> uprating them to critical was overlooked (which would have forced a lot more testing of them).
Was this "overlooking" accidental or designed specifically to ignore the forced testing? This is the question that might be hardest to answer, but would justify criminal charges for the deaths in my mind.
True. Would be impossible to prove without a whistle blower. Boeing hasn't been charged with a crime so there won't be any subpoenas of e-mail records, etc. It is within the realm of possibility, but I certainly hope that was not the case. I lean very slightly toward thinking it was an accident as the whole program was rushed and give the engineers the benefit of the doubt (innocent until proven guilty) - however, it would not take much evidence to sway my mind.
I’m not sure a type rating works that way. A Boeing 717 has the same type rating as a Douglas DC-9. The two aircraft have similar airframes and some common systems but are quite different as one was built decades later. What the FAA does in these cases is requires more training. A DC-9 rated pilot who has only flown DC-9s will not fly a B717 without several hours in a simulator. The MAX will certainly fly again. I suspect some form of a redesigned MCAS in addition to more training for pilots will be the end result. The airlines may have to spend some money training but they’ll still be left with B737 type rates pilots who can fly B737NG and B737MAX which is where the cost savings is derived.
Is there any indication that MCAS is a problem when it is provided with accurate AOA data? I don't think so, but I'm willing to be educated.
My understanding is that MCAS only goes haywire when it gets haywire sensor data--specifically, data that indicate an AOA that is much higher than reality.
So much depends on the investigations, which are currently in progress. But it seems possible that the technical solution could simply be: require 3 AOA sensors and require agreement between 2 out of 3 to activate MCAS. Of course, there are probably significant business and regulatory issues to consider/fix as well.
Yes, but there's also the problem of just how haywire it goes --- forcing the horizontal stabilizer in an unflyable position from which it can't be moved (the subject of this article).
I'm not sure how easy it is to just bolt a third AoA sensor onto an existing airframe. OTOH, the MCAS system could at least be redesigned to take BOTH AoA sensors into account.
> I think the plane will fly without MCAS with a new type rating
Without the 737 type rating, the value prop of the 737 MAX is lost. There would be no point in continuing the program (at least at anywhere near its current scale).
That's a red herring, the MCAS issue has nothing to do with the type rating. MCAS is not something the crew was even supposed to be aware of, even less trained on, so no effect on type rating at all there.
What it may cause problem with is the certification of the plane as such - MCAS was required to correct the behavior of the plane in high speeds/high thrust and high angle of attack situation where the new larger engines made it difficult to bring the nose down (and avoid stalling).
However, that's not something that cannot be fixed. Airbuses fly full of similar augmentation systems and there is no problem with them, so it is certainly doable.
> MCAS is not something the crew was even supposed to be aware of, even less trained on, so no effect on type rating at all there.
Exactly this is the argument for a type rating problem. If pilots would have to undergo heavy training for the MAX there would probably have been a new type rating. There are some experts out there who argue that this is why Boeing tried to hide the existence of MCAS. Pilots should have been informed about MCAS and the reason it is there before all this started.
Yes but the point is that it was thought this isn't something that was to be required - and had the system been correctly implemented (and handled by the crews), there wouldn't have been a problem.
I don't think Boeing was intentionally "hiding" the existence of such a system, they just didn't consider it critical enough to bother the crews with it because it was supposed to come into play only in a very unusual flight regime - manual flight, flaps up and high angle of attack. There was no special "MCAS training" needed because any failures fall into the "runaway stabilizer" emergency procedure.
It just didn't work out exactly that way, unfortunately.
If I recall correctly, the system has been mentioned in the documentation - not prominently and not something the crews would have been forcibly made aware of, but given that the plane certification depended on it and it had to be certified, they couldn't exactly "hide" it from the regulators.
Keep in mind that the procedure for the stabilizer problem is the same since 737 has been first introduced in the late 60s. Had it been correctly followed, likely neither of the crashes would have happened, MCAS or no MCAS.
The Lion Air crew didn't identify the problem as a possible runaway in the first place for whatever reason and the Ethiopian crew made several (understandable, but still) mistakes, such as cutting the trim system out while heavily out of trim, not reducing the excessive speed and then finally re-activating the electric trim (which is against the procedure) when they couldn't re-trim the plane manually due to the resulting excessive forces. Which has directly caused the crash once the MCAS has re-activated.
