Boeing is imposing yet more losses on carriers that were hapless enough to have purchased its 737 Max.
Early on Sunday, United Airlines and American Airlines announced they were keeping Boeing 737 Max aircraft out of their flight schedules through early November. That’s two months later than their most recent plans.
The Wall Street Journal swung into action and got more detail. Its take? That the scandal-plagued model may not fly until 2020. That would hit airlines that own the plane with lower capacity during the big volume holiday season. From the Journal:
Boeing Co.’s 737 MAX planes are unlikely to be ready to carry passengers again until 2020 because of the time it will take to fix flight-control software and complete other steps, an increasing number of government and industry officials say….
The situation remains fluid, no firm timeline has been established and Boeing still has to satisfy U.S. regulators that it has answered all outstanding safety questions. But under the latest scenario, the global MAX fleet is now anticipated to return to the air in January 2020, a full 12 months after the plane maker proposed its initial replacement of software eventually implicated in a pair of fatal crashes—one in October and one in March—according to some Federal Aviation Administration officials and pilot-union leaders….
Boeing executives, FAA engineers and international aviation regulators have steadily expanded their safety analyses to cover a growing list of issues spanning everything from emergency recovery procedures to potentially suspect electronic components. Some of those assessments are further complicated because they also cover earlier 737 models.
The article acknowledges that Boeing and the FAA are shooting for an earlier date but increasingly recognize that January is more realistic. And given all the missed deadlines, that supposedly conservative target may again prove to be optimistic.
The hurdles the 737 Max has to surmount include:
Making the MCAS correction to a potential stall less aggressive.
Changing the inputs to MCAS so it looks at both angle of attack sensors
Giving pilots a warning when the angle of attack sensors disagree (this was an upcharge as an extra safety feature; one assumes it becomes standard)
The Journal rather blandly described two issues that are known to be holding up giving the 737 Max an FAA green light. As we’ll discuss shortly, they could be difficult to remedy:
The topics included concerns about whether the average pilot has enough physical strength to manually crank a flight-control wheel in extreme emergencies.
In late June, Boeing and the FAA disclosed still another flight-control problem on the MAX, involving failure of a microprocessor that meant test pilots couldn’t counteract a potential misfire of MCAS as quickly as required.
Moon of Alabama provided very informative posts on each issue when they were first reported in the press. The first is that the manual trim wheel may require too much force under certain circumstances for some pilots to be able to operate it. This is a problem not just for the 737 Max but also for NG. From Moon of Alabama:
In the NG series a new Flight Management Computer (FMC) was added to the plane. …The lengthy FMCs did not fit on the original central pedestal. The trim wheels on each side, used to manually trim the airplane in its longitudinal axis or pitch, were in the way. Boeing’s ‘solution’ to the problem was to make the manual trim wheels smaller.
Smaller trim wheels = more pilot force required. Back to the article:
But the problems described above are 737 NG problems. The 380 or so existing 737 MAX are currently grounded. But some 7,000 737 NG fly about every day. The record provides that it is a relatively safe airplane. But a runaway stabilizer is a well known electrical malfunction that could by chance happen on any of those flights.
The changes from the 737 Classic to the 737 NG make it more difficult, if not impossible, for the pilots to recover from such a situation:
- The smaller manual trim wheels on the 737 NG make it more difficult to trim a runaway stabilizer back into a regular position.
- The larger stabilizer surface makes it more difficult to counter a runaway stabilizer by using the elevator which was kept at the same size.
- 737 NG pilots no longer learn the rollercoaster maneuver that is now the only way to recover from a severe mistrim.
Simulator sessions demonstrate(video) that a runaway stabilizer incident on a 737 NG can no longer be overcome by the procedures that current Boeing manuals describe.
It is pure luck that no NG crash has yet been caused by a runaway stabilizer incident. It is quite astonishing that these issues only now become evident. The 737 NG was certified by the FAA in 1997. Why is the FAA only now looking into this?
Erm, sounds like more training will be required…something Boeing has been very keen to avoid.
Moon of Alabama covered the second problem in June, citing the work of former Boeing flight control engineer Peter Lemme. If his take is correct, it sounds intractable, at least on time frames Boeing and its 737 Max customers would consider to be reasonable.
