Lion air, B-737 and "AI" again.

[Anonymous]

I saw Lion Air being discussed earlier on this forum so there's probably some on this board that find this relevant

https://spectrum.ieee.org/aerospace/aviation/how-the-boeing-737-max-disaster-looks-to-a-software-developer

Now that all reports are out, and there's more information.

It's a long read. to summarise he says that

1. they put bigger engines on the plane to reduce fuel usage.

2. these engines affected the stability.

3. it was too expensive to fix the airframe (re-certification and re-training required) so they made a software patch to stabilise the plane.

4. there's only one sensor for this system and the computer automatically believes it no matter what (a big WTF for me)

5. because of a sensor problem the computer raised the nose and pulled on the stick to alert the pilot. EDIT: After re-reading I don't know which way around it is. I read elsewhere the planes stalled, but here he stays the computer lowers the nose.

6. it pulls so hard on the stick that the pilot can't fight it (an even bigger WTF for me) - should have asked on the intercom "do we have any gym-rats on board?")

7. there should be some way to turn it this system off, but there wasn't, or no-one knew how to do it.

8. because the nose kept going up, the plane slowed to a stall and crashed. EDIT: After re-reading I don't know which way around it is. I read elsewhere the planes stalled, but here he stays the computer lowers the nose.

9. this system would not have passed certification in the old days, but now, like the Dreamliner batteries, the authorities just trust the factory when the factory says it's fine.

 

[Condottiere]

Checks and balances, plus incompetence and parochial interests, something our North Americans cousins should be feeling particularly sensitive about now.

1. They needed a competitive solution to Airbus offerings.

2. All things being and remaining equal, shifting the centre of gravity doesn't turn out well.

3. Like the FCC and FDA, the FAA has been gutted and puppetted.

4. GIGO.

5. Veteran pilots appear to understand the issue and compensate for it.

6. Reportedly, those that did report it were censured.

Looking at it from a software viewpoint, go arounds may cause a virtual crash, and the blue screen of death may appear, not an actual crash, nor onrushing ground.

 

[Hakkonen]

1. they put bigger engines on the plane to reduce fuel usage.

Sorry, run that bit by me again?

 

[Anonymous]

1. they put bigger engines on the plane to reduce fuel usage.

Sorry, run that bit by me again?

In an airliner, the engine is mostly a giant fan that blows air out the back. The actual jet coumbustion component is low. This is the opposite of a fighter plane engine. Accelerating a large amount of air by a small amount is more efficient than accelerating a small amount of air a large amount. So larger fans that turn slower. Plane is slow but saves fuel. Opposite of fighter plane.

In physics, if you have two heat engines of the same power output, the one that burns hotter is more theoretically efficient becuase an engine that burns hydrocarbon works off the heat difference between the components. I'm not completely sure about the engineering of the engine but I suspect being larger allows it get hotter before it breaks (thicker parts maybe); this part about size vs heat could be wrong as I'm not an jet engineer, but the part about higher heat giving more efficiency is well understood physics.

 

[AndrewW]

4. there's only one sensor for this system and the computer automatically believes it no matter what (a big WTF for me)

Agreed, ran across this myself:

It is astounding that no one who wrote the MCAS software for the 737 Max seems even to have raised the possibility of using multiple inputs, including the opposite angle-of-attack sensor, in the computer's determination of an impending stall. As a lifetime member of the software development fraternity, I don't know what toxic combination of inexperience, hubris, or lack of cultural understanding led to this mistake. But I do know that it's indicative of a much deeper problem. The people who wrote the code for the original MCAS system were obviously terribly far out of their league and did not know it.

So Boeing produced a dynamically unstable airframe, the 737 Max. That is big strike No. 1. Boeing then tried to mask the 737's dynamic instability with a software system. Big strike No. 2. Finally, the software relied on systems known for their propensity to fail (angle-of-attack indicators) and did not appear to include even rudimentary provisions to cross-check the outputs of the angle-of-attack sensor against other sensors, or even the other angle-of-attack sensor. Big strike No. 3... None of the above should have passed muster. None of the above should have passed the "OK" pencil of the most junior engineering staff... That's not a big strike. That's a political, social, economic, and technical sin...