That's a red herring, the MCAS issue has nothing to do with the type rating.
In the 60 Minutes Australia piece a Boeing employee explains that MCAS was set up to only use one alpha vane specifically to avoid being classed as a difference that required simulator time for pilots transitioning. Whether or not that's true (or how that even works), I couldn't tell you.
How do you "insert yourself in the loop" on a FBW plane when the computers conk out? There is manual reversion on an Airbus but it is very minimal - if you ever end up in a situation where that is needed you are having a very very bad day.
On the other hand, 737 is actually fly-by-wire - literally, because actual steel wires are still connected between the cockpit controls and the control surfaces. So if the electronics dies and the hydraulics isn't working neither for whatever reason, the plane can still be controlled.
These "magic boxes" are there to augment the control characteristics, such as the "feel" of the control column (which is normally lost due to the hydraulics driving the control surfaces). Or, in the case of MCAS, to help with an aerodynamic issue. Both are hardly unprecedented - the last large transport plane that didn't rely on any computers/electronics for controls was maybe the 707.
The problem is the poor design of the MCAS software/system, not that it exists as such.
Type ratings don't really work like that. 90% of the type rating is systems knowledge, emergency procedures etc. If a new type rating was needed there would be a conversion course from the existing type rating.
Unflyable or unstable isn't nearly the same thing as uncertifiable. People have flown a whole lot of time in uncertifiable aircraft (see the whole experimental aircraft sector) but transport category aircraft are supposed to exhibit as full proof as possible behaviour.
I bet you it flies again with MCAS. I'll also bet it flies with barely a few hours of conversion training and that there aren't any accidents directly related to the updated MCAS system.
This isn't true though, it seems perfectly possible to certificate transport-category airplanes with sketchy stall characteristics, since T-tail planes can end up in unrecoverable situations by blanketing the tail in a stall.
Twin-engine planes also are not required to be recoverable from an engine failure at speeds below Vmc.
I don't know what the certification requirements are, but it seems they rely on being able to guarantee that you never get below a certain speed since in multi-engine airplanes there's always a low power/high thrust regime that's potentially uncontrollable.
While the FAA has allowed it, not all aircraft have been certified to fly in all airspace without safety systems to prevent stalls. Give the D. P. Davies Interview, recorded by the Royal Aeronautics Society a listen, and you'll learn how the 727 was not allowed to be certified in the U.K. without a stick pusher equipped.
I am not very concerned with this particular problem anymore since it will get a satisfactory solution due to its publicity.
What I am worried about and think should get more focus is other potential issues and corners that has been cut by the Boeing management. I have note read a single thing that Boeing has handled well in this massive disaster. This lack of judgement, short term profit seeking & cover up mentality is what media should spend more time on. It feels to me that they used up a lot of their trust capital to be allowed to perform any type of self-certification
More to the point, rushing execution in order to meet economic promises, and arbitrary cost-cutting, are symptoms of large systems. Capitalist, socialist, same diff, when it comes to the behavior of bureaucracies.
It turns out that the problem isn't capitalism or communism, it is arseholes, and they are evenly distributed across all economic systems. What we need to develop is some sort of arsehole distraction system to keep them away from anything important or fragile.
I say this often: “it turns out the problem was people the entire time”.
It is very popular to imagine that the negative circumstances we find ourselves in are enforced upon us by some malevolent outside force, such as “politicians” (scare quotes). And while sometimes certain social structures are to blame, more often than not it’s merely the failings of humanity in general that’s to blame. Greed and laziness know no cultural bounds, for example.
Psychopaths in leadership/management create problems in all economic systems. Don't blame capitalism. Evil people are evil. There are good people at Boeing but psychopaths have a natural ability to gain power in all kinds of social structures.
From my perspective, it seems evil behavior is rewarded and encouraged more often in this system.
I want to blame capitalism, but you are right. Ultimately people are to blame, and no one system can protect or prevent an evil person from doing evil things.
More often compared to what? What is your knowledge of history? Go read Solzhenitsyn and learn about the hundreds of millions killed and tortured under Communism.
I doubt that in a non-capitalistic system the 737 MAX would not have had similar problems as in no system resources are infinite. Budgets, even in the most socialist country, are limited. The question is just WHO limits the budget: private enterprises (or their shareholders), co-ops or politicians.