The trigger was that the “newer, better” MCAS software failed a test in one of its trials, catastrophically, because the chips couldn’t process the input quickly enough (emphasis original):
Meanwhile a new problem that will cause further delays was revealed only yesterday:
…In at least one instance, an F.A.A. pilot was unable to quickly and easily follow Boeing’s emergency procedures to regain control of the plane. The pilot rated that failure as catastrophic, meaning it could lead to the loss of an aircraft midflight, the people said.
…
The issue discovered last week is linked to the data-processing speed of a specific flight control computer chip, according to the two people with knowledge of the matter. In the test, the F.A.A. pilot encountered delaysin executing a crucial step required to stabilize an aircraft.It seems that the additional signal processing and calculations needed for the MCAS fix overload the Flight Control Computer’s (FCC) processor and delay its reaction.
Boeing has been developing a software update for the Max for eight months, [a Boeing spokesman] said. It is unclear whether the new flaw can be resolved by reprogramming the software or requires a hardware fix, which would be costlier and could take much longer.
The interpretation by the author, who coded in assembly language on the very type of chip in question back in the day:
The processors in question are said to be Intel 80286 type CPUs. The original Intel version of that CPU, sold between 1982 and 1991, had a maximum clockrate of 4, 6 or 8 MHz. It was later manufactured by a number of other firms, including by AMD and aeronautics company Harris, with a clockrate of 20 and 25 MHz. It is likely that the Boeing 737 FCC uses these or similar types.
These old processors are very reliable and error free. But they have less than 1/1000nds of the computing capacity of a modern cell phone. According to Lemme one CPU in the flight computer runs up to 11 different processes. All need to receive the input of external sensors, run through their algorithms, and signal a command to the relevant actuators that control the moveable flight surfaces of the plane. That the FAA pilot “encountered delays in executing a crucial step” caused by the computer points to a capacity overload….
Programs written for flight control purposes are already highly optimized. To further optimize them ‘by hand’ would break the regulated process that production of such software requires.
Boeing says that it can again fix the software to avoid the problem the FAA just found. It is doubtful that this will be possible. The software load is already right at the border, if not above the physical capabilities of the current flight control computers. The optimization potential of the software is likely minimal.
Readers discussed this problem at length in comments when the Moon of Alabama post on the chip problem first ran. If his take is correct, this is a nasty problem. The chips are so old that any software on them has to be highly optimized, so there’s almost certain to be no headroom to ask it to do more. And for efficiency sake, the code is written in a low-level language, and also “optimized” for the physical characteristics of that chip. That means it can’t be ported to a newer chip (if it was not a horrifically big task to move to more advanced chips, Boeing would have done so a while back) without starting from close to zero. This means not just heavy-duty development and testing, but new certification….which is what Boeing has been desperate to avoid.
The Journal, at a 50,000 foot level, seemed to be on the same page:
Since the 737 MAX and its earlier version, called the 737 NG, share the same flight-control computer, fixes related to the microprocessor also apply to NG models, thousands of which remain in service around the world. Boeing also faces the task of convincing the FAA that a software fix, instead of physically replacing the suspect electronic component on all MAX planes, will suffice.
Now it may be that things are not as dire as the foregoing suggests. The press is full of reassuring noises that the 737 Max will be flying soon….just not as soon as hoped.
But at this point, Boeing deservedly has no credibility. At every turn, CEO Dennis Muilenburg has understated the severity of the problems with the 737 Max. Boeing also kept its customers in the dark about key issues, like the famed “AOA disagree” alert and gave them inconsistent information. In keeping, in the Wall Street Journal’s comment section, a few readers maintained they’d never fly the 737 Max, and while that sounds like a difficult vow to honor, one said he’d stick to Delta which hasn’t bought the plane.
And the longer this fiasco drags on, the greater the damage to the airlines that own the plane and to Boeing. That famed order of 200+ planes at the Paris Air Show is likely to go poof or be cut way back if 737 Max approval drags on too long. And we also have the non-trivial issue that foreign regulators will be more hard-nosed than the FAA.
If the latest delay doesn’t lead more investors to successfully pressure the board to have Muilenburg’s head, one will have to wonder if he has Jeffrey Epstein level dirt on some directors. There’s no good explanation for his continued survival.
64 comments
Comments are closed.
