737 Stuck Manual Trim Technique
Join Date: Nov 2018
Location: madrid
Posts: 47
Likes: 0
Received 0 Likes
on
0 Posts
I am also quite uncomfortable with the fact that the pilots struggle to attain the pitch up attitude they wanted. Why let stab have more power than elevator by design?
FDR shows unused elevator travel, so it was a force problem.Feel system being 100% artificial (up to the point in which the hydraulic system does not have enough power to lift the elevator), it leaves two possible scenarios:
1) Feel system may overpower a human (I hope not). If it can, it is in this situation (a lot of airspeed, a lot of deflection, boost because of stall (false)). If it does, why not reduce its power a bit?
2) Elevator could not fight the aerodinamic forces (possible). It may be possible to enlarge the balance tab / increase the size of the hydraulic actuators at the expense of elevator speed?
Not saying those are sensible modifications, only trying to understand that specific design decision (which by the way is quite popular among other manufacturers, but far from standard).
FDR shows unused elevator travel, so it was a force problem.Feel system being 100% artificial (up to the point in which the hydraulic system does not have enough power to lift the elevator), it leaves two possible scenarios:
1) Feel system may overpower a human (I hope not). If it can, it is in this situation (a lot of airspeed, a lot of deflection, boost because of stall (false)). If it does, why not reduce its power a bit?
2) Elevator could not fight the aerodinamic forces (possible). It may be possible to enlarge the balance tab / increase the size of the hydraulic actuators at the expense of elevator speed?
Not saying those are sensible modifications, only trying to understand that specific design decision (which by the way is quite popular among other manufacturers, but far from standard).
Plastic PPRuNer
So many stories of potentially dangerous situations arising from autotrim (particularly those that fail to inform the operator that trim is way beyond expected limits). Autotrim is a nice thing to have - saves constant retrimming as fuel is pumped around and SLF move around the aircraft.
Good to have trim-wheels that clack-clack as they rotate, but I'm sure that that repeated noise moves into the background.
Not good when manual operation of trim wheels virtually becomes cosmetic because of aerodynamic resistance away from the middle-ground. And not good when "grab and hold" is virtually guaranteed to take the skin off your palm.
There is so much (often life saving) automation now that it takes a constantly "aircraft-aware" crew to know what and why the aircraft is doing what it is doing. But automation was designed to reduce this need (and often does). Rather too often for my taste, a minor (and perhaps transcendent) anomaly, leads to a cascade of alarms which the operator has to define and prioritise very quickly indeed using memory items and paging through the vast amount of information not instantly available on the various displays.
Nor is this confined to aircraft [although there is no comparison in complexity], modern automated anaesthetic machines can on occasion lead to the same confusion (although it is much easier to turn all the automation off and revert to manual). The same will apply to increasingly automated cars.
Ever since the days when some sensible fellow decided to put wheel-shaped knobs on the "undercarriage down" levers, research has continued on making the wo/man-machine interface more intuitive and giving it more "common-sense" but it is very far from complete. Writing software, as I sometimes have to do, 80% of the time goes into imagining and dealing appropriately with corner cases and black-swans.
It isn't easy and there is always another black-swan hidden unseen and waiting to be triggered - it has to be hypothesised and the appropriate information provided and acted upon (either by the machine or the operator). Automation isn't tough.
Just supposing that on AF447 a sensible girl's voice had said, "Hi guys. I've just lost all my airspeed indicators. Everything else seems to be OK, so I've turned off the autopilot and suggest that you just fly pitch and power until I get the ASIs back."
Do you think Bonin would have firewalled the throttles and hauled back on the side-stick?
Somehow it doubt it
Mac
Good to have trim-wheels that clack-clack as they rotate, but I'm sure that that repeated noise moves into the background.
Not good when manual operation of trim wheels virtually becomes cosmetic because of aerodynamic resistance away from the middle-ground. And not good when "grab and hold" is virtually guaranteed to take the skin off your palm.
There is so much (often life saving) automation now that it takes a constantly "aircraft-aware" crew to know what and why the aircraft is doing what it is doing. But automation was designed to reduce this need (and often does). Rather too often for my taste, a minor (and perhaps transcendent) anomaly, leads to a cascade of alarms which the operator has to define and prioritise very quickly indeed using memory items and paging through the vast amount of information not instantly available on the various displays.
