Ethiopian Airlines Crash/ B737MAX grounding

errlloyd

Damien360 wrote: »

I'm more surprised that Lion Air is still awaiting delivery of any orders. I thought they would be first to cancel.

I think they're pretty far up the list. So it's either take orders of these planes in 2 years and your business grows, or cancel them and wait an extra 8 years for Airbus.

banie01

EASA have submitted a list of requirements to Boeing to be fulfilled before Max can resume service.

https://samchui.com/2019/07/10/easa-identifies-737-max-autopilot-fault/

Reported by Bloomberg on the 6th of July 2019, the article notes the aircraft will not be allowed to fly again until the requirements are addressed and are tested accordingly.

The five requirements the European Aviation Safety Agency have listed so far are as follows:

Reduce the difficulty manually turning the trim wheel.
Address the unreliability of Angle of Attack sensors.
Address the training situation.
Investigate software issues with a lagging microprocessor.

However the biggest and newest to join the list of problems involves the autopilot system in the aircraft, which reportedly fails to disconnect in some emergency scenarios.

Investigations have concluded that pilots who encounter a stall scenario relating to the Manoeuvring Characteristics Augmentation System (MCAS) or similar flight control emergencies have difficulty disconnecting the autopilot.


Pretty damning that further flight safety and control issues have arisen during the review other than those to which the crashes have been attributed.


EDIT:
Only just saw that Inquitus has already shared this story.
Still...
Makes one wonder what confidence if any can be left for industry in both Boeing and the FAA that these issues were not recognised and rectified during initial certification.

Blut2

The Economist's Gulliver blog (which specialises in travel stories, and is excellent - I'd recommend it to any frequent flyer) had a short post/update on this yesterday:

https://www.economist.com/gulliver/2019/07/10/the-end-is-not-yet-in-sight-for-boeings-737-max-crisis

BOEING’S LATEST order and delivery numbers, revealed on July 9th, made for some grim reading. The company delivered just 239 commercial aircraft in the first half of the year, down 37% from the same period in 2018. And it had no new orders whatsoever in the second quarter for its 737 MAX aircraft, from which it makes around a third of its revenues and profits.

The news is hardly surprising, given that the 737 MAX was grounded in March by safety regulators around the world after two crashes in just five months killed 346 people. This has hit Boeing in the pocket, as it only gets paid for planes when they are delivered, and airlines are refusing to take any MAX jets until they are allowed to fly again.

Furthermore, carriers are losing confidence that the aircraft’s problems will actually be fixed at any point soon. In May Boeing revealed a fix for the software issues that contributed to those accidents, but more issues are emerging. A new problem with the plane’s microprocessors emerged in late June, which will take months to fix. And on July 5th European aviation regulators revealed another new software fault. The Federal Aviation Administration, America’s safety regulator, does not want to unground the MAX until other countries’ regulators do so. As a result carriers are already cancelling more flights this autumn.

The rumour among airline executives is that the MAX may not get airborne again this year. And their patience with Boeing is starting to run out. On July 8th flyadeal, a low-cost airline from Saudi Arabia, became the first airline to cancel an order for 737 MAXs—it had requested 30—in favour of Airbus’s competing A320neo. It may be the first of many more.

Its starting to sound like it might be well into 2020 before its fully air-born again.

Henry Ford III

Blut2 wrote: »

The Economist's Gulliver blog (which specialises in travel stories, and is excellent - I'd recommend it to any frequent flyer) had a short post/update on this yesterday:

https://www.economist.com/gulliver/2019/07/10/the-end-is-not-yet-in-sight-for-boeings-737-max-crisis



Its starting to sound like it might be well into 2020 before its fully air-born again.

If it takes until then for the Max to be properly safe it'll not be a day too long.

plodder

banie01 wrote: »

The five requirements the European Aviation Safety Agency have listed so far are as follows:

Reduce the difficulty manually turning the trim wheel.
Address the unreliability of Angle of Attack sensors.
Address the training situation.
Investigate software issues with a lagging microprocessor.

However the biggest and newest to join the list of problems involves the autopilot system in the aircraft, which reportedly fails to disconnect in some emergency scenarios.

