OT: Boeing 737 MAX

[goonster]

Originally Posted by MattTuck

https://twitter.com/trevorsumner/sta...rc=twsrc%5Etfw

Do not like, and here's why:

1. Listing all those factors in this way creates the illusion that they equal/similar meaning and weight. Many of these are not "problems" at all. Demand for more efficient aircraft is not an "economic problem," it's why Boeing makes airplanes. Putting big engines on a low plane isn't an "airframe problem," it's a design challenge, and all projects have constraints. Nobody is saying the 737 airframe is inherently unstable with the LEAP engines.

2. The "not a software problem" conclusion might be true in a very limited sense, but it's wrong in a larger sense. Say that with a straight face and you might be a programmer, but you're not an engineer.

[MattTuck]

Originally Posted by goonster

Do not like, and here's why:

1. Listing all those factors in this way creates the illusion that they equal/similar meaning and weight. Many of these are not "problems" at all. Demand for more efficient aircraft is not an "economic problem," it's why Boeing makes airplanes. Putting big engines on a low plane isn't an "airframe problem," it's a design challenge, and all projects have constraints. Nobody is saying the 737 airframe is inherently unstable with the LEAP engines.

2. The "not a software problem" conclusion might be true in a very limited sense, but it's wrong in a larger sense. Say that with a straight face and you might be a programmer, but you're not an engineer.

Yes, I was the lay person that I referred to in that post. I thought it was useful to understand that the reason this stuff was needed was caused by upstream decisions (some of which Boeing controlled, and some they didn't). If you believe that sequence of events, (which I do), it makes sense as to why each step was taken. It is incrementalism.

So, yes, I agree based on what I've read that the software implementation was terrible and not well thought out (no redundancy, no limits test in the software to say, "Are we telling the plane to dive at an angle that would be unreasonable for normal trim adjustments?" and little or none communication to ensure pilots are able to cope with this situation when it does arise.

But the reason that such software was necessary was not clear to this lay person from the get go, and helps me understand the bigger context.

[cnighbor1]

Boeing engineers are nuts
I am a Architectural engineering grad. LTU 1958 I worked with Architects like myself for my whole career. I consider Architects nuts but with a passion to create something great always in the front
of their thoughts. which is fine
Living in Seattle and knowing Boeing engineers and reading articles on the company, they tended to say and do things with no purpose in mind that that any sane engineer would never even think off.
For instance strikes by workers against Boeing. They always lasted at least 90 days why? Boeing contract with plane buyers always stated if a strike happens and it lasts over 90 days the delivery date of the planes can be revised. That almost makes sense. I could go but I quit here

[Heisenberg]

Originally Posted by goonster

Do not like, and here's why:

1. Listing all those factors in this way creates the illusion that they equal/similar meaning and weight. Many of these are not "problems" at all. Demand for more efficient aircraft is not an "economic problem," it's why Boeing makes airplanes. Putting big engines on a low plane isn't an "airframe problem," it's a design challenge, and all projects have constraints. Nobody is saying the 737 airframe is inherently unstable with the LEAP engines.

2. The "not a software problem" conclusion might be true in a very limited sense, but it's wrong in a larger sense. Say that with a straight face and you might be a programmer, but you're not an engineer.

i mean...the issue is that the 737 is an ancient airframe. boeing nixed creating a new platform in the late 2000s a la dreamliner and band-aided the MAX into existence.

tangentially, i sent that twitter thread to my father, who was an avionics engineer for a company that built business jets. he replied with an amusing story of his own:

Was wondering what the underlying issues were... We worked on similar flight control systems for business jets, the aerodynamic principles are the same. We were called on to correct for air-frame/control system deficiencies on many occasions. Quite curious that there is only a single angle of attack sensor. Seems especially for this aircraft, the flight control software needs to know for sure what that angle is to provide safe operation of the aircraft.

We had an issue with our flight controls on a number of early flights during certification of one of the business jets I worked on. We would deploy ground spoilers if we detected weight on wheels by two separate sensors. (In this case switches.) One that sensed the nose gear and the port gear and the other sensed only the starboard gear. Electrical control system would deploy ground spoilers (sometimes called lift dump) only if both sensor paths agreed the aircraft was on the ground. Typically, on larger aircraft and even some smaller jets, there is a ground sensing radar that is used in conjunction with this system. The Weight on Wheels (WOW) sensors are also typically an analog value that indicates exactly how much weight there was on the wheel. (Instead of just a switch state)

I got called back to headquarters for an emergency when they did not get the ground spoiler deployment when they thought they should. This was an issue since the length of runway the aircraft could land was such that these spoilers needed to deploy as soon as the plane's landing gear touched the runway. I looked at the strip chart to see when the WOW sensors indicated that the aircraft was on-ground and the spoilers deployed and was able to show that the spoiler deployment closely tracked the WOW switches.