Nor is this confined to aircraft [although there is no comparison in complexity], modern automated anaesthetic machines can on occasion lead to the same confusion (although it is much easier to turn all the automation off and revert to manual). The same will apply to increasingly automated cars.
Ever since the days when some sensible fellow decided to put wheel-shaped knobs on the "undercarriage down" levers, research has continued on making the wo/man-machine interface more intuitive and giving it more "common-sense" but it is very far from complete. Writing software, as I sometimes have to do, 80% of the time goes into imagining and dealing appropriately with corner cases and black-swans.
It isn't easy and there is always another black-swan hidden unseen and waiting to be triggered - it has to be hypothesised and the appropriate information provided and acted upon (either by the machine or the operator). Automation isn't tough.
Just supposing that on AF447 a sensible girl's voice had said, "Hi guys. I've just lost all my airspeed indicators. Everything else seems to be OK, so I've turned off the autopilot and suggest that you just fly pitch and power until I get the ASIs back."
Do you think Bonin would have firewalled the throttles and hauled back on the side-stick?
Somehow it doubt it
Mac
Mac,
But ‘sensible girl’ (AF447) did not know, could not know, that all sources of speed were corrupt; that was the underlying problem that the situation was beyond the comparator system’s (allowable) certification specification.
In many ways this is similar with the 737 accidents. The crews were presented with a tactile warning of approach to stall, low speed awareness of the same condition, and several alerts from other systems which at the time might not appeared to be related. What would ‘sensible girl’ say, based on those features, and having to consider individual or combined failures.
And again, with flaps up when the trim started to move - MCAS was operating as designed, and again MCAS continued to move trim according to the ‘allowable’ as-designed system. ‘Sensible girl’ thinks OK, and without means of cross comparing AoA, because with only two installed it is impossible to determine which one is correct (another comparator issue). Thus there was no means by which to enter the system logic to identify design weaknesses and conclude that AoA was the initiating factor, and thus correctly alert the crew.
“Good to have trim-wheels that clack-clack as they rotate, but I'm sure that that repeated noise moves into the background.”
Yes, but why was this feature used in the orrigional design? History, but we forget hard lessons learnt.
Was this feature (and others) robust enough to alert the crew against MCAS, or did the failure condition to which the crew should be alerted overcome all warning systems’ and the crew’s ability to deduce the nature, severity, and consequences of the failure?
Back to the drawing board.
But ‘sensible girl’ (AF447) did not know, could not know, that all sources of speed were corrupt; that was the underlying problem that the situation was beyond the comparator system’s (allowable) certification specification.
In many ways this is similar with the 737 accidents. The crews were presented with a tactile warning of approach to stall, low speed awareness of the same condition, and several alerts from other systems which at the time might not appeared to be related. What would ‘sensible girl’ say, based on those features, and having to consider individual or combined failures.
And again, with flaps up when the trim started to move - MCAS was operating as designed, and again MCAS continued to move trim according to the ‘allowable’ as-designed system. ‘Sensible girl’ thinks OK, and without means of cross comparing AoA, because with only two installed it is impossible to determine which one is correct (another comparator issue). Thus there was no means by which to enter the system logic to identify design weaknesses and conclude that AoA was the initiating factor, and thus correctly alert the crew.
“Good to have trim-wheels that clack-clack as they rotate, but I'm sure that that repeated noise moves into the background.”
Yes, but why was this feature used in the orrigional design? History, but we forget hard lessons learnt.
Was this feature (and others) robust enough to alert the crew against MCAS, or did the failure condition to which the crew should be alerted overcome all warning systems’ and the crew’s ability to deduce the nature, severity, and consequences of the failure?
Back to the drawing board.
What rule permits
- two pilots to be on the controls at the same time?
- a necessity to match speed to trim error speed by an abnormal untaught, manoeuvre?
- a design that incorporates a trim change that exceeds the certified trim error under 25.255, (10 seconds v 3 seconds)?
Respectfully, the out of trim standards are inadequate, on brand A or B in the real world, and public safety is hardly being preserved. As an industry, this is pretty untidy.