In over three decades working with them, I've never heard the term "lagging microprocessor" before. It sounds like the hardware is not able to cope with whatever additional load the MCAS fix places on it, which is what the link below suggests.

https://www.moonofalabama.org/2019/06/boeings-software-fix-for-the-737-max-problem-overwhelms-the-planes-computer.html

which is pretty alarming really, but not surprising if it's really the case that the computer in question uses Intel 80286 chips (the type used on the very first PC I bought around 30 years ago). I think Boeing is in a wholly different and deeper level of mess than has been hitherto established.

Bob24

plodder wrote: »

which is pretty alarming really, but not surprising if it's really the case that the computer in question uses Intel 80286 chips (the type used on the very first PC I bought around 30 years ago). I think Boeing is in a wholly different and deeper level of mess than has been hitherto established.

You’re being kind to the 80286 saying 30 years, it was released in 1982 :-)

Dumb question, but is using 37 years old CPUs common practice due to reliability concerns or is it just Boeing not coping with moving on to newer tech? For exemple would Airbus use similar CPUs on their latest planes?

(I can understand not using the latest tech as they need things with a proven track record, but it seems like between 80286 and the latest intel Core CPUs there could be a middle ground ... even an 80486 would provide a huge performance boost - 20 to 30 times faster - and having been released in 1989 it’s not exactly like it doesn’t have a proven track record)

Inquitus

Bob24 wrote: »

You’re being kind to the 80286 saying 30 years, it was released in 1982 :-)

Dumb question, but is using 37 years old CPUs common practice due to reliability concerns or is it just Boeing not coping with moving on to newer tech? For exemple would Airbus use similar CPUs on their latest planes?

(I can understand not using the latest tech as they need things with a proven track record, but it seems like between 80286 and the latest intel Core CPUs there could be a middle ground ... even an 80486 would provide a huge performance boost - 20 to 30 times faster - and having been released in 1989 it’s not exactly like it doesn’t have a proven track record)

From what I can google the A350 uses more modern processors, as did the A380, but the A320's still use 80's chips for fly by wire the same sort being used in the 737. It's not an easy subject to get answers on from google, but that is what I found.

plodder

Bob24 wrote: »

Dumb question, but is using 37 years old CPUs common practice due to reliability concerns or is it just Boeing not coping with moving on to newer tech? For exemple would Airbus use similar CPUs on their latest planes?

(I can understand not using the latest tech as they need things with a proven track record, but it seems like between 80286 and the latest intel Core CPUs there could be a middle ground ... even an 80486 would provide a huge performance boost and having been released in 1989 it’s not exactly like it doesn’t have a proven track record)

There's nothing wrong with old technology per-se. What's alarming is discovering that you have hit a limit with its capability, at this stage of a product's life-cycle, ie after it's already been delivered to customers and the production line is ramping up. I can see advantages of using old tech in some respects, eg much lower power consumption, well established development tools, reliability etc and [edit] to be fair, so long as the CPU can do the job then the age of the design is irrelevant. What's shocking is having a system running so close to its rated capability.

No idea what Airbus uses. I don't work directly in the aviation field.

Tenger

I remember reading (5-6 years back) that the ISS used pentium chips on its PCs.
Software used wasnt high demand, they were rugged and proven tech.


Looks at aircraft seatbelts. I think they date back to the 1960s. Aviation likes proven products that dont cost a lot over new tech which is expensive and requires costly testing.

Those CPUs were probably included in the MAX under a 'grandfathering' clause in the certification process. New chips mean new tests.
Boeing have used grandfathering to place B737MAX and the B747i under similar certification as the B737 Classic and the B747-100.

Hotblack Desiato

A320 first flew in, what, 1987? and the tech it used would have to have been a few years old at that stage, yet they were able to implement a full FBW system with envelope protections etc.

As I said on this thread previously, a lot of people were iffy about Airbus going full FBW at the time, but it looks a very wise move indeed now.

At that time the 737 was about 20 years old, selling like hot cakes and most of its development costs long paid for, so you can see why Boeing were unwilling to make the massive investment in a whole new airframe to compete with the 320. The MAX really does appear to be as far as they can take the 737 concept or, perhaps, a step too far. If they start having to certify new FMCs, or the MAX goes onto a different type certificate, then the economic case for it - a cheap upgrade to the NG that allows you to keep a common pilot pool - starts to evaporate.