Thought this was an interesting article and comments, covering the impact on a small airline like Icelandair. Also mentions storage which I haven’t seen mentioned before.
https://simpleflying.com/icelandair-737max-dilemma/
Re: Intel 80286 coding
It’s not just the (by modern computing standards) ancient technology levels of performance from this old chip — but comprehensively debugged and robust — design. Apart from speed of execution, another reason why the software is written in assembler is that it compiles to (comparatively) tiny executables. It’s been about a hundred years (well… 25 maybe) since I did any coding but I remember that a program written in assembly language that was fairly sophisticated had a 10 kilobyte (yes, kilobyte) file size.
By way of comparison, this Naked Capitalism web page (which is just a web page) is going to hit between 500kb and 1Mb. That’s not the fault of the site, it’s merely that all the endless bloat of runtime scripting, style sheets, component libraries (which have to be loaded, even if you only need one function from them) which get called are just now boiler plate and what started out as helpful convenience is unshiftable coding dross that you can’t get rid of.
Why does this matter? It’s not just available CPU cycles which need to be conserved in an old processor like the 80286. It has limited address space for memory and storage. If I recall correctly, that chip had 16bit memory addressing which gives an upper limit of 16Mb memory, which seemed ridiculously huge in 1990 but seems ridiculously tight today. Storage probably tops out at 512Mb (depending on how file allocation was done, but this was a common limit at the time).
One suspects that shoehorning in any more subroutines and, worse, whole new control applications will be hitting fundamental constraints inherent to that chip design. These simply cannot be overcome by more code optimisation.
Moving to a, say, 80386 or 80486 chipset-based flight control system will take years and years of testing.
I tried to dig around, and this tells me that the newest version of the FMC is actually using Motorola 68040.. Which should be way more capable than 286 (it’s 8 years younger than 286, which in HW space is a lightyears).
Just to make clear – it’s still a 30 years old CPU, just not 40, and Motorolas were (at that time) better than x86 which didn’t give up the backwards compatibility for ages, costing them ease of programming and performance.
But clearly even those will have a performance ceiling, and TBH, it may be harder to find a Motorola assembly programmer than Intel.
Yes, I agree, getting Assembly Language skills is always hard, this Motorola dependence would make it a niche within a niche. Adding avionics, certification requirements and particular testing demands means you’ve got a niche within a niche within a niche. I’ll bet Boeing outsources everything and anything in their software development (“it’s just a case of dragging someone with that skill set off the street, how hard can that be..?”). We’ve certainly seen that movie before.
But doesnt this simply point to really bad investment in R&D. If Boeing knew that these chips have reached limits and making up for their limits is requiring them to compromise here, there and everywhere at the cost of increasing complexity and ultimately lives, then might as well assign R&D boffins to come up with a new chip. Certainly it cant be that difficult considering how much money Boeing makes and can command at the capital markets, or its influence in the US defense industry? Even NASA uses C and recently invested in using c++ on mars rover – https://www.youtube.com/watch?v=3SdSKZFoUa8
Well, the fact that C++ turned 40 this year (well, the original idea. C++ as such was first mentioned in 1983) might have persuaded NASA it’s a mature technology ;).
C is “only” 7 years older (or 11, if you take it from that 1983 date).
Joking aside, 386 or 486 could easily run any code written for 286. Even current Pentiums support native x86 code in the 32bit mode.
The real problem is though that you’d have to redo the PCB, and interaction will all the HW kit. So basically, rip out the whole FCC and redo from scratch.
Not even FCA would just nod that through.
And almost certainly you’d still have to rewrite your code anyways, so another expense (in dev and testing).
In fact, even “just” building a faster perfect 286 equivalent may not be simple, because if the code is really optimised, it can rely on things like execution timings (compared to peripherals for example), so speeding it up could break it horribly in unpredictable ways.
Ain’t no easy or cheap way here.
Arguably, the best thing here is to build the electronics in a modular way, but it’s easy to say for a system that existed for a long time – but very hard to implement.
IIRC, Airbus could have an advantage here, as it has all systems in triplicates with different providers, so persumably it’s more modular.
Plus people are noticing that this triplicate seems useful in keeping air craft returning to earth earlier than scheduled.
And no matter what the fixes turn out to be, this perceived advantage will remain.
I used to work in safe sofware development for transport.
One of the reasons it is difficult to go past the 68000 or the early Intel, is that the new CPU have become highly complex and also have secret designs.
When you knew how a processor was built, you could prove some failure modes were impossible. Now, with secret multicores and predictive forking, there are no safe ways to predict what could happen at worst.
The Intel designs, I suspect, were not really made for real time. And the complexity is a good point as well.
Perhaps an older or simplified ARM design?
