OT: Boeing 737 MAX

[goonster]

Originally Posted by Tony T

It's may not be just the software, but possibly how the software responds to erroneous information

Yeah, that's a software problem.

How a control system responds to a sensor failure, or redundant instrument signal mismatches, is a core piece of the functional scope.

Based on the very limited, preliminary information trickling out, it sounds like there are three issues going on:
[+]AOA sensor failure
[+]MCAS misinterprets input array
[+]Pilot cannot assess active system state* in time

(* = MCAS is engaged, but operating in error)

Only the second item would be a previously unknown system defect.

I don't work in aviation, so am unfamiliar with the design standards and safety culture there, but in my world if operators are repeatedly confused by automation, that's poor system design. Grounding is justified on technical merits, imho.

This smells very similar to AF447, where there was no system malfunction per se, but the pilots failed to assess the active autopilot state.

[AngryScientist]

Originally Posted by benb

Most accidents come down to pilot error...

right on. i bet if oldpotatoe, saab, or sully were flying those planes they wouldnt have gone down

i can just hear it on the black box recorder now:

"who designed this autopilot? SRAM? - Pull the fuse"

[echappist]

I’ll add that the panel that certified this plane in the first place included many company employees. Had to be done, apparently, b/c it made such panels cheaper to staff.

While the plane may very well be airworthy, that the people who approved it have inherent conflict of interest should raise flags about objectivity

[oldpotatoe]

Originally Posted by AngryScientist

right on. i bet if oldpotatoe, saab, or sully were flying those planes they wouldnt have gone down

i can just hear it on the black box recorder now:

"who designed this autopilot? SRAM? - Pull the fuse"

Great morning giggle..thanks but my era of flyin was pulley, cable, actuators. I did fly a fly-by-wire aircraft(F-16N) but preferred the manual, A-4, F-4.

Autopilot? What is that? Never had to worry about the little lady or guy 'in the back' who didn't like the idea that their 'room/seat' was moving. My GIB(Guy IN Back), RIO..185 pound radio tuner and somebody to borrow money from when ya got stuck in BumFarge, USA...I did fly with a couple who got yorppy every flight..oh well..

2 things to never say to your RIO

"what was that" and
"watch this"...

[Hardlyrob]

The Wall Street Journal today reported that the original MCAS angle of attack system was based on input from one sensor. Boeing is rushing a software fix for the 737 Max 8 that will take input from multiple sensors. This reportedly has been in the works since the Lion Air crash, but FAA slowed the process in arguing over engineering specifications and details. The update reportedly takes about an hour to install and test.

I find it hard to understand why a critical system would be based on one sensor, when multiple sensors are already available.

Then again, I'm not a pilot or aircraft or systems engineer - I just fly a lot.

Rob

[saab2000]

Originally Posted by AngryScientist

right on. i bet if oldpotatoe, saab, or sully were flying those planes they wouldnt have gone down

i can just hear it on the black box recorder now:

"who designed this autopilot? SRAM? - Pull the fuse"

Don’t assume we don’t make mistakes. Every pilot makes mistakes on every flight. Our job is to catch them through crew coordination, procedures, checklists, etc.

Ethiopian Airlines is a serious airline with global operations. The captain was an experienced pilot.

Like everyone, I hope the issue is discovered and resolved ASAP.

[oldpotatoe]

Originally Posted by Hardlyrob

The Wall Street Journal today reported that the original MCAS angle of attack system was based on input from one sensor. Boeing is rushing a software fix for the 737 Max 8 that will take input from multiple sensors. This reportedly has been in the works since the Lion Air crash, but FAA slowed the process in arguing over engineering specifications and details. The update reportedly takes about an hour to install and test.

I find it hard to understand why a critical system would be based on one sensor, when multiple sensors are already available.

Then again, I'm not a pilot or aircraft or systems engineer - I just fly a lot.

Rob

I think the issue is not having only one sensor but what that one sensor tells the computer and then what the computer then has the aircraft do. What slays me(and SAAB can talk more about this) is the airline reliance on the 'computer' via an autopilot, to fly this thing and the inability of aircrew to see that the 'autopilot/autotrim' is fighting them and they should just turn it off..