I'm not sure about that. There is genuine value created by avoiding the need for a significant fraction of the world's airline pilots to use scarce and expensive simulators all at the same time, and Boeing's claim that they could pull it off would have been just as attractive under socialism. Too many people trusted that claim without verifying it, but wishful thinking is a pretty universal human trait, and I don't think that socialists are immune to it.
> The U.S. crew tested this by setting up a 737-Next Generation simulator at 10,000 ft., 250 kt. and 2 deg. nose up stabilizer trim. This is slightly higher altitude but otherwise similar to what the ET302 crew faced as it de-powered the trim motors 3 min. into the 6 min. flight, and about 1 min. after the first uncommanded MCAS input
That's not slightly higher, that's a lot higher. The graphs in the preliminary report [0] note that the flight starts at around 7,500 ft. since Addis Ababa Bole International Airport is 7,625 ft. above sea level [1]. Their radar altitude "about 1 min. after the first uncommanded MCAS input" was ~1,000 ft. The highest they got above ground was around 6,500 ft. around 5 mins 30 seconds into the flight.
> The excessive descent rates during the first two steps meant the crew got as low as 2,000 ft. during the recovery.
If you assume they meant they were 10,000 ft. above sea level in the first quote, then this quote means they'd be 5,625 ft. under ground at their lowest point.
So if you run into this problem at 7,500’ above ground level you might just be okay.
This is why having the ability to disable MCAS without disabling electric assisted trim is so important and such a big factor in this accident. Additionally its raised another possible accident scenario:
The runaway trim cutouts were put in place after electric trim in case the switch (or some other aspect of the electric trim) got stuck in an on position that would cause the trim to go to full deflection one direction or the other, hence runaway trim. And hence if you notice this happening you cutout the power to the electric trim completely.
But now we know you can’t manually trim the aircraft above certain speeds with some trim levels. Should there be an expectation you can? What if runaway trim happens and ran to full deflection before a pilot cut it out. That would be equivalent to the same conditions experienced in this accident - the pilots now needing to manually trim but being unable to.
So on top of needing separate cutouts for the MCAS and electric trim, should the manual trim not be investigated for installing some higher ratio gearing/pulleys to enable over coming the forces experienced here?
Right, which is why I'm saying that means there's been a failure scenario that existed even prior to MCAS. This is how they have demo'd the problem in some flight simulators for this. The electric trim can without MCAS erroneously drive the trim into this situation before being cut out. This is why the cutout switches were put in place decades ago. If this happens in some situations it could leaving the plane in a position where the trim couldn't be manually adjusted due to forces. We now know this situation also exists. Is it so rare, or resolvable in other ways, so it shouldn't be addressed? Or does there also need to be a fix for the manual trim by adjusting the gear/pulley ratio such that the pilot could overcome these forces?
What I read and makes sense the way to adjust the manual trim if the force on the stabilizer is too high is to push the nose down to unload it and crank on the wheel. Probably when this was needed the pilots killed the electric trim long before it got too far out.
So I agree the manual trim has always been kinda scketch.
Perhaps the intent was to perform the test at the same air density, as many aerodynamic properties, including stalling speed (and also engine performance) depend on density.
Also, it is more useful to know how much height was needed to recover, than that there was not enough. The latter can easily be determined from the former.
Is anyone concerned that even in the face of re-training, that two aircraft could find themselves in this position within the first year of operations?
Let's say you were on one of these aircraft and the pilots were able to recover. How terrifying would that be? This is the designed behavior?
Besides retraining, it seems that there must be upper-body strength tests that 737 MAX pilots will have to pass to ensure that they will be able to manually trim the stabilizer.
Poe's law 'n all that -- are you serious? I mean, yes, I've seen the video, but presumably it won't be safe to require pilots to execute the manual trim, yeah? IMO either the design is unsafe and must change or there exists a way to safely fly without ever needing manual trim.
I'm serious. If the final backup system is to revert to mechanical cranks, pulleys, and cables to move the horizontal stabilizer, then the pilots have to be able to exert sufficient torque and power on the control wheels to do so.
