Aeronautical Insanity: switching to secondary tolerances

The Something Awful Forums > Discussion > Automotive Insanity > Aeronautical Insanity: switching to secondary tolerances

«‹›2241 »

bull3964: Nov 18, 2000; DO YOU HEAR THAT? THAT'S THE SOUND OF ME PATTING MYSELF ON THE BACK.

Would knowing the AOA disagreed really have helped Lion Air? They knew MCAS was acting up and trimming nose down, knowing it was caused by disagreeing AOA wouldn't have helped them fly out of the situation. They needed to prevent MCAS from pitching nose down and the only way to do that on the MAX was apparently to disable electronic control of the trim altogether and use the manual wheel. Whereas on the previous gen 737, their contrary control inputs would have disabled automatic trim adjustment while keeping electronic control intact and they could correct trim quickly with the column switches and without having to take a hand off the control column.

I mean, that's my understanding of the situation from reading the info that's been posted.

# ? Mar 13, 2019 19:47

Adbot: ADBOT LOVES YOU

# ? May 27, 2024 18:39

Jealous Cow: Apr 4, 2002; by Fluffdaddy

# ? Mar 13, 2019 20:50

KodiakRS: Jul 11, 2012

So how long until someone is slow to react to a real stall because of all the emphasis on erroneous stall indications and colgons a 737?

# ? Mar 13, 2019 20:57

PT6A: Jan 5, 2006; Public school teachers are callous dictators who won't lift a finger to stop children from peeing in my plane

bull3964 posted:

Would knowing the AOA disagreed really have helped Lion Air? They knew MCAS was acting up and trimming nose down, knowing it was caused by disagreeing AOA wouldn't have helped them fly out of the situation. They needed to prevent MCAS from pitching nose down and the only way to do that on the MAX was apparently to disable electronic control of the trim altogether and use the manual wheel. Whereas on the previous gen 737, their contrary control inputs would have disabled automatic trim adjustment while keeping electronic control intact and they could correct trim quickly with the column switches and without having to take a hand off the control column.

I mean, that's my understanding of the situation from reading the info that's been posted.

Yeah, I have no idea, but if the MCAS is still running, based off the AOA sensors, when they are in disagreement, that sounds like a really poor design choice. If you're going to have that system be always-on, then you want at least three AOA sensors as was already pointed out.

# ? Mar 13, 2019 20:59

movax: Aug 30, 2008

Eagerly awaiting any surfacing of e-mail or paper trail(s) where some systems engineers pointed out the perturbing effects of the change to the control law and that the customer notification / training had to be done differently. Do they have customer advocates at Boeing?

# ? Mar 13, 2019 21:04

slidebite: Nov 6, 2005; Good egg

So thought I'd share something that I noticed last night booking some flights to Panama.

I used google flights to check fares and found a good rate on business class @ 817 CDN. Everything else was quite a bit more expensive and less than $100 difference between economy and business. By the time you have a couple bags, price difference it literally gone plus you get the better check in/seat/food/drinks and since it's international you get lounge access too.

Checked the same flight directly on united.com and it was more than $1100 higher per person. Exact same flights.

Any idea how that works other than typical airline pricing voodoo?

# ? Mar 13, 2019 21:10

Ola: Jul 19, 2004

As I've read it (Norwegian paper citing NY Times), MCAS reads off only one AoA at a time, which one depends on which autopilot computer is currently active. So if that AoA fails, MCAS fails. Perhaps it's possible to switch autopilot computers, as you'd do if one failed, but it might not have been a known solution by them at the time.

Hardware failure leading to weird software conditions is exactly what needs to be designed out in a safe way in this type of automation, looks like they haven't done so. There's plenty of computer screens that can display "Warning: Thing A disabled because of error in Thing B". Not just "Thing A disabled", but that's preferable to "Looks fine to me, time to pitch the gently caress down!"

Jealous Cow posted:

It's not great when the stock price does the same thing as the FR24 data.

# ? Mar 13, 2019 21:17

Nerobro: Nov 4, 2005; Rider now with 100% more titanium!

Worth noting, is that we've had at least 5 reports of this happening with american pilots, and without crashes.

I believe there is something to be said about the attitude of american pilots. As noted by flight instructors here, foreign pilots can be... a bit iffy. Especially the asian ones. I don't know anything about the african pilots though.

Maybe.. they're fighting it instead of trying to fix it?

Finger Prince posted:

If you're flying into the ground at 400kts, what do you think your AOA is going to be?

400kts true? Probally about 1deg. Maybe 0, as the airfoil is a lifting profile. it's also worth noting that most airliners have their wings attached at a significant angle of incidence.