I felt the same thing with the autopilot in the F-14..when you try to override it..manually move the aircraft while it's on, it certainly fights you, trying to keep the aircraft straight and level....so, turn the thing off..

Most all military and civilian(pretty sure) aircraft have one AOA 'probe'...multiple pitot static sensors but one AOA.

Just forward of the '2' on the F-4, just aft of the '2' on the A-4.

[goonster]

Originally Posted by Hardlyrob

I find it hard to understand why a critical system would be based on one sensor, when multiple sensors are already available.

Because you have to distinguish which one is good and which one is bad.

In my experience (again, non-aviation) sensors provide signal status, so you default to one, and switch to another if the status goes bad. It is very, very rare for sensors to go bad and provide a significantly bad process value without also throwing a bad status somewhere.

You can also generate an alarm if redundant signals deviate from each other by a certain amount, but then you have to decide how to respond, i.e. continue to operate or go to held/faulted state. It would be poor design to just take an average of multiple signals, without determining which one is bad, and by how much.

[zmudshark]

Originally Posted by Hardlyrob

The Wall Street Journal today reported that the original MCAS angle of attack system was based on input from one sensor. Boeing is rushing a software fix for the 737 Max 8 that will take input from multiple sensors. This reportedly has been in the works since the Lion Air crash, but FAA slowed the process in arguing over engineering specifications and details. The update reportedly takes about an hour to install and test.

I find it hard to understand why a critical system would be based on one sensor, when multiple sensors are already available.

Then again, I'm not a pilot or aircraft or systems engineer - I just fly a lot.

Rob

It was reported that work on the software fix was halted during the government shutdown.

[benb]

There is so much conjecture from journalists who know little that it's really hard for me to believe much of anything.

I have also seen mention that the co-pilot on the Ethiopian flight had only 200 hours of flight time. That would be a red flag, if it was true, but I have no idea if it's true, cause you can't trust much these days.

The operating rule of everyone in the press or otherwise on the web is "get it out there now to get ad dollars, worry about facts later." So it's hard to get to worried about anything till more time occurs.

It is definitely a fact the press leaves millions on dollars on the table if they don't sensationalize this, get stories out right away, or wait for hard facts.

[goonster]

Originally Posted by benb

It is definitely a fact the press leaves millions on dollars on the table if they don't sensationalize this, get stories out right away, or wait for hard facts.

346 people died, and you're whining about the press?

[unterhausen]

sometimes there needs to be a fuss in order for people to do the right thing. I know that I would have been somewhat reluctant to ground this airplane given the fact that the real fix is pilot training. But governments are forcing their hand.

Lion Air pilots had a 3 hour online course to qualify for this plane, that's not enough. It's clear that the issue is serious enough that simulator time is needed. It happens at a bad time in the flight, it's not an easy thing to recognize in time. But I'm pretty sure pilots are going to know where that switch is now. Might even be taped over on some airlines.

[benb]

Originally Posted by goonster

346 people died, and you're whining about the press?

The press wants to point all the fingers at the US and/or the FAA and/or Boeing.

Most of these accidents get investigated and it turns out to be Pilot Error, but no one in the press wants to point that finger.

Lion Air in particular.. you should be pointing a lot of fingers at them before you point at the airplane. Low training requirements, meth in the cockpit, blacklisted from flying into large areas of the world until very recently, crashes/hour rates orders of magnitude higher than US airlines. And yet it is instant knee jerk to assume it's the airplane.

The other thing about all this hysteria is it's still probably orders of magnitude safer to hop on a Lion Air or Ethiopian flight than it is to hop in the car and drive across town and we are fine ignoring the auto risk, and hopping on one of our domestic airlines is vastly safer than Lion or Ethiopian.

[jet sanchez]

The secretary of Defense is a longtime Boeing exec, there's no way the FCC will be grounding any Boeing jets

[unterhausen]

I feel like the pros at the FAA would ground this plane if they thought it was the right thing to do. I am sure they are somewhat hesitant, as I would be.

I'm more concerned about the AOA sensor failure rate. An airframe I used to work on would crash if the AOA sensor went out. I can't imagine too many companies make them.