The 777 and 787 are full fly by wire. The control system has all the sensor inputs, all the actuator outputs, and a model of how the aircraft is supposed to behave. If the aircraft isn't doing what's expected, the flight control system will fault and drop to a dumber control mode, giving the pilot more control rather than driving control surfaces to their limits trying to correct the wrong problem. Those systems are managing trim, like MCAS, plus a lot more.[1]
MCAS is dumb. It has few inputs and one output. It has no overall model of aircraft behavior. It just detects a bad angle of attack and cranks the trim to bring the nose down.
The 777 is supposed to handle like a 737. The cockpit controls are reasonably similar, although different enough that transition training is required. Unfortunately, there's no "small 777", a gap in Boeing's product line the 737 Max was supposed to fill. (Or small 757, 767, or 787 variants. How did they go half a century without a new sub-200 seat aircraft?)
Link to the forum post explaining how to execute the yo-yo manoeuvre, mentioned in the article. What is interesting was one pilot mentioned that the yo-yo procedure dates all the way back to the 707 for a runaway trim. The 737 was derived from the 707 and 727. The 707 was the first Jet Airliner.
The 707 is interesting as the UK air Registration Board refused to certify it in its original form. It had insufficient rudder authority in the case of certain engine-out scenarios because Boeing hadn't implemented full rudder boost, and the pilots therefore couldn't overcome yaw. That had killed two crews in training. In order to save the big BOAC order Boeing had to fit full boost and a tail strake, which they quietly adopted for all subsequent 707s
Then there was the 727 that had four fatal crashes within six months of entering service, which was put down to inadequate training. But it was later found that the high fatality rate was because Boeing had routed fuel lines along the belly, which severed in a hard landing and caused immediate fires. And they severed easily because bizarrely they were aluminium, in order to bring the weight within target...
A little pedantic, but: the 707 was one of the first jet airliners. Britain's de Havilland Comet beat it into service by six years, and the Soviet Union's Tupolev Tu-104 by two.
This focus on MCAS only is tantamount to a red-herring because there are many more changes to the 737 NG and MAX that the FAA allowed with little or no oversight. The absurdly-crude Ducommun critical structural parts debacle gained almost no media attention in 2010, and several people have died due to fuselages breaking-up in runway overruns and hard landings where previous and similar aircraft maintained fuselage integrity. 737 NG, MAX, 787 and all other Boeing aircraft developed in the past 25 years need to be seriously reexamined for safety risks and deficiencies.
1: multi billion dollar U.S. aviation company that employs thousands of employees statewide screws up
2: the aviation company offers to send consultants to "help" investigate the cause of the crash, and not falsify or change the narrative of the story in of the African operated flight whatsoever
3: the reputable Airlines that used their plane, decided to do their own untampered investigation and found the aviation company at fault
4: leaks get out via a few ethical employees at the aviation company that the plane was not tested properly and was rushed to market to compete with a rival European aviation company
5: CEO semi apologizes and the company has nightmare of pr
One thing I don't understand is how the pilots can override the operation of the electric trim system by holding the manual trim wheels and yet the electric trim appears to be able to overcome airloads on the jackscrew that the pilots can't.
In light aircraft the design of autopilot servos normally includes a clutch so that the pilot can override the servos. But that limits the force that the servo can apply.
I guess it could have some kind of torsion sensing system which cuts the power to the trim motor but I've been unable to find this detailed anywhere.
Well, the "Stabilizer wheel - grasp and hold" item of the procedure doesn't necessarily mean the crew will be able to override it. That's the last ditch option on the checklist, when everything else has failed, even turning the motors off (there could be e.g. a short circuit powering the motor).
There is no clutch on the 737, AFAIK, because the forces are such that it wouldn't work.
And this simulator video shows both how the 737 horizontal stabilizer is controlled and what kind of forces are on the trim wheel and control column when severely out of trim:
Should give you an idea what were the Ethiopian pilots facing when unable to use the electric trim, plane trimmed nose heavy by MCAS and at high speed ...
My impression is that, at least partly, it's a result of the trim wheel being connected through a very long mechanical linkage to the tail, while the motor is right there. The pilots also don't have great mechanical leverage to turn a fairly small wheel in the direction they have to.
How do these simulators work? What is the underlying model that gives sim users assurance that the simulated behaviour corresponds to real world aerodynamics and eg engine behaviour?