Join Date: Nov 2007
Location: dublin
Posts: 2
Likes: 0
Received 0 Likes
on
0 Posts
Just supposing that on AF447 a sensible girl's voice had said, "Hi guys. I've just lost all my airspeed indicators. Everything else seems to be OK, so I've turned off the autopilot and suggest that you just fly pitch and power until I get the ASIs back."
my view too. And the girl knew that pitch power and groundspeed did not equate with stall. 450 kts g/s at the time.
And the pilots should have seen this too. Everything is in the right place except for a lot of distracting warnings.
Join Date: Nov 2007
Location: dublin
Posts: 2
Likes: 0
Received 0 Likes
on
0 Posts
On the NG Speedbrakes cause small pitch up but not too significant. Can’t see that they are relevant to recent incidents
Join Date: May 2010
Location: Boston
Age: 73
Posts: 443
Likes: 0
Received 0 Likes
on
0 Posts
Just supposing that on AF447 a sensible girl's voice had said, "Hi guys. I've just lost all my airspeed indicators. Everything else seems to be OK, so I've turned off the autopilot and suggest that you just fly pitch and power until I get the ASIs back."
my view too. And the girl knew that pitch power and groundspeed did not equate with stall. 450 kts g/s at the time.
And the pilots should have seen this too. Everything is in the right place except for a lot of distracting warnings.
This device would be totally independent, 'read only' with no ability to manipulate controls, it's sole role would be to correlate all the data and offer an opinion when asked (or possibly swear if it detected something really scary?)
Since it would be advisory only it would not need to meet the highest level reliability standards. This would allow for more complex functions that would not be easy to certify to the highest level.
It would not be subject to sensory overload or other human foibles and would be able to incorporate history and trends; AoA was 15, one second later it is 75 strongly suggests that is a root problem.
It could also be provided with a system model to allow testing scenarios to determine the best match for the observed warnings and data.
Last edited by MurphyWasRight; 13th Apr 2019 at 00:11. Reason: typos add more complex
Psychophysiological entity
You know, if I was young again and offered a nice new MAX, I'd feel a lot happier if MCAS was not 'connected' to anything except one of the afore mentioned nice voices. "MCAS, lower the nose . . . lower the nose . . . This is your MCAS speaking." I would then decide how my charge was to be flown.
While in this dream state, I could become the boss of Boeing. 'We're going to have to bite the bullet on this one. What's it going to cost us to hinge the Horizontal Stabilizer at the front?'
The strange thing is, I'm not kidding, though I imagine I wouldn't be the one to make the calls.
While in this dream state, I could become the boss of Boeing. 'We're going to have to bite the bullet on this one. What's it going to cost us to hinge the Horizontal Stabilizer at the front?'
The strange thing is, I'm not kidding, though I imagine I wouldn't be the one to make the calls.
Join Date: Mar 2001
Location: Left seat of a Boeing... mostly!
Posts: 45
Likes: 0
Received 0 Likes
on
0 Posts
I don’t usually post here, however seems there’s pool of knowledge (some historical) regards the B737 trim system in this thread. The following observation for your interest..
A simulator experiment … Try this in the sim if you get chance, requires some ‘mishandling’ and in this order:
- Conduct GA (Flap 15) with A/T disengaged – set full TOGA N1
- ALT ACQ at say 2500’ AGL manual flight, leave thrust at TOGA N1
- Electric Trim forward for 10-12 seconds continuously (agreed not a normal trim run by any means, quite difficult to force yourself to run electric trim AND for this period)
- Speed continues to accelerates +/- 220-230 KIAS, SFP1 simulation in flap load relief to F10 (from F15)
- Close thrust levers to idle, aircraft will now pitch down, attitude below horizon, KIAS further increases.
- Elevator full aft required to arrest nose drop (which may not be possible)
Now ‘try’ to apply ANU electric trim, stab trim motor will NOT run ANU unless the elevator is moved to a neutral position. With aft elevator input removed the stab trim ‘progressively’ starts to run ANU with control wheel thumb switches held in ANU throughout this maneuver.
This effect occurs in a similar fashion with two different Level D B737-800 FFS devices it’s been tried in, one FFS is a SFP1 simulator the other straight -800.
