Well, yes, probably so

Ethiopian Airlines Boeing 737 pilots ‘could not stop nosedive’

Given the consequences of not stopping it we can be pretty sure they tried, given the consequence we can be pretty sure they could not.

37 thoughts on “Well, yes, probably so”

  1. It seems the line is that the computer thinks the wings have stalled and tries to correct it buy going into a dive that the pilots cannot override.

    Now, if I wasn’t a right-on progressive ‘liberal’ I might mention the elephant in the room that of all the 737 maxs in the world this problem seems to occur only in shithole 3rd world countries and question the competence of the flight crew or maintenance teams. But I’d never do that.

  2. If a pilot wants to over-ride a system because its going to cause a crash then god help the company who doesn’t allow them to do that.

    OTOH if the system wants to overide a pilot because they’re going to cause a crash then god help the company who doesn’t allow it to do that.

  3. There is a question, I think, over whether they knew how to disable MCAS.

    If I understand it correctly (I might not), pilots who are used to other 737s don’t need to undergo so much training when they move to a 737 MAX 8 because it is still a 737, and as a result they did not all know about MCAS, because it’s a system designed to make a MAX 8 fly like any other 737 despite the different physical characteristics.

  4. @Dongguan John

    ‘… all the 737 maxs in the world this problem seems to occur only in shithole 3rd world countries…’

    As did all the early problems with Airbus planes and computerised systems.

    Odd that.

  5. ‘… all the 737 maxs in the world this problem seems to occur only in shithole 3rd world countries…’

    As did all the early problems with Airbus planes and computerised systems.

    Odd that.

    Shit hole 3rd world countries, Like Spain / Or Australia

  6. Reading the news article it appears that not only did they try, but they tried repeatedly, and each time they tried the followed the instructions from the manufacturer on how to do it.

    That is they tried the right and recommended way and it still didn’t work. It wasn’t that they tried but did the wrong thing (push up instead of down).

  7. There have been 737-MAX incidents in the US but the pilots diagnosed the problem and successfully took avoiding action. That probably says quite a lot about relative airmanship levels.

  8. Dongguan John, on a business trip to Lusaka in the early 90s we stayed at the same hotel as the Italian maintenance crew for Air Zambia. They were all rat-arsed in the hotel bar every night. Was absolutely bricking it all the way back to Johannesburg on our Air Zambia (DC10) flight especially with Action Jackson at the controls.

  9. My understanding is that the only sure way to stop erroneous MCAS inputs is to “pull” the relevant circuit breakers. Simply resetting it won’t stop it trying another nose down input if the sensor is still giving the wrong information. I’m sure I read that another Max 8 (might even have been the same airline?) suffered similar problems only the day before, but an extra company pilot – who was hitching a ride in the jump – seat knew about this, and was able to help the flight crew who didn’t…

  10. What puzzles me is the autopilot should be getting both altitude & rate of decent information & will be aware of the pitch angle & airspeed. Putting those 4 together you can easily work out:
    1) the plane isn’t in a stall configuration
    2)diving the plane is going to intersect with the geography
    It is, after all, what a pilot does if he can’t be bothered looking out the window. You can, in fact, land a plane safely with that information. Although it does help if there’s a runway conveniently underneath you when you do.

  11. I’m surprised that planes (especially hundreds of millions of dollars worth of aircraft like a large airliner) don’t have a GPS system talking to the control system.
    Most phones have GPS these days, accurate to within 10m or less.
    Surely there should be some check going on constantly – system input say we’re in a terminal stall, GPS says that’s unlikely – result = alert pilot to system problem and hand control back to them (after all that’s what they’re paid for…)

    Why isn’t something like that done?

  12. Why isn’t something like that done?

    To my very limited knowledge, the Boeing beancounters were adamant that no new pilot facing systems were installed on the new models so that pilots wouldn’t have to undergo type conversion training; and also that the automated systems that were supposed to make it fly like the old ones meant the type could still be certified as an update rather than a whole new model.

  13. If you want more information to add to your speculation, there’s quite a bit here (https://arstechnica.com/information-technology/2019/04/ethiopian-air-pilots-turned-off-737-max-anti-stall-system-then-it-turned-on-again/).
    Though half of the information is in the comments…

    As I understand it, you can turn off the system, but then you have to manually turn the trim wheel, which is very difficult as you’re fighting against the wind pressure on the flight surfaces. And you can’t get power back on to the trim wheel without also turning on the MCAS system, which will once again try to kill you.