Ditch the speculative execution and anything like hyperthreading, and go with a fairly simple 8 core 32 bit processor at a reasonable clock rate.
Slower than current general purpose ARMs but a lot more predictable, and a few hundred times faster than 286… and you can spread out the processes among cores to reduce context switching.
Issues with stuff like cosmic rays and other stresses can be dealt with by staying with an older scale and redundancy, and maybe a more robust transistor technology or material.
Relatively easy to send the files to a fab and get more printed up if you need them, a lot more cache, and more RAM.
Stick a good open source RTOS on it – there are a number to choose from, and then do a clean sheet implementation based on functional specs, written in a high level (well C or C-like) language.
That should be a lot more ‘future proof’ than what they have now.
yea but do you guys really think Boeing couldnt have solved these problems for themselves if they put enough money in R&D?
You are mixing up the flight management computer (FMC) and the two flight control computers (FCC).
The FMC is for route planning, load calculations etc. The FCC controls the planes actuators. It also includes the autopilot.
The new FMC was introduced with the NG, the FCCs are older.
Ok, I see, that makes sense. Thx.
Well if the smaller trim wheel and old processor are critical problems shouldn’t the 7000 or so 737NG planes be grounded as well (even though they’ve been flying for many years)?
IMHO yes. The trim wheel issue is absolutely safty critical. The trim wheels are the only alternative when the single electrical system that runs the stabilizer fails.
Interesting story I read about the Apollo 11 LEM descending to the moon – they were early in their descent when they got a 1202 error code from their on-board computer (think one-step beyond abacus). Nobody knew what that was until they found the error code cheat sheet- turned out to be an executive overflow when the chip and RAM have too much info thrown at it and it simply starts to dump what it considers to be the less important info. My guess is that everything they were working with was probably 8 bits at the time with maybe 1 Kb of RAM. My old TI calculator in 1976 probably had more computing power than the LEM.
They were still a go. In the end, Armstrong did the last part of the landing by hand as they were going to land in a boulder field and he had to guide the LEM past it. The various course corrections meant that they were miles away from the projected landing spot so nobody actually knew where Tranquility Base was and Collins couldn’t see it from orbit using a telescope.
The various procurement, engineering testing, and regulatory approvals for things like Apollo, military hardware, and FAA means that there is a lot of old computing technology being used in things that look very new. So your kid playing computer games on the couch probably has vastly more computing power at his disposal than jet fighters and 737 MAX.
Actually, it sort of depends:
https://www.curtisswrightds.com/products/electronic-systems/program-specific-systems/flight-control-computer.html#tabbed-table2
That for example runs on SBC (single-board computers), where you basically plug (and pray).
But of course, tested and tried solutions tend to live for a looong time.
I’m surprised that there is a source for ‘286 chips.
You can still buy a Z80 8bit chip.
There are a ton of older chips still in production that we use in embedded hardware engineering. Just in consumer manufacturing, for instance, in you coffee pot, microwave, washer/dryer, car, &c. Most of my experience runs to military equipment, though, aerospace being part of it.
“component libraries (which have to be loaded, even if you only need one function from them)” No, you don’t have to ship the whole shebang to browsers: that’s what packagers are for (google “tree shaking” for more details if you are curious). Unfortunately because they are yet another rigmarole to put up with in development their use is not as widespread as it should be.
Re: Muilenberg – I’ve seen a few apparently well informed commentators say the reason he is quite safe is that its accepted that most of the mess was inherited, and that he (even before the two accidents) was believed to be trying to shift Boeing back to its days as an engineer-led company. So while his crisis management has been poor, he may be seen as the right man in the longer term to repair the problems.
I’ve commented before that Ryanair (by far the biggest European customer) is one of the most important bellwethers to look out for – so far they are doubling down on their commitment to the MAX. No doubt they will expect to extract a good price from Boeing for their faithfulness. But the company is facing some unexpected troubles, so if the MAX is delayed for longer, they could find themselves in serious trouble. Couldn’t happen to a nicer bunch of people.
PK – On the subject of Ryanair, I totally agree with you. I flew with them once. And promised myself never to fly with them again (a promise I kept). Horrible flying, the scheduling was shocking, customer service almost non-existent, flights to basically nowhere. It looks cheap, but add all the “extras” (including having to get an expensive bus from the middle-of-nowhere airport) and it isn’t worth it. Seriously, it couldn’t happen to a nicer bunch of people!