Would have to assume from that the stab screwjack loading and it’s effect on the trim motor is part of these particular FFS’s aero modeling? Only these Level D devices or other B737-800 FFS too?
A simulator experiment … Try this in the sim if you get chance, requires some ‘mishandling’ and in this order:
- Conduct GA (Flap 15) with A/T disengaged – set full TOGA N1
- ALT ACQ at say 2500’ AGL manual flight, leave thrust at TOGA N1
- Electric Trim forward for 10-12 seconds continuously (agreed not a normal trim run by any means, quite difficult to force yourself to run electric trim AND for this period)
- Speed continues to accelerates +/- 220-230 KIAS, SFP1 simulation in flap load relief to F10 (from F15)
- Close thrust levers to idle, aircraft will now pitch down, attitude below horizon, KIAS further increases.
- Elevator full aft required to arrest nose drop (which may not be possible)
Now ‘try’ to apply ANU electric trim, stab trim motor will NOT run ANU unless the elevator is moved to a neutral position. With aft elevator input removed the stab trim ‘progressively’ starts to run ANU with control wheel thumb switches held in ANU throughout this maneuver.
This effect occurs in a similar fashion with two different Level D B737-800 FFS devices it’s been tried in, one FFS is a SFP1 simulator the other straight -800.
Would have to assume from that the stab screwjack loading and it’s effect on the trim motor is part of these particular FFS’s aero modeling? Only these Level D devices or other B737-800 FFS too?
Only half a speed-brake
In short: (on your FSTDs) When mis-trimmed ND, the electric trim ANU would not engage as long as there is back-pressure on the yoke to keep the A/C from diving?
Last edited by FlightDetent; 14th Apr 2019 at 12:20.
Join Date: Mar 2001
Location: Left seat of a Boeing... mostly!
Posts: 45
Likes: 0
Received 0 Likes
on
0 Posts
Correct (with these FSTDs)...., in the circumstances described, the electric stab trim will not run ANU until aft elevator loading is reduced/or elevator/control column moved towards a neutral position.
Join Date: Nov 2005
Location: Zulu Time Zone
Posts: 730
Likes: 0
Received 0 Likes
on
0 Posts
...interesting. But the data from ET320 shows that at VMO and mis-trimmed ND, the electric trim worked normally in the ANU direction immediately before they set the switches to cutout.
Join Date: Mar 2001
Location: Left seat of a Boeing... mostly!
Posts: 45
Likes: 0
Received 0 Likes
on
0 Posts
I'm no expert on FSTD qualification, I'm sure someone who is much more knowledgable will be along soon.... would the aero loading effect observed in these two 737-800 Level D certified devices be subject to QTG / MQTG (manufacturer data?) or other approved data source for the simulation aero package ?
Last edited by The Mad Russian; 14th Apr 2019 at 12:41. Reason: typo
Thread Starter
I'd be surprised if sim modelling goes as far as that.
I think we're at risk of losing sight of the wood for the trees.
The angst and indignation that the trim system behaves in the way it does presupposes that Boeing could and should have anticipated that an aircraft could ever get into such a bizarre part of the flight envelope in the first place. The fact that , afaik, this is the first time in 60 years there's ever been an accident attributed to the previously well known jackstall shows that it is so vanishingly unlikely to occur that its hardly fair to have anticipated it. Indeed, Boeing considered it so unlikely they removed mention of the roller-coaster technique from the manual. That certainly looks like an error now, but no one raised theis as a critical failing before these accidents so no one elase had the awareness of it either.
Any aircraft is likely to become uncontrollable if mishandled as comprehensively as the Mad Russioan's interesting post - and 'm sure we could all come up with endless different gross-mishandling scenarios resulting in an accident. Thais doesn't, in my mind, equate to a design flaw of any great proportions as some are suggesting. Aircraft can and will bite if mishandled. You can't design that out of them, and of course it was a measure to mitigate against this that resulted in MCAS in the first place.
I see considerable parallels between the Air France stall crash and this one. In both cases an errant indication was not isolated or properly dealt with leading pilots to mishandle the aircraft to such an extent that they flew it from a perfectly recoverable situation into one that was extremely difficult to recover from, and didn't realise what they had done. There was none of the indignant howlings we hear today saying that Airbus should be sued into penury and it's directors jailed for manslaughter over the AF accident because they failed to anticipate anyone would point the nose at the moon and trim the aircraft into a deep stall with full power on - but what's the difference?