# ? Mar 13, 2019 21:24

evil_bunnY: Apr 2, 2003

Finger Prince posted:

a busted AOA sensor shouldn't cause the plane to crash if the pilots don't take extraordinary action.

bing bing bing. It's truly precious watching boeing attempting to talk their way out of this dumpster fire after all the crap everyone gave airbus.

# ? Mar 13, 2019 21:28

bull3964: Nov 18, 2000; DO YOU HEAR THAT? THAT'S THE SOUND OF ME PATTING MYSELF ON THE BACK.

PT6A posted:

Yeah, I have no idea, but if the MCAS is still running, based off the AOA sensors, when they are in disagreement, that sounds like a really poor design choice. If you're going to have that system be always-on, then you want at least three AOA sensors as was already pointed out.

Isn't this all starting to add up to some unsettling hints at deeper issues with the 737 MAX flight characteristics? Boeing took away an indirect override of the system for some reason, one that helped mitigate issues arising from disagreeing sensors while at the same time doing nothing to disable the system if the disagreement happened.

It's clear now that a third sensor is needed with the current behavior, but is the shift in CoG by the engines that severe that they prioritized keeping MCAS running unless very explicitly disabled?

Is it possible they are worried about the rate of change of trim via electronic means when MCAS is disabled that they are forcing the use of the manual wheel with a slower change rate?

The outcome of this one is going to be interesting as the things that lead to the accident all seem like process issues that I thought we would have been beyond at this point.

# ? Mar 13, 2019 21:28

Nerobro: Nov 4, 2005; Rider now with 100% more titanium!

So, lets talk about "why" this is a thing.

The 737 has a tail sized for the 737-100. At those weights. At those design constraints.

The 737-MAX is.. a much bigger plane, with a CG that's in a place where you can get the plane into an unrecoverable state. I suspect it's because the tail can't provide enough control authority. By checking AOA, you can determine what sort of control forces you need, and by enforcing AOA, you can ensure that tail can provide the control forces needed to keep the plane flying where you want it to go.

I am left, somewhat wondering, what departure from control is the point of fear for Boeing. If the plane stalls... Is the tail stalling first, before the wing? That's a big problem, with capital letters. If that tail stalls, the nose is going to go in some wild direction, up or down, and everything then goes into the how does a 737 regain control after it's departed controlled flight. Are we at the point, with the new wing, that the tail stalls first?

So now we've justified the MCAS being on, all the time, because a straight line stall means departing controlled flight. (I suspect accelerated stalls, are done at a high enough airspeed that the tail won't stall first..)

I was discussing in another forum, and perhaps the logic tree for the MCAS is to short, and lacks enough PID's to handle normal flight. Are gusts enough to trip it up? Or... the conditions they set it up for, were just fine for the magical man robots that do flight testing, but puts typical pilots a little to close to MCAS stepping in. Say someone who's used to a 737 Neo has been flying a bit steeper than they really should be, now tries that with the MAX and is getting caught up in it? Say... a pilot with 6000hours in type, and "knows what this plane can do"?

# ? Mar 13, 2019 21:41

Ola: Jul 19, 2004

Nerobro posted:

So, lets talk about "why" this is a thing.

The 737 has a tail sized for the 737-100. At those weights. At those design constraints.

No. http://www.b737.org.uk/techspecsdetailed.htm

Horiz Stabilizer span:
100-series: 10.97 m
Max 8: 14.35 m

But old design limitations has something to do with it. The new engines give more of a pitch up moment when you apply power, so they had to add MCAS.

https://theaircurrent.com/aviation-safety/what-is-the-boeing-737-max-maneuvering-characteristics-augmentation-system-mcas-jt610/

# ? Mar 13, 2019 21:55

Infinotize: Sep 5, 2003

FAA et al should get real about type ratings. Narrow the constraints that determine whether a new rating is required. Stop these bullshit games the industry gets to play by forcing engineers to design a mutant airplane because of business requirements and then slap on unbaked software band aids.

# ? Mar 13, 2019 21:58

e.pilot: Nov 20, 2011; sometimes maybe good
sometimes maybe shit

I still can�t believe a CRJ200 is the same type as a CRJ700/900, the systems and the way they land are so different.

# ? Mar 13, 2019 22:02

Nerobro: Nov 4, 2005; Rider now with 100% more titanium!

Ola posted:

Horiz Stabilizer span:
100-series: 10.97 m
Max 8: 14.35 m

AAaaand I'm wrong.