According to Wikipedia:
Early A320s used the Intel 80186 and Motorola 68010.[97] In 1988, the flight management computer contained six Intel 80286 CPUs, running in three logical pairs, with 2.5 megabytes of memory.[98]

Thirdfox

Bob24 wrote: »

You’re being kind to the 80286 saying 30 years, it was released in 1982 :-)

Dumb question, but is using 37 years old CPUs common practice due to reliability concerns or is it just Boeing not coping with moving on to newer tech? For exemple would Airbus use similar CPUs on their latest planes?

(I can understand not using the latest tech as they need things with a proven track record, but it seems like between 80286 and the latest intel Core CPUs there could be a middle ground ... even an 80486 would provide a huge performance boost - 20 to 30 times faster - and having been released in 1989 it’s not exactly like it doesn’t have a proven track record)

Hey if they upgrade it to the intel i960 (80960 first released in 1984) they could power a F22 raptor...the best air superiority jet fighter in the world today

https://www.quora.com/Lockheed-Martin-F-22-Raptor-How-fast-is-the-CPU-of-an-F-22-of-the-latest-model
https://en.wikipedia.org/wiki/Intel_i960

cnocbui

Back in the late 70's, I bought a HP 41C calculator, that reputedly had equivalent computing power to that onboard the Apollo 11 command module, 8 years previous.

Modern small feature size processors would be intrinsically less reliable than the processors made late last century.

L1011

Hotblack Desiato wrote: »

or the MAX goes onto a different type certificate

There are plenty of things that would not be allowed on a new type cert and are grandfathered in - so this would kill the product.




I thought the 737 used Motorola 680x0 series rather than Intel, which are similarly ancient though.

Bob24

plodder wrote: »

There's nothing wrong with old technology per-se. What's alarming is discovering that you have hit a limit with its capability, at this stage of a product's life-cycle, ie after it's already been delivered to customers and the production line is ramping up. I can see advantages of using old tech in some respects, eg much lower power consumption, well established development tools, reliability etc and [edit] to be fair, so long as the CPU can do the job then the age of the design is irrelevant. What's shocking is having a system running so close to its rated capability.

Yeah agree there’s nothing wrong with old tech as such, but what I find surprising is the lack of evolution altogether for decades even though it seems clear that the currently used tech has basically reached its limits in terms of controlling a modern aircraft. As I said a 486 architecture still easily qualifies as old tech and would give them a lot more leeway. Not having upgraded the tech before actually hitting its limits would seem like a massive lack of foresight.

Also to be honest I doubt power consumption is a factor when they decide to use older chips, in fact compared to a 286 architecture, a modem chip would probably handle the same task both in a much faster way AND using less power. And in terms of having well established development tools and proven reliability it is true if you compare a 2 years old chip to a 20 years old chip, but they are way past that stage and would have many equally proven options which are significantly faster, so I don’t think it applies in this case: if you look at development tools and reliability track record I don’t think anyone can honestly say there is a difference between a 286 and let’s say a 486 which would be be way faster.

Bob24

Hotblack Desiato wrote: »

According to Wikipedia:
Early A320s used the Intel 80186 and Motorola 68010.[97] In 1988, the flight management computer contained six Intel 80286 CPUs, running in three logical pairs, with 2.5 megabytes of memory.[98]

Would be interesting to know if the 320neo still uses those, but I couldn’t find the answer.

cml387

I would guess (speaking as an electronic engineer) that older processors make it easier to proof the software for reliability. It would be still possible to go down to machine code level and demonstrate 6 sigma reliability.

goingnowhere

A320 followed best practice

Triplicate systems
Use majority voting
Use different hardware/software developed independently
The odds of multiple CPU's from different vendors based on different architectures running software written by different teams all hitting the same bug at the same time is effectively zero


The legacy processors, 286/386 and the 680x0 all have a lot in common.
1. There is a lot of other people using them so bugs would be found (sales of 680x0 outstrip x86 by a good margin)
2. They are incredibly robust
3. They don't need fancy cooling
4. There provide deterministic execution, none of the multi threaded, cached, predictive execution stuff. We know exactly what the processor will do

The math required to do FBW on a passenger airliner is straightforward no way does it need anything like the CPU performance of today to crunch the numbers. This is all written in assembly and Ada and other ugly stuff, its light weight none of the bloat of an OS.