Neither of these accidents would be credible if postulated before the event, you'd be laughed out of town for suggesring such a thing might happen - so I'm rather sympathetic towards Boeing's stance on the trim matter. Some aspects of MCAS are another thing altogether of course, but again how could Boeing have anticipated anyone would fail to carry out necessary drills, fail to reduce power after levelling out, and fail to trim the aircraft sufficiently? They aren't psychic, no one could possibly have inagined this situation in advance so blaming them so harshly as some are doing seems to me a somewhat OTT application of 20/20 hindsight.
I think we're at risk of losing sight of the wood for the trees.
The angst and indignation that the trim system behaves in the way it does presupposes that Boeing could and should have anticipated that an aircraft could ever get into such a bizarre part of the flight envelope in the first place. The fact that , afaik, this is the first time in 60 years there's ever been an accident attributed to the previously well known jackstall shows that it is so vanishingly unlikely to occur that its hardly fair to have anticipated it. Indeed, Boeing considered it so unlikely they removed mention of the roller-coaster technique from the manual. That certainly looks like an error now, but no one raised theis as a critical failing before these accidents so no one elase had the awareness of it either.
Any aircraft is likely to become uncontrollable if mishandled as comprehensively as the Mad Russioan's interesting post - and 'm sure we could all come up with endless different gross-mishandling scenarios resulting in an accident. Thais doesn't, in my mind, equate to a design flaw of any great proportions as some are suggesting. Aircraft can and will bite if mishandled. You can't design that out of them, and of course it was a measure to mitigate against this that resulted in MCAS in the first place.
I see considerable parallels between the Air France stall crash and this one. In both cases an errant indication was not isolated or properly dealt with leading pilots to mishandle the aircraft to such an extent that they flew it from a perfectly recoverable situation into one that was extremely difficult to recover from, and didn't realise what they had done. There was none of the indignant howlings we hear today saying that Airbus should be sued into penury and it's directors jailed for manslaughter over the AF accident because they failed to anticipate anyone would point the nose at the moon and trim the aircraft into a deep stall with full power on - but what's the difference?
Neither of these accidents would be credible if postulated before the event, you'd be laughed out of town for suggesring such a thing might happen - so I'm rather sympathetic towards Boeing's stance on the trim matter. Some aspects of MCAS are another thing altogether of course, but again how could Boeing have anticipated anyone would fail to carry out necessary drills, fail to reduce power after levelling out, and fail to trim the aircraft sufficiently? They aren't psychic, no one could possibly have inagined this situation in advance so blaming them so harshly as some are doing seems to me a somewhat OTT application of 20/20 hindsight.
Only half a speed-brake
The Mad Russian's profile has the shapes and sounds of the Rostov accident. While there is little if any connection on why and how, the "what" i.e. being unable to trim (fast) enough or overcome with the elevator a speed-dive to ground is identical.
With regards to the observed SIM behaviour: sounds like a training device quirk. If it was correctly replicating a jack-screw siezure due excessive loads, someone would have needed to code that behaviour into the simulation software - meaning it must had been a recognized and documented state in the first place. By all accounts, it is not.
Now, in this thread and elsewhere I read about
- (a) elec trim not able to go all the way to extreme NU/ND (but able to come back from within that range)
- (b) manual trim that does not have enough leverage in the mechanism to move THS under major mis-trim and simultaneous elevator load (the yo-yo recovery technique)
WILD SPECULATION: If either or both of the above two got incorrectly programmed for the SIM, a scenario that The Mad Russian describes could ensue.
With regards to the observed SIM behaviour: sounds like a training device quirk. If it was correctly replicating a jack-screw siezure due excessive loads, someone would have needed to code that behaviour into the simulation software - meaning it must had been a recognized and documented state in the first place. By all accounts, it is not.
Now, in this thread and elsewhere I read about
- (a) elec trim not able to go all the way to extreme NU/ND (but able to come back from within that range)
- (b) manual trim that does not have enough leverage in the mechanism to move THS under major mis-trim and simultaneous elevator load (the yo-yo recovery technique)
WILD SPECULATION: If either or both of the above two got incorrectly programmed for the SIM, a scenario that The Mad Russian describes could ensue.