# ? Mar 13, 2019 22:06

StandardVC10: Feb 6, 2007; This avatar now 50% more dark mode compliant

Nerobro posted:

AAaaand I'm wrong.

The 737-300 and newer have a fillet to the vertical tail that the 737-100/200 don't; the 737NG has a wider horizontal stabilizer than the earlier models, but IIRC the same vertical fin. The Max has a new tailcone but I don't think they did anything to the control surfaces that the NG didn't?

# ? Mar 13, 2019 22:12

azflyboy: Nov 9, 2005

Infinotize posted:

FAA et al should get real about type ratings. Narrow the constraints that determine whether a new rating is required. Stop these bullshit games the industry gets to play by forcing engineers to design a mutant airplane because of business requirements and then slap on unbaked software band aids.

When Bombardier was designing the Q400, it was originally going to be a significantly different airplane from the -1/300 models, with a heated wing, simplified flight deck, and other changes to update the design.

However, airlines insisted that it keep the same type rating, so the end result was something that weighs twice what the Dash 8 -100 did, carries twice as many passengers, and has more than double the horsepower, but kept the same type rating because it still has the same flight deck layout (including some switches that basically do nothing now) as the much smaller versions of the airplane.

# ? Mar 13, 2019 22:31

simplefish: Mar 28, 2011; So long, and thanks for all the ~~fish~~ gallbladdΣrs!

So is type rating more about cockpit layout than handling characteristics or checklist similarity then?

# ? Mar 13, 2019 23:21

hobbesmaster: Jan 28, 2008

Nerobro posted:

Worth noting, is that we've had at least 5 reports of this happening with american pilots, and without crashes.

I believe there is something to be said about the attitude of american pilots. As noted by flight instructors here, foreign pilots can be... a bit iffy. Especially the asian ones. I don't know anything about the african pilots though.

Read the preliminary accident report for Lion air. The previous flight had the behavior that sound like the NASA reports - namely that shortly after take off (400 ft) the plane started pitching down. The Lion Air crew handled that condition fine. In the accident flight the computer start fighting the pilots with everything it could (stick puller and automatic trim) the instant they rotated.

hobbesmaster fucked around with this message at 23:27 on Mar 13, 2019

# ? Mar 13, 2019 23:21

Cat Mattress: Jul 14, 2012; by Cyrano4747

Sagebrush posted:

I feel like there's probably an automated test that could be done by looking at the AoA when the airplane is rolling but still has weight on the nose gear and ensuring that it's within spec.

Then again, the middle of the takeoff roll is probably the worst possible time to be checking for hardware failures

What if you put a multiGNSS (GPS, GLONASS, Galileo, Beidou, etc.) receiver in the nose, another in the tail, for redundancy one in each wingtip, and use the coordinates they all give you to get a relatively precise idea of the position of the plane relative to the Earth? By combining the data you get from several GNSS (and there re four main ones now) you get increased precision and accuracy, and that's even truer near airports thanks to WAAS, EGNOS, and similar augmentation systems.

So you should be able to get in real time the position of at least two static points on the aircraft, and from them the position and orientation of the aircraft itself; which should be able to help the flight computer determine when another sensor is malfunctioning.

# ? Mar 13, 2019 23:39

vessbot: Jun 17, 2005; I don't like you because you're dangerous

Nerobro posted:

AAaaand I'm wrong.

Yes, but not fundamentally so. It's a bigger tail than the 100, but still the same size as a bunch of versions ago and inadequate for the rest of the airframe as-is.

# ? Mar 13, 2019 23:46

vessbot: Jun 17, 2005; I don't like you because you're dangerous

Ola posted:

The new engines give more of a pitch up moment when you apply power, so they had to add MCAS.

This is also an issue with the 737 in general, but not the issue behind MCAS.

MCAS is because of aerodynamic instability from the nacelles, which are bigger and more forward than previous versions. That's on the principle that any surface area aft of the CG is stabilizing, and forward is destabilizing. That's why seaplanes that were originally landplanes, and therefore weren't originally designed with all the forward area of the floats, have extra tail surfaces to compensate.

The thrust-pitch couple, which you refer to, may actually be less on the MAX since the vertical offset of the thrustline is less.

# ? Mar 13, 2019 23:54

Finger Prince: Jan 5, 2007

The 400' AGL is notable because that's when the PFD disagree flag will appear. The AOA could be borked before then, but it won't display until its reached 400'. If what Ola said is true, that the MCAS uses only the AOA Sensor currently used by the autopilot in use, that could be interesting if they switch AP in use each leg. Might explain the different experiences by the two crews. Getting into speculation land now though.

