On the treadmill just now I came up with a nice “thought experiment” to help explain better why the crashes were such an outrage and also why the planes can still be among the safest airliners in the sky. Give me a couple of hours…
As a non pilot, but a person interested in technology and its failure, I am very interested in the cause and outcome of this. I believe these statements made for example by you, smoke. Or Juan Browne (blancolirio) on YouTube.
The interesting thing is: How will stoopid hoomans deal with this massive fail, and how will its aftermath look like. Will we learn more than just correct a plane model. Pretty sure we will, but to what extend? Prices going up is one indicator I‘ll be watching.
Money is the only thing.
Boeing designed the MAX instead of a clean-sheet because Airbus was getting 32xneo sales that were also not clean-sheet. It had customers saying “we need something NOW.”
Unfortunately, to get the plane to have the efficiencies it needed and be cost-effective meant the plane wouldn’t fly exactly the same. They could have probably designed it that way eventually, but it would’ve taken longer and/or cost more. Neither was acceptable.
However, making the plane fly different was also not acceptable. The airlines didn’t want to have to train their 737NG pilots like it was another plane, they wanted minor changes only.
So instead automation was introduced to make it fly the same (or almost) so the airlines could save time and money on training. We don’t want them off in training, we want them flying! We don’t want to pay for that extra training both coming and going!
The problem is, naturally, pilots are not engineers and engineers are not pilots. Both make assumptions the other does not. The engineer that assumed a single AOA sensor was good enough was crazy. Then you had the marketing guys saying that critical flight indicators that could assist the pilots in diagnosing a serious issue should be an add-on for just a few hundred thousand more, per plane! For a light! Guess what? Management says no thanks, our fleet of a hundred would cost too much, we don’t need that.
Why were the pilots not informed? Was that marketing? Management? Airline management? Who knows. But they weren’t because everyone who did know assumed it was a nonissue.
Then the airlines where the pilots and maintainers get less training because of the costs naturally ran into it first. I’m sure if they hadn’t grounded them at some point pilots in the better funded airlines would’ve run into it as well.
After both Boeing and the airlines were so concerned over saving money, both have now lost money, some more than others. Textbook example of penny wise, pound foolish.
I think this is a bit simplistic. Yes, Boeing obviously assumed that pilots would just treat this as a runaway trim situation, which is a pretty benign affair, and follow their training / checklists to sort the problem. However I think it has become increasingly clear that this was a mistake, and they failed to account for MCAS being more powerful and the failure not presenting itself in the same way, making correctly reacting to the problem more difficult. There was also no slack in the system to allow for human factors.
Training and maintenance are of course pertinent issues, but given that the pilots and airlines were not made aware of the existence of MCAS and that even the best quality parts can fail it seems to me that even those airlines with the best training / maintenance programmes were at risk.
As for the automation vs human debate, this is less clear cut. An objective assessment would lead to the conclusion that you should put the most reliable system in charge, as this would result in the fewest accidents. Historically, this has been a well trained human, as they have generally coped better in an environment that throws up unknown errors that need to be troubleshot (troubleshooted?) and reacted to.
However, we are now living in an age where technological solutions can be created that react more quickly and with better insight than even the best trained human (as an example look at Google’s Alpha Zero AI, which taught itself chess from first principles in only 4 hours, and is unbeatable by any human player). Airbus have demonstrated that it is possible to design a safe and reliable automated system (on par with boeing), and most of their automation is based on 1990’s technology. They are not without their own design issues too, though.
It therefore comes down to three questions:
- Can you supply enough sufficiently well trained humans to meet demand?
- Can you put sufficient care and attention into the design and testing of automation to ensure it performs better and more safely then a human?
- Can the passengers you are transporting psychologically accept that they are travelling in an automated craft and that the computers are in charge?
In my mind I have no doubt that it is possible to design and build a completely automated system that is safer and better than one flown by a well trained human. It is unlikely to be done though because of the effort required for design / testing, and (at least for the foreseeable future) because of the answer to question 3 above.
As always, I am no expert and may well be completely off base with all of this.
I still think that the worst part of it was that the plane got certified. Still mind boggling to me.
I wrote what I hope might be a helpful parable…
My question…coming from a mind of a simple pilot…why didn’t they trim the forces out? The trim system worked just fine. It even disabled MCAS for 5 seconds.