There are several levels of "Full Flight Simulators" [1] defined by the FAA: "There are currently four levels of full flight simulator, levels A - D, level D being the highest standard and being eligible for zero flight time (ZFT) training of civil pilots when converting from one airliner type to another."
The requirements and process for testing simulators for certification is documented as well [2]. Its not clear from the article exactly what simulator they were using though.
This is slightly off topic, but I read write ups like this and wonder how flight simulators like this work. Do they have to program in the details of the accident, or is it just a standard Max 737 simulator?
In flight simulators, every aspect of the environment is able to be controlled by the user. Programming in the details of the accident involves setting up the same conditions as were present during the accident including altitude, airspeed, etc. Then the user/instructor would activate specific malfunctions that will cause the plane to behave like it did during the accident. Now the pilot is, for all intents and purposes, flying the same plane in the same conditions that the pilots were during the accident.
Would a possible recovery method be to roll the plane somewhat so that when the nose is forced down, it doesn't take the plane directly toward the ground?
No. Now a substantial part of the lift no longer points up because you use it for turning instead. That's why when you turn you have to increase lift (i.e. you pull back a bit, or adjust trim if it's a shallow turn that takes a while to complete) or the nose drops, to compensate for the loss of lift by not using all of it to remain at the original altitude.
Remember that unless you have negative or zero g you always have positive lift even with a nose pointed down, so taking away some of it won't help you, you'll just help gravity. You could slow down (forward speed) - if you can, that is. Banking won't help your descend rate all else remaining equal.
Well, I'm sure they'd prefer being upset over being dead... although the likelihood of successfully rolling over and flying a negative-g recovery in a transport airplane seem dubious at best. Transport category airplanes are only certified to -1G so an inverted pullout would definitely exceed those limits.
The reason for the nose down is to prevent a stall.
Rolling the plane in stall conditions can cause a spin, which is very, very bad. A spin is an aggravated stall which is harder to recover and puts a lot of stress on the airframe.
While almost all planes can safely recover from stalls given enough altitude and pilots are trained for it, spins are normally limited to aerobatics.
While that's why the system exists, in this case we're not near a stall, we're just bringing the nose down because the system erroneously thinks we're near a stall.
Found a forum thread about it from some aviation enthusiasts [1]. Some choice excerpts they made:
‘A simulator session flown by a U.S.-based Boeing 737 MAX crew that mimicked a key portion of the Flight 302 accident sequence suggests that the crew faced a near-impossible task of getting their 737 MAX back under control, and underscores the importance of pilots understanding severe runaway trim recovery procedures.’
‘What the U.S. crew found -. Keeping the aircraft level required significant aft-column pressure by the captain, and aerodynamic forces prevented the first officer from moving the trim wheel a full turn. They resorted to a little-known procedure to regain control.’ (YoYo Roller Coaster)
The excessive descent rates during the first two steps meant the crew got as low as 2,000 ft. during the recovery.
...
‘The simulator session underscored the importance of reacting quickly to uncommanded stabilizer movements and avoiding a severe out-of-trim condition, one of the pilots involved said. “I donʼt think the situation would be survivable at 350 kt. and below 5,000 ft,” this pilot noted.’
“This is the sort of simulator experience airline crews need to gain an understanding of how runaway trim can make the aircraft very difficult to control, and how important it is to rehearse use of manual trim inputs,”
That's why it is so essential to recognize and stop the runaway trim (for whatever reason) ASAP or you could get into a situation where it would be physically impossible for the crew to do so.
Most likely aerodynamically relieving trim wheel pressure by intentionally letting the nose of the plane drift down. This allows the pilots to more easily crank the trim wheel and fix the issue.
If I remember correctly, this procedure was specifically described in older 737 manuals, but had been dropped more recently.
It's a "roller coaster" procedure where the plane is pitched down to lessen aerodynamic forces on the stabiliser, allowing the trim wheels to be manually turned, then pitched up to regain altitude. The process is repeated as necessary until the trim wheels can be manually moved without it.
Interesting, it seems similar to how large sailboats need to head into or down wind when attempting to trim sails in some conditions? (to reduce pressure on the sails to allow trimming to be doable by hand)
Yeah, with the important difference that in sailing you lose speed (and heel) when heading into the wind, whereas a plane will go even faster towards the ground... Scary!
Yes. Mentour mentions it in the video I have posted above.