TBH, I found Ryanair good – as long as you limit your expectations to no customer service in problems and a pushy website. I fly about four times a month, usually with one of Ryanair, EasyJet or BA.
Personal experience below, YMMV.
BA is by far the worst value-for-money. The flight crews are still nice and professional, even though I do detect a bit of resentment of being converted to snack salespeople. But I’ve had so many problems with the “customer service” (with some exceptions), and their web is simply awful UI (buying a ticket leaves you the last page, with no clear way of getting back to homepage or the FF page????).
EJ has its ups and downs, it was awful 7 years ago, reasonable to actually good 3-6 years ago, and is now in the pretty-awfull stage again. I used to sit in the first row, and chat to the flight crew on takeoffs and landings, and interestingly enough, they were usually way more stressed when I saw EJ in its “good” phase and way less so when it was horrendous. Go figure.
Ryanair is ok if you have no expectations, and enough experience to know their systems a bit so you know how to work them.
That “no expectation” excludes an expectation of nil customer service if you have a problem, and extremely pushy website (where it takes 10 clicks to get to payment bit).
Right now, I’ll take Ryanair over either of the two competitors (famous last words, ask me tomorrow midday again..).
I don’t really fly enough to compare, but after a particularly horrible Ryanair experience a few years ago I vowed never to use them again. I use Aer Lingus by default (there isn’t much choice if you live in Dublin) – they’ve gone downhill somewhat, but they are still much more pleasant to use, with far nicer staff. I also feel a little safer with them – I know from people ‘in the industry’ that Aer Lingus pilots are of significantly higher quality than Ryanair. Aer Lingus unions are still quite strong.
I’ve found that so long as you plan carefully, you can find alternatives to Ryanair that usually don’t cost more – indeed, sometimes when you include everything it can be cheaper.
Rule no 1 with RA – never pay for anything except the flight :) [Once I flew all the way to Oz for two weeks with just a cabing bag + a laptop backpack. Smart packing is your friend]
There’s only two alternatives to those three I listed above for me. One flies from Luton (WizAir), so not really and alternative, and the other is SmartWings, more expensive and service comparable (i.e. none), and has much worse departure times.
My first experience with RA was horrible and I said never again – but, when you fly often enough, you find that what really matters is a) how often are they on time (and here the flight I usually take with RA is the first flight on the day, so answer tends to be “pretty often”) b) price. I fly very early in the morning, so have to take a taxi, which is often more expensive than the RA flight. Average RA flight works for me at about 30GBP, but I have paid as little as 10 if I can buy reasonably in advance. At this price, I don’t really expect customer service. Also, the only morning alternative is BA, which is at least twice as expensive, and lands at Heathrow, which is a major pain because of the air congestion (the plane usually spends the last 15-20 minutes in holding pattern before it can land).
https://www.youtube.com/watch?v=l6pj3Fdbwng
50p Airline.
Mary Barra was in a similar position at GM when the ignition switch debacle mushroom clouded. The bean counting executives had baked that cake years earlier.
I don’t see how he’s created that image. See this recent story, which praises him to the sky for doing the reverse, making acquisitions and squeezing the supply chain, both of which are the antithesis to focusing on nuts and bolts:
https://www.bizjournals.com/seattle/news/2019/01/04/boeing-ceo-dennis-muilenburg-the-transformer-named.html
A reader also pointed out that Muilenburg’s efforts on the defense side of Boeing were huge failures. This came via e-mail:
My money is that this bird won’t be ready to return to the air until the other side of next June. There are way too many problems that have stacked up on them that previously were just fobbed off. Last I heard, Boeing is doing everything in its power to be able to say that this plane does not need any more real pilot training nor simulator time. I bet that if they rolled over and accepted the need for extra training for pilots using this plane, that some of their other problems might disappear.
The interesting event will be when it has to be re-certified as being airworthy enough to return to the air. Not just the United States but the rest of the world for those countries that want to operate this plane as well as those countries which will have it overflying it. Boeing will be keen to see it returned to worldwide service as soon as possible so I wonder if Boeing will try to get it certified with the FAA as well as the aviation authorities of other countries at the same time. Previously many aviation authorities accepted the word of the FAA whether a plane was fit to fly but because of the 737 MAX fiasco, they may want to do their own examinations instead.
With the 2020 Presidential election getting closer, China may use their re-authorization of the 737 MAX for leverage with Trump so that is one example of how things could go wrong for Boeing. This must be Boeings’ year of hell.