This jet had an interesting aerodynamic issue; it would create a cushion of air underneath once it was very close to the ground, called ground effect. This held the aircraft up off the runway just a little bit higher so that the landing gear WOW switches would not all indicate that the aircraft was on the ground. The spoilers would not deploy, and the aircraft would not be able to stop in time on a short runway. A very serious issue which a capable pilot should be able to recover from by throttling up and doing a go around.

The design of the system was changed so that the pilots could manually deploy the ground spoilers rather than rely on the automatic deployment. Made me nervous since mid air deployment can result in a serious event, especially if the aircraft is at cruising speed...

[unterhausen]

If a Boeing software engineer had written that series of tweets they would justifiably be looking for a job at a game company. There was a requirements problem, and the system engineering people didn't really think things through enough. However, I imagine the fix will be pretty much all software, and I see nothing wrong with that.

They somehow decided that this was not critical enough to require more than one sensor. That was a big mistake, and a requirements error. Nobody was ever dumb enough to let me get close to the people actually writing flight software, but it seems they do have a responsibility to make sure that requirements are reasonable.

[Louis]

Originally Posted by unterhausen

They somehow decided that this was not critical enough to require more than one sensor. That was a big mistake, and a requirements error.

I wonder how often this is done. (relying on only one sensor) Ideally you'd have three (or more) so they can "vote" but I doubt that they always have the luxury.

[fa63]

I can't find the article now, but if I recall correctly, there was mention that the two AOA sensors in the Lion Air 737 were off by 20 degrees while the plane was taxiing... And also that Boeing charged extra to have a cabin warning light installed to indicate that the two sensors are reading differently...

This also made me wonder; do pilots not check for consistency between the sensors before take-off?

[unterhausen]

I thought I read today that the second sensor was extra. There is a lot going on in the cockpit before takeoff. A 20 percent error might go unnoticed.

I don't think single point failures are generally allowed in critical flight control systems. It doesn't seem like fully redundant sensors are needed in this case.

[fa63]

Not 20 percent error, 20 degrees.

Edit - found the article:

http://www.seattletimes.com/business...ystem/%3famp=1

[cloudguy]

Originally Posted by unterhausen

I thought I read today that the second sensor was extra.

This seems absurd, from a safety standpoint.

[Louis]

Originally Posted by cloudguy

This seems absurd, from a safety standpoint.

Only if it's highly unusual to have designs where data from only one sensor is used.

[oldpotatoe]

Originally Posted by unterhausen

I am not sure I would consider a uncommanded max nose down input a "trim," so I guess it doesn't surprise me that the pilots involved didn't immediately think to flip that particular switch off. Especially if the MCAS system wasn't mentioned in the training materials or the flight manuals.

Flying with autopilot on..BIG nose down moment..the FIRST thing I would do(yes, not an airline guy, never have been, never seen their training and my experience is manual flying 100%)..is turn off the thing that's 'between' me and aircraft, auto-anything..IMHO, of course.

[oldpotatoe]

Originally Posted by fa63

I can't find the article now, but if I recall correctly, there was mention that the two AOA sensors in the Lion Air 737 were off by 20 degrees while the plane was taxiing... And also that Boeing charged extra to have a cabin warning light installed to indicate that the two sensors are reading differently...

This also made me wonder; do pilots not check for consistency between the sensors before take-off?

AND the company using manufacturer recs will delineate whether or not that is a grounding malfunction or....

Again...yes, manual flying type pilot..IF the AOA was goofy on any of the jets I flew(only one sensor..AOA, NOT Pitot static systems)...I'd use the airspeed indicator...AOA works so that the attitude and stall margin is proper regardless of weight..airspeed changes..15 units AOA..heavier-faster, lighter-slower but as long as the AOA is right, you won't stall BUT if AOA doesn't work..figure out airspeed for the approach for your weight(134 knots plus 2 knots per 1000 pounds over 34,000 pounds..Yikes, I still remember the F-4 numbers), fly that airspeed..Again, back to 'systems' flying w/o reliance on 'stick and rudder' skills when all the whizbang stuff Sh__sthe bed..

[goonster]

Originally Posted by Louis

Ideally you'd have three (or more) so they can "vote" but I doubt that they always have the luxury.

Here's a case where such a system malfunctioned, for reasons never conclusively determined.

[Louis]

Originally Posted by goonster

Here's a case where such a system malfunctioned, for reasons never conclusively determined.

I'm not claiming that redundant systems are 100% safe and realiable, just that I bet that they're more reliable and fault-tolerant than non-redundant systems.