You unload the stabilizer by pitching down for a few seconds so that you can trim it back somewhat.
However, that is not something you can do close to the ground (you lose altitude, obviously) and the "vomit comet" ride won't be appreciated by the passengers neither.
Basically in this situation, even though the nose of the plane is uncontrollably pointing down, to get it to rise again requires that you momentarily push the nose down even further to allow the trim control to unjam and start working again.
If I remember correctly, this is a procedure that once was on the 737 flight manual as an recommended procedure but later Boeing removed it. I recall reading about in an article here at HN.
Sounds more like we're blaming the manufacturer for concealing the fact that pilots needed to be specifically (re-)trained for this plane if they were to have a chance of handling it successfully.
Exactly. But such training should not be necessary because this problem was created by Boeing.
Pilots really should have extensive emergency training, but that doesn't mean it's ok for the manufacturer to Add a new failure case to a previously fine aircraft.
[edit: leaving this here because I wrote so much :D
I totally misread/interpreted the comment I was replying to, and am entirely in agreement :D
]
In this case the "fix" is called the "roller coaster", but Boeing has removed the documentation regarding this from the manuals - specifically the Pilots could have 100% perfect recall of the entire 737-MAX documentation and it wouldn't have helped.
This is 100% Boeing's responsibility:
* They designed a system that overrode pilot control so that they could avoid a requirement to retrain pilots. MCAS existed specifically to maintain the perception of identical flight behaviour entirely to avoid the need for recertification and training. That is the only reason MCAS existed.
* They had minimal (no?) documentation indicating it existed
* They made a warning system to report that the system was faulty - literally they shipped a broken safety system: the warning was intended to be freely available, but it was incorrectly tied to a costly system, so even though it was meant to be a standard feature, it was not present.
* The recommended recovery technique from their documentation was not physically possible, even when you knew exactly what you were meant to do.
* The only technique to recover - the roller coaster - had been deliberately removed from the training and flight manuals. The pilots who managed to recover were only able to recover because they knew of the "solution" from old manuals that were published online after both crashes had occurred, by some non-Boeing website. Even then the pilots in the article said that they didn't believe it would be possible to recover once you got below 5000 feet.
Now, exactly how was this in any way the pilot's fault? They were fully trained, they had clearly read the manuals provided by Boeing (they knew they were mean to be using the manual wheel thing), the warning system was non-functional due to a construction/software error in Boeing's control software, etc.
No, the only way to recover as far as I can tell is to do the roller coaster thing.
From what I can make out MCAS points the plane at the ground, so it accelerates - I don’t know if it also increases thrust but if it’s meant to be anti-stall maybe it does?
The only way to reduce load once you’re at those speeds with a faulty plane (as far as I can understand it) is to keep pointing to the ground and move the trim control a bit, pull up again (which reintroduces the load) to gain altitude, then point down again and repeat until you’ve defeated the unsafe aircraft.
That's pretty much the opposite, right? That the pilots would need to know an extreme emergency procedure. In other words you'd need a Sully level pilot to survive (and unsurvivable below a certain speed and height). If you ask me that's a damning indictment when a flawed autopilot results in a state like that.
Sully didn't do anything hard from a piloting standpoint. He made the right decision to ditch in the Hudson. Once that decision was made, it was just a long glide. Plenty of room. He had the full Airbus flight control system doing most of the work. Read "Fly By Wire", by Langewiesche, which goes into this in great detail.
The Ethiopian Air pilots faced a very tough problem. They were nosing down for an unknown reason, the standard recovery procedure didn't work, and they didn't have much altitude.
A Sully level pilot is one with extensive experience, particularly in edge cases. There are many such pilots... They have decades of experience that include pre-automation days.
Don't underestimate the accumulated knowledge and experience that gives one the ability to make a series of very important decisions in an extreme stress situation.
This was disproven. And is a major plot point in the movie about Flight 1549 (Sully).
From Wikipedia:
> The NTSB used flight simulators to test the possibility that the flight could have returned safely to LaGuardia or diverted to Teterboro; only seven of the thirteen simulated returns to La Guardia succeeded, and only one of the two to Teterboro.
> Furthermore, the NTSB report called these simulations unrealistic: "The immediate turn made by the pilots during the simulations did not reflect or account for real-world considerations, such as the time delay required to recognize the bird strike and decide on a course of action." A further simulation, in which a 35-second delay was inserted to allow for those, crashed.