The 737 Max saga teaches us not only about the limits of technology and the risks of complexity, it teaches us about our real priorities. Today, safety doesn't come first -- money comes first, and safety's only utility in that regard is in helping to keep the money coming. The problem is getting worse because our devices are increasingly dominated by something that's all too easy to manipulate: software.... I believe the relative ease -- not to mention the lack of tangible cost -- of software updates has created a cultural laziness within the software engineering community. Moreover, because more and more of the hardware that we create is monitored and controlled by software, that cultural laziness is now creeping into hardware engineering -- like building airliners. Less thought is now given to getting a design correct and simple up front because it's so easy to fix what you didn't get right later.

 

[Anonymous]

Software standards have dropped, industry-wide.

In the programming industry there is this idea that embedded code in vehicles is somehow quality better because "people die if it fails". After several high profile fails we now realise this isn't as true as it once was, though it's obviously better than a random small web app. Embedded also has a huge security (hacking) problem waiting to happen because they can't easily update and ignored the issue for years.

The ability to provide digital updates and the need to grow startups fast has caused the quality of certain software to plummet.

I don't know what's happening in aviation but I just looked at software engineering jobs for it and it's paying a full $30K lower than the tech sector at entry level. They have already started a shift away from ADA (a programmming language common in aviation and military) as a result of the difficulty of attracting new graduates (getting kids to learn a niche language not commonly used in tech is hard), and this pay gap can't be helping.

 

[Linwood]

The most disturbing part of this for me personally is an apparently complete miss by Boeing on the functional safety side. Like all commercial aircraft manufacturers Boeing is obligated to adhere to ISO 26262, which mandates a review and update of the aircraft’s safety case for design changes like these. That assessment - if performed properly - should have flagged MCAS as a system requiring additional work. The high frequency of use - every flight! - the inability to cross-check between sensors, the lack of a mandated second indicator to the flight crew (Boeing had an AoA indicator as optional equipment) - all should have stood out as potential issues.

Functional safety is meant to be a check on the concerns Moppy and AndrewW point out. That Boeing failed to flag them in this instance - or worse, put other concerns first - doesn’t speak well for the company or its products right now.

 

[phavoc]

I think there is something to be said for the training and experience of the pilots. In the Lion air crash it was reported that the day before the plane crashed the same exact thing occured by a different flight crew. A 3rd pilot riding in the jump seat told them how to address the issue and the plane didn't crash.

The Korean air flight that crashed in SF a few years ago was because the pilots were using the autopilot to land and did not realize that they were coming in too low. They trusted the autopilot and did not trust their eyes.

Many pilots around the world aren't former military pilots. The US airline industry loves to steal AF trained pilots. I think that the human factor is just as much of an issue as the automation factor. There probably should be a master cut-off switch that disables all automation controls and lets the pilots control everything manually. But there isn't, you have to know exactly what to do quickly to disable things. And that's another thing - Boeing sold this plan as a easy upgrade from existing 737 with almost no re-training required. That was a giant sales factor they used, and now it's coming to bite them in the ass. There's no substitute for training, not even software.

 

[Condottiere]

There's a lot of momentum behind the seven three seven family, it's a cheap short range jet that may make up a quarter of the commercial airliners currently in operation, but is based on nineteen sixties technology.

They needed an intermediate range boost to compete with Airbus, since they aren't ready to invest in a brand new plane development. Problems might lie deeper than a stingy board of directors, as complaints from both military and commercial customers of left behind tools and debris like metal shavings have surfaced, which apparently could short circuit electrical systems.

Unless there is a revolution in engine technology, the seven three sevens may be quietly end of lifed.

They'll probably remain the primary short range passenger transport, but challenged by Bombardiers in that role, especially if oil prices climb up again.

It's hard to believe anyone airline outside of the United States will buy a Max again, or any further series of seven three sevens.

If I had to speculate, Boeing will buy Embraer.

 

[Condottiere]

US regulators allowed Boeing to fly 737 MAX
by NICK FARRELL on12 DECEMBER 2019

Even though they knew one would crash every two or three years

US regulators decided to allow Boeing's 737 MAX jet to keep flying after its first fatal crash last fall even after their own analysis indicated it could become one of the most accident-prone airliners in decades without design changes.