I don’t have any particular insight into the 737 but it’s certainly possible for a plane to get a bad reputation and sales never reach the point where its a viable product and the manufacturer abandons it. The DC-10 is an example of this; despite being a competent aircraft, after a number of high profile crashes no one wanted to buy or fly it. You still see some of them flying cargo routes.
It’s easy to imagine the 737 MAX being abandoned altogether by Boeing at this point, given the specific problems with the model, it’s history, the cost of keeping the production line going, and the general lack of credibility of Boeing after recent events. I have a hard to imagining it becoming a profitable product that customers, pilots, and passengers are enthusiastic about.
It’s difficult to ignore the analogy with de Havilland Comet, and what subsequently happened to that plane and that company.
Or the Sukhoi Superjet, which was looking good until they flew a demo flight into a mountain that wasn’t on the charts.
I’ll bet that the reason Boeing is fighting the additional pilot training and simulator time is that there are substantial contractual penalties they would have to pay the airlines that bought these lemons^H^H^H^Hplanes. Should the additional training become mandatory in a significant number of markets then the liability to Boeing might run up to a number big enough to leave a mark on all those executives stock options.
Southwest to pick one example has their entire business model dependent on flying one type of plane (737) and should they need to re-train their pilots, the entire company might not be a going concern any longer (hello, bankruptcy court!)
Watching the stock prices of Southwest and United it seems that investors are betting that they can “claw back” their losses from Boeing should the regulators demand significant retraining. Yet Boeing stock while down a bit hasn’t cratered either.
Somebody is sitting on some pretty substantial losses that Mr. Market seems to be ignoring, for now.
Southwest has a contract with Boeing that gives it an extra $1 million rebate per plane should the MAX require extra simulator training,
Thanks … not sure that is big enough to leave a mark on Boeing but good info. Of course other airlines may have similar contracts.
Per the NC dark ages it was discussed the market would have a share price sad if a CEO gets their wings clipped … like losing a demigod ….
Consider the climatic benefits of grounding – less airplanes polluting the skies. ha ha !
On the contrary: older less efficient planes being used for longer.
On the contrary, Tim. Jevon’s Paradox.
My initial thoughts were that this would be a single problem that would be eventually fixed. I mean, unlike the crappy software made by companies like Microsoft, Boeing airplanes are extremely complicated machines that routinely perform flawlessly without ‘crashing’ (literally and figuratively).
But: we hear that Boeing has been outsourcing software development to low-cost contract engineers. Leaving aside the issue of using the dire poverty in places like India to hammer down wages for American workers, it’s not about whether these Indian engineers aren’t smart (they are). The issue is that programming was considered to be a commodity, like buying fill dirt by the ton. If they really lost control of this process, if they didn’t have experienced engineers overseeing the entire integrated system, there could potentially be a lot more potential ‘gotchas’ buried in Boeing’s code. I could see a death spiral, where digging into the code for MCAS uncovers more possible problems, which causes more digging into the code, and more potential problems… if Boeing’s control software has become a layered monstrosity of hack upon hack, this really could be the end of Boeing (at least as a manufacturer of civilian craft: it will still do fine in military contracting).
For the Dreamliner, Boeing followed the standard business school model of outsourcing everything, and lost control of the project: current projections are I think that they will never make a profit on that plane. Now they are following the standard business model of outsourcing the software development – need I say more?
737 MAX and Dreamliner are two different problems.
The 737 MAX problem is that the 737 is a very old airframe but new aircraft certifications take time and money. As a result, there is an immediate short-term profit motive to continue to tweak the existing airframe instead of designing a new modern plane from scratch.
The original 737 design criteria of having it low to the ground so that roll-up stairs could be used meant that the new larger-diameter engines had to be moved forward on the wing so they could be higher off the ground. That wreaked havoc on the center of gravity of the plane and its flight characteristics which is where the software comes in.
As a design engineer myself, the big question that you always have with a design if it starts getting complex is “Do I toss it and start over?” For complex vehicles with lots of regulatory hurdles like a plane, that is a very big question and more often than not the answer from the C-suite is “No. Just make it work”
So the 737 MAX solution was to take an increasingly complex problem and add complex software to save it. At the same time, little things like non-redundant sensor systems slipped into the design and programming. It was also desired to market the plane as not requiring much additional training, so training on the modifications was lacking (along with some notifications of how systems actually worked).