The link you included says the opposite of what you're implying.
> under the circumstances the captain's decision to ditch into the Hudson River was the better choice, documents released Tuesday by a federal safety panel said
> He also would have had no way of knowing that he would be successful, and therefore would have been risking the possibility of a catastrophic crash in a densely populated area.
No, we're back to, "why the heck didn't Boeing honor the regulatory requirement for simulator training, and tell American it was the best you could do?"
Clearly, the condition is recoverable, with simulator training. Throw in a triple redundant AoA sensor on top of the training, plus the inclusion of MCAS implementation details in the training material, and things would likely have had a much higher chance of being dealt with without loss of life.
Would Boeing have had a harder sell? Yes. Would there have been a non-zero chance of losing out on the order? Yes.
Would 300+ people currently be pulp? I think not. I have a great deal of respect for the skill pilots develop. They are only human however, and are therefore subject to the same weakness of not being omniscient.
If MCAS was the bullet, Boeing's opaque communication, and refusal to not game aviation certification regulations was the gun. Of that, I'm fairly certain.
What plays the trigger pull in this sordid affair is the last thing that remains to be declared, but there are some seriously bright neon signs pointing at either somewhere in Chicago, or at the culture as a whole at Boeing.
Let the digging continue. Let's see how deep the rabbit hole goes.
Didn't read the entire thing.
Going to register later.
I actually wonder whether anyone with access to the simulator has tried pulsing the trim switches in sub 4 second intervals. Which depending on how the trim switches work, (I.e. are they continuously coupled, or do they act by moving around a discrete set-point) might be able to fend off MCAS by constantly resetting the time until reactivation long enough to give the trim system time to neutralize.
If something like that were done, a recovery below 5000AGL may be possible.
It's not by a long shot the most intuitive approach out there, and it doesn't generalize well, but for the specific MCAS behavior that's been revealed it may work.
(If there's any 737 pilots out there with insight, I'd be giddy if you'd offer some perspective.)
It sounds like the simulators all this testing were on didn't actually simulate MCAS, so the "fault injection" was done manually. Testing in the way you describe wouldn't provide meaningful results without an MCAS-capable simulator.
Hmmm. Then unfortunately we'll have to wait for that to be implemented I suppose. If the algorithm is as it has been described, it may work, but at this point is moot. The damage has been done.
IIUC, those who did overcome the problem in the simulator lost around 8,000 feet of altitude in the process (despite presumably going into the simulator already knowing what they were going to be facing). It's not clear to me that the Ethiopian plane ever had enough altitude (above the ground) to allow for such a recovery.
If I recall, it did stay at around 12k altitude for a stable period of flight where MCAS was being successfully neutralized. It's in the Ethiopian report. The airspeed was way above limit, though, which would make everything harder, including the manual retrimming.
Remember that they left an airport about 7000 feet up and headed over mountains. It's unlikely the plane was ever more than a few thousand feet above ground level.
Thanks, I'd forgotten the numbers. They aren't quite as bad as you say. The Ethiopian report shows their radar elevation around 14k feet before the descent, so it seems they made it to around 7k feet AGL for a few minutes. Not enough to be able to give up 8k feet during a recovery, I agree.
Details of the session, shared with Aviation Week, were flown voluntarily as part of routine, recurrent training. Its purpose: practice recovering from a scenario in which the aircraft was out of trim and wanting to descend while flying at a high rate of speed. This is what the ET302 crew faced when it toggled cutout switches to de-power the MAX’s automatic stabilizer trim motor, disabling the maneuvering characteristics augmentation system (MCAS) that was erroneously trimming the horizontal stabilizer nose-down.
In such a scenario, once the trim motor is de-powered, the pilots must use the hand-operated manual trim wheels to adjust the stabilizers. But they also must keep the aircraft from descending by pulling back on the control columns to deflect the elevator portions of the stabilizer upward. Aerodynamic forces from the nose-up elevator deflection make the entire stabilizer more difficult to move, and higher airspeed exacerbates the issue.