The Apollo CPU (it wasn't a CPU but boards stuffed with NOR gates on IC chips, super easy to understand and visualise) ran at 2Mhz, the SSI processors running the Irish railway network are, 2Mhz. This works as there is a very clear task to perform. Much is made about my watch is faster than the Apollo computer, I doubt any pocket (non graphing) calculator would match its abilities. It was moving along at 25,000mph+ and 2Mhz was enough to get to the Moon and back and to do so with a degree of accuracy which we would struggle to beat today.


737 is basically a 727 with a 707 wing with 2 engines, there are still wires and pulleys. Robust, understood and easy to maintain tech down country. You cannot retrofit FBW, you got to go back to the start

Bob24

The CEO of AIG says the main factor for recently ordering the 737 Max is frustration with Airbus which has a 70 days delay with deliveries of the 320: https://www.irishtimes.com/business/transport-and-tourism/iag-chief-says-frustration-with-airbus-a-factor-in-boeing-purchase-1.3955112

So he’s not happy with a manufacturer experiencing 70 days delay in deliveries and in order to address that he goes for an airplane whose manufacturer hasn’t delivered a single unit for 4 months and which is currently not allowed to fly anywhere in the world? :-)

Sure, we all believe price wasn’t the main factor ;-)

banie01

goingnowhere wrote: »

You cannot retrofit FBW, you got to go back to the start

I was just going to make a similar post.

Boeing have been caught trying to game the certification system.
They have failed catastrophically and people have died because they tried to shoe horn a poorly implemented electronic solution into an airframe that is very much a mechanical control system.

FBW works because the entire flight control model is tested, certified and has intense scrutiny and quality control.
For an example of what happens with FBW goes wrong, take a look at the delays SAAB encountered with the Gripen flight software and induced rolling oscillation.
It added years to the in service time for the airframe.

Boeing took the path of least resistance in deciding to keep the mechanical flight control and just add fixes.
There was no in depth testing, seemingly no actual integration control and because these changes were grandfathered in on the 737 type cert...

No validation or confirmation that the electronic systems operated to the required standards that would be needed for an FBW system.

Penny pinching and lax regulation have allowed Boeing to push the very limits of what they could get away with.
That the issues only seem to roll on, rather than Boeing getting ahead of them yet is worrying.

I posted a while back that this could be Boeing's McDonnell moment.
The more it drags on, the more I believe it will be.

Damien360

In fairness, Boeing’s sales weren’t harmed for many years. Why would they change. A new era of oil, environmental awareness and more people flying has forced their hand. They were not prepared and that has to be levelled at the CEO.

Irish Steve

It was a long time ago, so I may have some things slightly wrong here, I seem to recall that the reason for the continued use of "old" processors was down to a problem found many years ago with the underlying operation of the 486 processor, which meant that it did not correctly process some specific instructions and provide a valid result, the error was somewhere deep within the architecture of the processor, and there was not a quick fix to resolve it.

At the time, it was held up as a "big deal", because it meant that the processors could not pass a specific validation test that was required before the chips could be certified for "mission critical" applications.

One of the consequences of this was a long time after production of 386 chips had been terminated, Intel had to de mothball the dies and jigs for the production of 386 processors to produce a specific batch of chips for Boeing and (I think) Smiths, who needed them for the 737 FMS system, due to the stocks of chips held by Boeing and Smiths having been exhausted by the continuing production beyond the originally anticipated time scale.

I'm not sure that the 486 was ever fixed, and the later processors that followed it were changed to ensure that the same error did not carry forward, but it meant a hiatus in the supply chain for a long time.

plodder

goingnowhere wrote: »

A320 followed best practice

Triplicate systems
Use majority voting
Use different hardware/software developed independently
The odds of multiple CPU's from different vendors based on different architectures running software written by different teams all hitting the same bug at the same time is effectively zero


The legacy processors, 286/386 and the 680x0 all have a lot in common.
1. There is a lot of other people using them so bugs would be found (sales of 680x0 outstrip x86 by a good margin)
2. They are incredibly robust
3. They don't need fancy cooling
4. There provide deterministic execution, none of the multi threaded, cached, predictive execution stuff. We know exactly what the processor will do

The math required to do FBW on a passenger airliner is straightforward no way does it need anything like the CPU performance of today to crunch the numbers. This is all written in assembly and Ada and other ugly stuff, its light weight none of the bloat of an OS.