Eventually the Rube Goldberg contraption proved to be too much and a couple of them dove into the ground with pilots not knowing what to do. It seems like the reduced experience and training of the overseas airlines meant that those pilots had less experience and knowledge on how to handle the situation. But as more has come out about the 737 MAX, it was probably just a matter of time before an unfortunate confluence of events would have resulted in a crash in the US or Europe as well.
I agree, in particularly with the last para.
But that in itself is a damning indicement – if the plane is not safe to fly by your average Joe pilot (and the average now includes a lot of fast-track pilots in third world countries, a lot of US pilots tend to have military experience for example), what it’s doing in passenger trade?
It should be possible (and am surprised that no-one dug into it yet) to look up all runaway stabiliser incidents that happened to NG fleet and go through that (I believe all incidents like that have to be logged with the relevant national authorities), to see how many close calls there were.
“Boeing airplanes are extremely complicated machines that routinely perform flawlessly without ‘crashing’ (literally and figuratively).”
You appear to know zero about how these planes are engineered – namely, rather than intended to perform flawlessly, they are engineered with the principle that all kinds of stuff can go wrong, multiple failures, but redundancies and flight crew response will prevent any catastrophic effect.
One of the earliest reliability and failure studies was Abraham Wald figuring out what parts of planes to armor in WW II: https://medium.com/@penguinpress/an-excerpt-from-how-not-to-be-wrong-by-jordan-ellenberg-664e708cfc3d
This is how engineers and mathematicians figure out how to optimize things for reliability. But it is not necessarily how business executives do it…..
You could say that about any airplane. Whether Boeing airplanes are more poorly engineered than Airbus planes (which have had their own automation problems leading to deadly crashes) is an open question. Where Boeing has clearly failed, however, is in the trust department by pretending that none of it is their fault rather than most of it. They have flunked the Tylenol test.
Tylenol Test? Someone else tampered with the tylenol and it killed. The right analogy with Boeing would be if Tylenol itself poisoned the pills… (Or are you downplaying the ‘killing, and trying to direct attention only to the PR afterwards?)
And are you saying that Boeing deliberately set out to kill hundreds of people and lose billions? On the stupd versus evil seesaw the initial coverage from the Seattle Times et al leaned toward evil but more recent revelations may tip the balance to stupid. They made an engineering mistake (for business reasons) but thought the MCAS would never come into play because the last “runaway stabilizer” was decades ago. Failing to account for the AOA sensor (which caused the whole mess) could have been a bureacratic snafu more than a matter of calculation.
Or not. But there’s way too much conclusion jumping IMO.
…Great article; and informative, articulate comments as well…oaf shut up now., back to hovel….
Thanks, Yves…
It looks like Boeing is already taking action – they are changing the name of the Max to… well, some numbers nobody will remember.
I was wondering about the reasoning behind having “optional” controls, warnings, etc… for flying the aircraft. WTF?
Champagne coolers would be optional. We are getting a demonstration of protected crapification under a bright light.
Simple: reduce entry-level package to make product seem more attractive upfront, probably to smaller East Asia airlines (which Boeing sold a TREMENDOUS # of 737 Max’s to).
Boeing CEO still has a job because Boeing stock is UP 10% from January of this year. It’s literally that simple.
As for whether it was his problem or not, looks as though the kickoff of this project happened before his time so its actually hard to pinpoint the blame. Boeing has really really idiotic management all the way from Director level up. They are all non-engineers, mostly Harvard grads from what I can remember. It was so bad, that I cant recall a SINGLE senior director level besides engineering, that was someone who worked his way up the ranks.
There was a joke when I worked in aerospace: when you hire one Harvard grad, its like you hire them all. (meaning, the first one will hire all of his buddies directly into upper management).
Seriously, this is what happens when finance guys run the show, and manage to maximize their own stock options, though to be fair, this is what economics tells us that they would do.
I spoke to Nader’s lawyer about Multi-National’s sales processes, wher I had some experience in a past life, and and explained there are two type of communication between the sales forces and the Company, formal and informal,
Informal depends on personal relationships among employees, and is probably more prevalent in the US, than between the US employees and Indonesia sales force.
And suggested a question, “What did airlines know during the sales process, where US airlines apparently they bought the “enhanced MCAS features, and Airlines in Asia, where they apparently did not buy the “enhanced” MCAS features.