According to the Wall Street Journal, the November 2018 internal Federal Aviation Administration analysis, released during a House committee hearing reveals that without agency intervention, the MAX could have averaged one fatal crash about every two or three years.

That amounts to a substantially greater safety risk than either Boeing or the agency indicated publicly at the time. The assessment, which came the month after a Lion Air crash in Indonesia, raises new questions about the FAA's decision-making in the wake of that disaster, along with what turned out to be faulty agency assumptions on ways to alleviate hazards.

In the wake of the analysis, the FAA took steps to put short-term and permanent measures in place to combat hazards, but Wednesday's hearing started off with challenges to some of those decisions.

Rep. Peter DeFazio chairman of the House Transportation Committee said that despite its own calculations, "the FAA rolled the dice on the safety of the travelling public and let the 737 MAX continue to fly."

https://www.fudzilla.com/news/automotive/49966-us-regulators-allowed-boeing-to-fly-737-max


Die-rolling R Us.

 

[Anonymous]

Why not make the builders fly in it? Seems like a simple rule that will improve the plane safety without a whole load of work on the regulator's part.

 

[Condottiere]

There's an anecdote that Stalin had his tank designers climb into their prototypes, and he'd machine gun it.

 

[phavoc]

What I don't see why Boeing is balking on getting the 737 back in the air is to integrate additional sensors into the MCAS system. They actually offer them as additional add-ons if the customer is willing to pay. They should just integrate them for free. The amount of money they are losing would be a lot less than the cost of the sensor (not to mention that they should have AT LEAST two sensors to feed MCAS).

 

[Anonymous]

There's an anecdote that Stalin had his tank designers climb into their prototypes, and he'd machine gun it.

I'd eat anything I cooked, because I probably wouldn't have taken a job I felt I couldn't deliver.

 

[Condottiere]

1. Group think.

2. Hoping technology will catch up.

a. Volkswagon diesel engines.

b. Nasrudin was caught in the act and sentenced to die. Hauled up before the king, he was asked by the Royal Presence: "Is there any reason at all why I shouldn't have your head off right now?" To which he replied: "Oh, King, live forever! Know that I, the mullah Nasrudin, am the greatest teacher in your kingdom, and it would surely be a waste to kill such a great teacher. So skilled am I that I could even teach your favorite horse to sing, given a year to work on it." The king was amused, and said: "Very well then, you move into the stable immediately, and if the horse isn't singing a year from now, we'll think of something interesting to do with you."

As he was returning to his cell to pick up his spare rags, his cellmate remonstrated with him: "Now that was really stupid. You know you can't teach that horse to sing, no matter how long you try." Nasrudin's response: "Not at all. I have a year now that I didn't have before. And a lot of things can happen in a year. The king might die. The horse might die. I might die.

"And, who knows? Maybe the horse will sing."

3. I eventually learned how to cook without food poisoning myself.

4. Dictators tend to shoot the messenger.

 

[Linwood]

What I don't see why Boeing is balking on getting the 737 back in the air is to integrate additional sensors into the MCAS system. They actually offer them as additional add-ons if the customer is willing to pay. They should just integrate them for free. The amount of money they are losing would be a lot less than the cost of the sensor (not to mention that they should have AT LEAST two sensors to feed MCAS).

The most likely reason is that this has forced them to review everything about the 737 Max. It’s possible other potential deficiencies have turned up - either elements of the controls architecture that weren’t adequately verified or actual components that don’t work as they should. It could also be that they’ve tried a fix and found it created new issues that they’re now trying to address.

Or the weight of the paperwork created to recertify the bird doesn’t equal the weight of the aircraft yet....

 

[Anonymous]

They actually did hook up other sensors. One of the original problems was Mcas was connected to only 1 of a pair of sensors. This is now resolved.

 

[Condottiere]

Probably cheaper (in relative terms) and faster to try and recertify the MAX.

If Boeing gave up now, they have nothing to replace it, and I hear there's now an overflow at the employees' parking lot.

If they can't deliver them, they have to be junked; wonder if insurance covers that.

 

[phavoc]

They won't junk the aircraft. They will be sold, though the bottom line will feel it.

 

[Linwood]

Boeing just announced they’re slowing production of the Max in the new year, presumably due to the delays in recertification.