# ? Mar 13, 2019 23:55

Ola: Jul 19, 2004

Cat Mattress posted:

multiGNSS

The problem is system complexity. The solution isn't more complex systems.

Angle of attack sensors are fine. There are simpler ones as well, like the lift reserve indicator: https://www.mountainflying.com/Pages/articles/alpha_systems_aoa.html It works on differential pressure alone, no moving parts. Boeing will probably come up with a perfectly good fix, maybe it's just a few lines of code and an MCAS training bit in the type rating. But it's really bad for aviation as a whole that it took another crash and then some before they responded the way they should respond.

# ? Mar 13, 2019 23:56

Ola: Jul 19, 2004

vessbot posted:

This is also an issue with the 737 in general, but not the issue behind MCAS.

MCAS is because of aerodynamic instability from the nacelles, which are bigger and more forward than previous versions. That's on the principle that any surface area aft of the CG is stabilizing, and forward is destabilizing. That's why seaplanes that were originally landplanes, and therefore weren't originally designed with all the forward area of the floats, have extra tail surfaces to compensate.

The thrust-pitch couple, which you refer to, may actually be less on the MAX since the vertical offset of the thrustline is less.

Ok, thanks, I just read a newspaper source and filled in some blanks wrongly. Makes sense, like the fletching on an arrow.

# ? Mar 13, 2019 23:59

movax: Aug 30, 2008

Ola posted:

The problem is system complexity. The solution isn't more complex systems.

Angle of attack sensors are fine. There are simpler ones as well, like the lift reserve indicator: https://www.mountainflying.com/Pages/articles/alpha_systems_aoa.html It works on differential pressure alone, no moving parts. Boeing will probably come up with a perfectly good fix, maybe it's just a few lines of code and an MCAS training bit in the type rating. But it's really bad for aviation as a whole that it took another crash and then some before they responded the way they should respond.

Yeah, I think the "fix" is easy and straightforward, it just shouldn't have taken killing ~300 people to correct a problem that (guessing) stemmed out of a suboptimal engineering culture / poor communication environment at Boeing.

# ? Mar 14, 2019 00:32

simplefish: Mar 28, 2011; So long, and thanks for all the ~~fish~~ gallbladdΣrs!

Why are the nacelles further forward anyway? I imagine it has to do with aerodynamics and fuel efficiency?

Or is it just that they wouldn't fit under the wing properly unless they were all the way out from under it?

# ? Mar 14, 2019 00:41

TTerrible: Jul 15, 2005

simplefish posted:

Or is it just that they wouldn't fit under the wing properly unless they were all the way out from under it?

From what I've read, I think it is this.

# ? Mar 14, 2019 00:45

FuturePastNow: May 19, 2014

bet Boeing wishes it hadn't shut down the 757 line

# ? Mar 14, 2019 01:00

e.pilot: Nov 20, 2011; sometimes maybe good
sometimes maybe shit

TTerrible posted:

From what I've read, I think it is this.

Yep, even the CFM56 had to have the bottom flattened out and the accessories moved to the side to get enough ground clearance.

# ? Mar 14, 2019 01:13

Brovine: Dec 24, 2011; Mooooo?

Sperglord posted:

How uncommon is it for a country to send a Boeing black-box to Europe for analysis?

Very common. It's a Honeywell unit IIRC and Honeywell Toulouse is probably the most likely repair/download shop. Or Basingstoke.

# ? Mar 14, 2019 01:14

Finger Prince: Jan 5, 2007

I feel like the endless models of 737 are like Ford just keeping on going with the Panther Platform and the A320 was like the Camry.

# ? Mar 14, 2019 01:19

simplefish: Mar 28, 2011; So long, and thanks for all the ~~fish~~ gallbladdΣrs!

e.pilot posted:

Yep, even the CFM56 had to have the bottom flattened out and the accessories moved to the side to get enough ground clearance.

I'm just gonna say it: stilt struts and monster truck wheels

# ? Mar 14, 2019 01:27

hobbesmaster: Jan 28, 2008

Finger Prince posted:

The 400' AGL is notable because that's when the PFD disagree flag will appear. The AOA could be borked before then, but it won't display until its reached 400'. If what Ola said is true, that the MCAS uses only the AOA Sensor currently used by the autopilot in use, that could be interesting if they switch AP in use each leg. Might explain the different experiences by the two crews. Getting into speculation land now though.

Something I�m curious about - what would an airbus do in this situation? Just kick you over to alternative law and turn the stall protection and everything off right?

# ? Mar 14, 2019 01:30