The U.S. crew tested this by setting up a 737-Next Generation simulator at 10,000 ft., 250 kt. and 2 deg. nose up stabilizer trim. This is slightly higher altitude but otherwise similar to what the ET302 crew faced as it de-powered the trim motors 3 min. into the 6 min. flight, and about 1 min. after the first uncommanded MCAS input. Leading up to the scenario, the Ethiopian crew used column-mounted manual electric trim to counter some of the MCAS inputs, but did not get the aircraft back to level trim, as the 737 manual instructs before de-powering the stabilizer trim motor. The crew also did not reduce their unusually high speed.
What the U.S. crew found was eye-opening. Keeping the aircraft level required significant aft-column pressure by the captain, and aerodynamic forces prevented the first officer from moving the trim wheel a full turn. They resorted to a little-known procedure to regain control.
The crew repeatedly executed a three-step process known as the roller coaster. First, let the aircraft’s nose drop, removing elevator nose-down force. Second, crank the trim wheel, inputting nose-up stabilizer, as the aircraft descends. Third, pull back on the yokes to raise the nose and slow the descent. The excessive descent rates during the first two steps meant the crew got as low as 2,000 ft. during the recovery.
The Ethiopian Ministry of Transport preliminary report on the Mar. 10 ET302 accident suggests the crew attempted to use manual trim after de-powering the stabilizer motors, but determined it “was not working,” the report said. A constant trust setting at 94% N1 meant ET302’s airspeed increased to the 737 MAX’s maximum (Vmo), 340 kt., soon after the stabilizer trim motors were cut off, and did not drop below that level for the remainder of the flight. The pilots, struggling to keep the aircraft from descending, also maintained steady to strong aft control-column inputs from the time MCAS first fired through the end of the flight.
The U.S. crew’s session and a video posted recently by YouTube’s Mentour Pilot that shows a similar scenario inside a simulator suggest that the resulting forces on ET302’s stabilizer would have made it nearly impossible to move by hand.
Neither the current 737 flight manual nor any MCAS-related guidance issued by Boeing in the wake of the October 2018 crash of Lion Air Flight 610 (JT610), when MCAS first came to light for most pilots, discuss the roller-coaster procedure for recovering from severe out-of-trim conditions. The 737 manual explains that “effort required to manually rotate the stabilizer trim wheels may be higher under certain flight conditions,” but does not provide details.
The pilot who shared the scenario said he learned the roller coaster procedure from excerpts of a 737-200 manual posted in an online pilot forum in the wake of the MAX accidents. It is not taught at his airline.
Boeing’s assumption was that erroneous stabilizer nose-down inputs by MCAS, such as those experienced by both the JT610 and ET302 crews, would be diagnosed as runaway stabilizer. The checklist to counter runaway stabilizer includes using the cutout switches to de-power the stabilizer trim motor. The ET302 crew followed this, but not until the aircraft was severely out of trim following the MCAS inputs triggered by faulty angle-of-attack (AOA) data that told the system the aircraft’s nose was too high.
Unable to move the stabilizer manually, the ET302 crew moved the cutout switches to power the stabilizer trim motors—something the runaway stabilizer checklist states should not be done. While this enabled their column-mounted electric trim input switches, it also re-activated MCAS, which again received the faulty AOA data and trimmed the stabilizer nose down, leading to a fatal dive.
The simulator session underscored the importance of reacting quickly to uncommanded stabilizer movements and avoiding a severe out-of-trim condition, one of the pilots involved said. “I don’t think the situation would be survivable at 350 kt. and below 5,000 ft,” this pilot noted.
The ET302 crew climbed through 5,000 ft. shortly after de-powering the trim motors, and got to about 8,000 ft.—the same amount of altitude the U.S. crew used up during the roller-coaster maneuvers—before the final dive. A second pilot not involved in the session but who reviewed the scenario’s details said it highlighted several training opportunities.
“This is the sort of simulator experience airline crews need to gain an understanding of how runaway trim can make the aircraft very difficult to control, and how important it is to rehearse use of manual trim inputs,” this pilot said.
While Boeing’s runaway stabilizer checklist does not specify it, the second pilot recommended a maximum thrust of 75% N1 and a 4 deg. nose-up pitch to keep airspeed under control.
Boeing is developing modifications to MCAS, as well as additional training. Simulator sessions are expected to be integrated into recurrent training, and may be required by some regulators, and opted for by some airlines, before pilots are cleared to fly MAXs again. The MAX fleet has been grounded since mid-March, a direct result of the two accidents."