    Horrible, horrible situation to be in, and sorry DJ, but this time the shithole appears to be Boeing.

  14. Amazing how people can draw the conclusions that fit their personal prejudice based on N=2. I guess the air France A340 pilots who seemed to think pushing the nose up, arcade game style, is the way to gain altitude, were also dogs. Stands to reason, innit?

  15. BiG. I live in Asia and travel all over the place. The levels of fuckwittery are astounding.

    Now I’m no pilot but given the levels of incompetents I see in my industry I wouldn’t be at all surprised if a competent pilot would have dealt with it with a flick of a switch.

    The shit I see everyday….

  16. As best we know, the Aircraft suffered damage on take-off (not sure how) to the critical AOA sensor (Angle of Attack) such that the system was receiving incorrect data. That is what they think happened to the Lion Air aircraft off Indonesia, a faulty sensor. The damage could have been a bird strike or something like that. Lion Air the sensor was faulty, “fixed” overnight by maintenance, but not properly fixed so it promptly failed again not long after take off. That might be more criticised as poor maintenance.

    The MCAS system works fine – with correct data. The issue may be that Boeing uses a single AOA sensor and if that is damaged then MCAS can command faulty actions like nose down when 450 ft up at full power and desperately trying to climb.

    BenS has also read the same sort of info and I believe that is correct, the pilots did in fact turn MCAS off correctly but the relatively high speed that the aircraft was travelling at meant that the pilots could not correct the faulty trim manually, at least not quickly enough. So apparently the turned the power back on, got control authority but then had the MCAS system cut in again and trim the nose down before they could prevent it. They had a small chance to switch power on, rapidly trim back then cut power again, but sadly couldn’t make it in time.

    Boeing does appear to be at fault in that the MCAS is an extended version of what already exists on earlier 737 models but doesn’t have an adequate fail-mode whereby the pilots can manually take control in the event of a sensor malfunction.

  17. Another (veteran) pilot perspective

    What seems abundantly clear is that a highly anomalous instrument reading was used as a control input to flight systems with little or no warning to the pilots and that the morass of automation went some way to concealing this fact – shortcomings in the control architecture contributed….

    The IRS inertial system should have been used to spot the insane reading from the AOA …..

    The AF447 crash is related in some ways…. when things go pear shaped in the air you don’t usually have much time to diagnose and batshit instuments should be I feel flagged at an earler stage….

  18. I was a computer scientist for 30 years.

    I described myself as a “systems archeologist,” trying to figure out what the heck people were thinking when they wrote the code.

    No one’s life depended on our systems, thank God!

    Computerizing control is difficult, beyond the capabilities of most humans to comprehend what can go wrong. Even some engineers. I worked with many engineers. Some were brilliant, some were dumb, “How did he ever get his degree?”

    What’s the answer? I believe Boeing and Airbus will work it out. Trial and error. Eventual comprehension. There will be bodies.

    Related: automation of automobile driving. IF it happens, it will follow the same path. Will probably work to some extent, some day. There will be bodies.

  19. Computerising control systems should be a done deal these days. Fighter aircraft have been designed to be aerodynamically unstable for a long time and are stabilised by computer control systems. I’ve seen a demo of a computer control stabilising not one stick balanced on a finger but two and three on top of each other, and reacting to small pushes of the sticks. The key is the integrity of the sensors measuring rod included angle at each joint. That was Boeing’s problem and they got the solution wrong.

  20. During WW II, Penn (Towne
    School) published “Theory of
    High-Speed Stall”.
    Could the Boeing problem be

  21. “Computerising control systems should be a done deal these days.”

    A hasty generalization.

    “Fighter aircraft have been designed to be aerodynamically unstable for a long time and are stabilised by computer control systems.”

    Which has double ought zero to do with controlling an air liner.

  22. override aft column cutout
    not fail-safe with single sensor malfunction (ssm)
    unlimited authority from ssm.
    trim without regard to resulting severe mistrim
    remove switch to cutout MCAS runaway
    force manual trim w/ssm
    No guidance: manual trim lockup
    MCAS not in FCOM

    Boeing are fucked – a situation predicted by many long serving Boeing folk who voted with their feet and pensions after the 1997 merger with Douglas …..

    Boeing’s liability is simply enormous

  23. @tomo
    The AF447 crash is related in some ways…. when things go pear shaped in the air you don’t usually have much time to diagnose and batshit instruments should be I feel flagged at an earlier stage….