Procuring a number of Airplanes in a Airline involves mountain of paper produced in the procurement process, complete with multiple meetings and memos, where various option were included and excluded, and would reflect what the Airline employees knew of the MCSA features.
Comparing the Knowledge in the US with outside the US might be illuminating.
The MAX seems like a sweater. The more they pull the string, the more it unravels. I don’t see it ever returning to the air. The modifications needed to fix it would require recertification. That would be a severe blow to Boeing and US aircraft manufacturing.
So what happens if/when Boeing manages to BS their way back to flying with some minimum expense ‘fix’ that doesn’t solve the fundamental problems…and another plane crashes?
Excellent post and comments. This isn’t just an issue with the 737 Max but is the inherent outcome of neoliberal globalization’s prime directive to increase stockholder value above all else. This led directly to deregulation, outsourcing, monopolization and labor cost cutting. Boeing management had no idea that they had built a death trap or how to correct the design errors. This is why Dennis Muilenburg is so flat-footed. All this to keep the 737 type certificate and transfer the extra money to the wealthy. The supreme irony is Boeing very existence as North America’s aircraft builder is now dependent on their suppliers who they screwed to cut costs to come up with the fixes they can longer do in house, the FAA which lost the workforce needed to regulate aviation, and the Civil Aviation Administration of China.
If Boeing had any other option, they might be better off to just write off the Max (8200), but wouldn’t it take them years to design a replacement?
If we had an actual ruling class, rather than a cluster of fragments of what used to be (up till maybe 1960 or 1970 or so) a ruling class, it would have come down on Boeing like a ton of bricks for destroying what used to be one of the crown jewels of American capitalism.
(The Pecora Commission in the 1930s is one example of what this can look like.)
…”the new larger-diameter engines had to be moved forward on the wing so they could be higher off the ground. That wreaked havoc on the center of gravity of the plane and its flight characteristics”…
And now we do have confirmed proof that the cost-saving (free money) engine option killed the air-worthiness. All this gratuitous exhumation of ancient code may be therapeutic, but many have grown out of the navel lint and are not distracted, but more curious about the software-framing reality-TV issue. It’s a multi-billion dollar hickey that Boeing refuses to acknowledge and move into the penalty box, not collecting the $200.
I imagine an affordable dream car that plants itself after some rare speed bumps but not to worry, it gets automatic upgrades of the fix online.
The MCAS programming was to make the 737MAX be operable just like earlier 737 versions so as to keep pilot training among various airlines common across various aircraft models.
Setting aside the extreme range of MCAS travel(as implemented is 4 times the range in the original specs), the abrupt change in handling characteristics when the MCAS system is turned off or disabled means that pilots will have to be retrained anyway.
I think it would have been better to bite the retraining costs, and make the planes controls operating more similarly with MCAS on as with MCAS off.
The pilots in the two crashes had no clue and no training to handle the aircraft when the MCAS was turned off.
I’ve been wondering lately if dropping the 757 was a mistake. The bigger 737’s are now in the same passenger capacity range as the 757 and the 757 was designed with high bypass ratio turbofans right from the start so was better arranged to accommodate them than the 737 with it’s short landing gear legs which were fine for the JT-8D’s but have resulted in the need for more and more modifications with newer engines.
The 737 MAX story sure has played out worse than I expected. It’s gotten almost as bad as the DC-10 cargo door fiasco – in that one the THY DC-10 that crashed leaving Orly was one that was on the production line when the door fault was recognized and the fix engineered and the proper mechanism was never built into the aircraft despite inspector’s stamps indicating that it was.
Not exactly, MCAS only turned on after takeoff, after retracting the flaps to zero.
MCAS is not needed during landing either, so for most of the flight MCAS was effectively turned off.
I often wonder what the outcome would have been, if they simply moved the flaps to 1. Then MCAS would not activate. I know they couldn’t as it is not in the manual as a solution.
What the pilots had trouble with was recovering from the out of trim situation they should never have been in.
It would be best to just remove MCAS completely, and train pilots to expect the nose up tendency during climb out to cruising level. Very similar to the NG, which those pilots handle just fine.
It appears the 737 Max simulated the earlier 737.
Simulations work until one reached some limit. The they break.
Boeing lost its way a long time ago
Now the chickens come home to roost
But that hundreds have perished horrible deaths like the 737 hard over rudder and now the 737s issues which are many.
Boeing Needs to be monitored by someone .