Last edited by FlightDetent; 14th Apr 2019 at 14:45.
Join Date: Mar 2001
Location: Left seat of a Boeing... mostly!
Posts: 45
Likes: 0
Received 0 Likes
on
0 Posts
Indeed this scenario described was observed whilst 'further exploring' UPRT 'stall/upset during go-around' training. The 'quirk' or 'effect' was quite a surprise to the 3x TREs present (at least one of which had flown most 737 Variants back to the -200) in the sim on this occasion, to the extent that the scenario was flown several times to ensure what we'd witness was indeed what we'd witnessed as an effect. Later on another FSTD (same FSTD manufacture).
Interesting all same... perhaps indeed a line of code from the 737s ancestry or perhaps indeed not valid data, albeit within the flight envelope (I.E. not the wrong side of the stickshaker or with excessive bank angles etc).
Interesting all same... perhaps indeed a line of code from the 737s ancestry or perhaps indeed not valid data, albeit within the flight envelope (I.E. not the wrong side of the stickshaker or with excessive bank angles etc).
Join Date: May 2010
Location: Boston
Age: 73
Posts: 443
Likes: 0
Received 0 Likes
on
0 Posts
Indeed this scenario described was observed whilst 'further exploring' UPRT 'stall/upset during go-around' training. The 'quirk' or 'effect' was quite a surprise to the 3x TREs present (at least one of which had flown most 737 Variants back to the -200) in the sim on this occasion, to the extent that the scenario was flown several times to ensure what we'd witness was indeed what we'd witnessed as an effect. Later on another FSTD (same FSTD manufacture).
Interesting all same... perhaps indeed a line of code from the 737s ancestry or perhaps indeed not valid data, albeit within the flight envelope (I.E. not the wrong side of the stickshaker or with excessive bank angles etc).
Interesting all same... perhaps indeed a line of code from the 737s ancestry or perhaps indeed not valid data, albeit within the flight envelope (I.E. not the wrong side of the stickshaker or with excessive bank angles etc).
The sim probably does include aeroloads on the trim system, the video of pilots unable to manually trim while 2.5 out of trim suggests that it does. Also appeared to have good modelling of the 'spring' in the long control cables.
How/if these loads affect the (simulated) trim motor is an open question although it certainly could and would not need to be part of a predefined error package/scenario.
Join Date: May 2010
Location: Boston
Age: 73
Posts: 443
Likes: 0
Received 0 Likes
on
0 Posts
and didn't realise what they MCAS had done.
Agreed, before Lion Air suggesting that a new undocumented feature relying on a single sensor would have a death wish would have indeed been laughable.
Truly tragic is the CYA manner the AD was written, rather than simple and forcefull checklist for MCAS misbehave.
Would be interesting to hear why mention of the 'yo yo' technique was removed along the way.
Additional point:
A true trim runaway (stuck relay/whatever) at high speed could get into this state before the pilots recognized and cut out the trim motors, so hard to see how this is unimaginable. About 5 seconds or so appears to do the trick.
Neither of these accidents would be credible if postulated before the event, you'd be laughed out of town for suggesring such a thing might happen - so I'm rather sympathetic towards Boeing's stance on the trim matter. Some aspects of MCAS are another thing altogether of course, but again how could Boeing have anticipated anyone would fail to carry out necessary drills, fail to reduce power after levelling out, and fail to trim the aircraft sufficiently? They aren't psychic, no one could possibly have inagined this situation in advance so blaming them so harshly as some are doing seems to me a somewhat OTT application of 20/20 hindsight.
Truly tragic is the CYA manner the AD was written, rather than simple and forcefull checklist for MCAS misbehave.
Would be interesting to hear why mention of the 'yo yo' technique was removed along the way.
Additional point:
A true trim runaway (stuck relay/whatever) at high speed could get into this state before the pilots recognized and cut out the trim motors, so hard to see how this is unimaginable. About 5 seconds or so appears to do the trick.
Last edited by MurphyWasRight; 14th Apr 2019 at 22:02. Reason: typo Add additional point.