That's what you would expect and maybe it is true of the Airbus systems, but this "lagging microprocessor" issue suggests they have run into some performance limit, which is something that should not happen on real-time, safety critical control systems.

I know we have posited some reasons why they might prefer to use old processors, but I think the most convincing reason as some have suggested, is just design inertia, and not wanting to have to recertify new designs.

It will be really interesting to find out what this issue is about in more detail.

The Apollo CPU (it wasn't a CPU but boards stuffed with NOR gates on IC chips, super easy to understand and visualise) ran at 2Mhz, the SSI processors running the Irish railway network are, 2Mhz. This works as there is a very clear task to perform. Much is made about my watch is faster than the Apollo computer, I doubt any pocket (non graphing) calculator would match its abilities. It was moving along at 25,000mph+ and 2Mhz was enough to get to the Moon and back and to do so with a degree of accuracy which we would struggle to beat today.


737 is basically a 727 with a 707 wing with 2 engines, there are still wires and pulleys. Robust, understood and easy to maintain tech down country. You cannot retrofit FBW, you got to go back to the start

Comhrá

Key Points

United Airlines on Friday said it will extend its Boeing 737 Max groundings through Nov. 3, amounting to 2,100 cancellations in September and 2,900 in October.
United, which has 14 Max jets in its fleet, had previously removed the planes from its schedule through Aug. 3.
The 737 Max has been grounded worldwide since mid-March, after its anti-stall software was implicated in two deadly crashes in October and March.


https://www.cnbc.com/2019/07/12/united-airlines-extends-boeing-737-max-cancellations-to-early-november.html

Muahahaha

Will the airlines with grounded aircraft sue Boeing for all their cancellations or what way does that aspect of the grounding get resolved between them?

Hotblack Desiato

They will absolutely be looking for compensation, but it's in the interests of both parties to keep this on the q.t. rather than go all guns blazing into court. Agreements will be reached behind the scenes if they haven't already.

It's going to be an expensive exercise for Boeing whatever happens.

Why didn't they make the landing gear longer so the engine could be placed where it should have been?

L1011

Deleted User wrote: »

Why didn't they make the landing gear longer so the engine could be placed where it should have been?

Would have required fuselage changes for larger gear compartments and would have pushed beyond what they could do on a grandfathered cert.

The entire aircraft is uncertifiable as a new design. The MAX11 is to have bending gear to reduce the risk of tailstrikes from the gear being so stumpy

Engines in front of the wing isn't inherently a bad idea, the Dassault Mercure was designed that way from day 1. But this isn't done right

Thirdfox

Muahahaha wrote: »

Will the airlines with grounded aircraft sue Boeing for all their cancellations or what way does that aspect of the grounding get resolved between them?

This should all be covered by the contract terms (speaking as a lawyer) if they used lawyers worth their salt (and they most definitely would have for multi-billion dollar contracts).

There would generally by exceptions for force majeure events but design flaws would clearly fall outside these exceptions. These contracts would be confidential and you could only guess what terms are but often you might find staged escalation of compensation paid i.e. delay of 2 months = X, delay of 10 months = Y. However Boeing made be able to get more favorable terms since they are such a huge company.

In essence - the parties imvolved will be having their lawyers pore over the contracts to 1) verify what happens in these situations and 2) ensure the wording is ironclad (or some law firm might be in trouble) - squeaky bum time for drafting lawyers

STB.

Bob24 wrote: »

Very few European airlines on that list. Is it that they just don’t buy many planes in general at the moment, or they don’t like that specific one?

Just not that specific one.

Norwegian, Ryanair and TUI were the Europeans interested in that model. IAG (Aer Lingus/Iberia/BA) ordered 200 last month probably at a bargain price.

Ryanair already used Boeing and their strategy seems to be to sell their 737-800 range young to avoid large maintenance costs.

GM228

IAG have not "ordered" anything just yet, there is just a LOI with Boeing, not quite the same and it does not necessarily mean 200 will actually be the number ordered in the end.