That’s exactly what they tried to do. But in hindsight that was the wrong response. The correct response would have been to treat it as a “Runaway Trim” (which it was not) and lift the cutouts. After that, they could trim manually with the wheels. The motor which MCAS utilized was very fast. So it only took a few seconds before the stick forces were beyond the ability of the yoke to track the nose back away from the drink. Try to imagine a cycle where you get things under control for a moment. You relax a moment then the plane starts trimming wildly again and you once more find yourself along for the ride. Totally terrifying.
I’ve hear opposite. MCAS is slower than normal trim. …
It is now.
Pilots both real and virtual for whom airliners are pretty dull machines have a backwards concept of elevator and trim. Most planes smaller than a midsized corporate jet have elevators that are big and trim tabs that are small. Airliners have movable stabilizers which are huge and elevators which are comparatively tiny. It is those big stabilizers which are moved by the trim. When you move the yoke you are manipulating the the elevator, the minor of the two pitch controls. If you watch an airline pilot land her jet in gusty conditions you might be surprised at how much yoke displacement is required. At low speed, it takes a lot. This makes for excellent controllability through a huge speed and CG envelope. But it also means that any automated system that is given access to the trim had better be very limited in its authority.
Now? Boeing updated MCAS logic after crashes and went through red/black label and FAA approval before fleet got grounded?
I am not sure what that means. But I have read that of the many changes to MCAS, one was to lower the trim rate, the other was to limit the length of each application.
Presumeably these are the changes that boeing is trying to get certified now?
Right. Remember that the crews in both crashed planes were dealing with stick shakers and stall warnings. The cockpit was filled with sources of confusion and sensory overload. This was because a single bad AOA sensor was all it took to activate the warnings and MCAS. Now it takes both sensors. There might be other logic built into the stall warning system as well to lower the odds of errant activation even further. MCAS now supposedly is so weak that, even if it does activate, it can easily be managed without the cutouts. I think I read also that MCAS will now honor the yoke displacement sensor. If the yoke moves a certain distance of travel opposite the trim direction, power to the trim motors is removed.
That sounds much, much better. An assist shouldn’t obtain primary control.
To make an analogy to a more common system most of us deal with - power steering, or in my case electric power steering. While driving it varies the assist to make turning easier or more stable depending on speed. That functionality makes driving easier and safer - but I don’t want it suddenly trying to steer on my behalf (without going into autonomous cars).
I had read the second AoA sensor was an option, so will all MAX’s be retrofitted with the second sensor if they didn’t have it already?
My Honda Pilot has 2 features - lane keep assist and lane departure warning.
The LKA works when cruise control is on (if you activate it) and will try and keep the car between the lines on the road. It will nudge the wheel left or right to do so. It’s not great if the lines are faded, the ground isn’t clean, construction has made the lines confusing, etc. When that happens, it can try and steer you left when you need to go right or vice versa. I don’t use it because it gets confused too much as the real world roads are rarely as neat as the program assumes.
LDW doesn’t try and steer. If you’re too close to the left or right side of the road, it just starts shaking the steering wheel, presumably because it thinks you’re dozing off or not paying attention.The problem is the wheel shake is far too severe, and on occasion it has almost pulled me over the line it was alerting me about! Instead of the rumble you can feel in a FFB wheel while racing Project Cars or Dirt Rally, it’s just a left-right-left-right-left-right action of the wheel that is almost too strong to overpower. So I turned that off too.
Imo all the driver assist functions in cars these days just makes everyone a poorer driver.
A few weeks ago I was in the crewroom printing out the paperwork for a flight. A few feet away a Captain was singing the praises about his new Tesla to his crew. His drive to Newark from Long Island is 2-3 hours in traffic. The new car lets him read, plan the flight, catch a little Netflix. The Tesla “autopilot” requires that the steering wheel is able to sense the driver or it will sound a warning (or maybe vibrate) after a period of time. To het around that, he hangs a weight on the column. I personally can’t believe that anyone would trust the automation to that degree. He then chuckled as he recounted how the car sometimes suddenly brakes for bridge shadows.
“Hey, how funny! I might get rear-ended but at least it won’t be my fault!”
Nature will always come up with a better idiot. What scares me more is that a person with that attitude is a plane captain.
I tried the Tesla autopilot in a friends car… I just couldn’t relax. I felt it steered way to close to the line and I just couldn’t shake the feeling that it could turn the wheel over before I could react. I can’t understand how people can show such blatant disregard of safety that they won’t even hold on to the steering wheel.
Especially when it’s still in development.