    It can be a huge problem if you’re still taking off, close to the ground, and not far above actual stall speed. But in the Air France case, they were safely in the cruise at 37,000 feet. If the two First Officers had just selected a sensible thrust level and attitude, and then simply kept it there for 5 minutes until they were through the icing conditions that had caused their unreliable airspeed indication, they’d have been fine.

    Instead, the pilot flying firewalled the engines and hauled back on his stick, which is how he’d been taught to handle a stall warning near the ground (the only time he’d probably ever experienced one in the sim), but a really bad idea when you’re only a few thousand feet below ‘coffin corner’. And his colleague had little idea what was happening because (unlike in a Boeing) the joystick gives no direct feedback of what the other pilot is doing.

  24. Chris Miller


    The batshit pitots didn’t help. Aftet making the comment I found this. Misinterpretation of sensor malfunction….

    I was ignorant of the non linking of the sidesticks on an Airbus – is there absolutely no connection or force feedback between the two controls? What happens if control inputs conflict?

    An acquaintance paracuted into electric steering software for trucks informed me that he was amazed that there hadn’t been incidents with the system as the production code (pre-compilation) had so many logical flaws and bugs in all areas.

  25. @tomo April 5, 2019 at 11:31 am

    when things go pear shaped in the air you don’t usually have much time to diagnose

    ~40 seconds on 737 Max crash for pilots to understand and override plane’s “secret” self-destruct system.

  26. Pcar


    I worked in Seattle in 2000 / 2001 and met veteran Boeing people who were spitting feathers at the merger with McDonnell-Douglas and the HQ move to Chicago. A couple were so miffed that they handed in their notice and took early retirement. There seemed a unanimous opinion that the culture coming in would inevitably lead to some calamitous engineering decisions from the changes in emphasis being made at that time.

    I don’t know if Boeing had a prior blemish free record and that the MD culture was that corrosive – but there are undoubtedly some retirees in Washington state who will be repeating a “told you” mantra at the moment.

    This could be Boeing’s DC10 moment.

    Paul Lewis will be scampering around trying to PR the hell out of all this. 5000 arcraft …..

  27. @tomo (NB I am not a pilot :))

    AIUI there’s no linkage between the controls, but it’s possible to ‘see’ the inputs from the pilot’s side on one of the screens. There’s a ‘take control’ button on the Captain’s sidestick, but otherwise it’s the arithmetical sun of the two inputs (although only one pilot at a time is meant to be in control).

    On any Boeing, the fact that the joystick was pushing into your lap would be a strong indication that your fellow pilot had ‘lost it’.

  28. Chris Miller

    I’m not a pilot either but I have seen the joystick innards – not the operational implementation – I had assumed (yeah, that old one) that the sidesticks mimic-ed the mechanical predecessor = wrong it seems…

    That said – “iced up pitots!” should be the first warning ….. not STALL!, STALL!, STALL!

  29. Pcar

    blancolirio (777 pilot) on youtube does a better job I feel of portraying the workload in the cockpit at takeoff.

    The left AOA instument was clearly insane and should’ve clearly been dropped out of the control logic – it had the aircraft AOA at 74.5 degrees and was flapping around

    Beyond the fact that the batshit instrument had unlimited control authority the failure of a brand new instrument raises quality control issues.

    I know from direct personal avionics experience that the production bureaucracy and production deadlines can overide a non-complying part – the primary reason I quit avionics some 30 years back…. At the moment few seem to be asking about the failure rate on the AOA instrument….. There have been claims that the Lion Air unit that crashed had a 20 degree bust between AOA indicators – imho that should’ve grounded the aircraft until resolved – there’s more to the AOA sensor story I suspect.

  30. I guess it wouldn’t be that difficult to make the ‘passive’ sidestick emulate the movements of the ‘active’ one, but the point is that the non-flying (monitoring) pilot would not normally have their hand on the sidestick, so there’s little point in doing do.

    As for AF447, handling unreliable airspeed indication should be taught on about day 3 of initial training.

  31. One might have thought that the position of the controls in a dual control aircraft might be part of the “at a glance” situational awareness thing. The way Airbus do it must have been thoroughly worked out and tens of thousands of pilots seem happy enough with it…

    The failing sensors were iirc the subject of an upgrade program – sensing the onset of pitot icing / ramping up the heaters being the issue addressed – again iirc.

    Relying on instruments mandates reliable instruments.

Leave a Reply

Your email address will not be published. Required fields are marked *