Air India plane crash

[///AMG]

Sorry guys, I should have been more clear re: THR HOLD. I was talking about a scenario where AT was not actually selected but was thought to have been……and then a rollback airborne. I dont think I have found the smoking gun with this one, just curious. Obviously we have several checks that should catch this where i work, but i dont know what they are working with at AI of course. Lord knows only some people are starting to look at the FMAs

Obviously flaps before gear is theoretically possible, but i cannot imagine myself doing that, based on the ergonomics alone. So I wouldn’t imagine someone else would either. But yeah, it’s possible, and maybe supported by UFO footage level granular data as well.

I just cant think of a reason for two engines rolling back at the same time other than 1) birds/FOD, or 2) aircrew f*** ery and some automation failure. But then again I have a lot less airline flying experience than the average cat here, and I have a tiny brain sometimes too

 

[Cherokee_Cruiser]

So they retracted the flaps instead of the gear and then deployed the RAT? All in that time frame? I just don’t get this one. The captain was very experienced and my guess would have seen the FO do something like that pretty quickly considering that’s not a very easy ‘mistake’ to do with the flap and gear handles.

What world do we live in that an 8000 total time as a widebody CA is considered “very experienced.”

As a 14,000 hr guy, I’m just now figuring out what makes an airplane go weeeeeee:

 

[schloopy]

Really good comment elsewhere from a current B-787 captain.

Crew set reduced thrust parameters incorrectly, but had enough to get off the ground.

They had 200' set instead of 2000 in ALT window.

At positive rate call, the PF called for gear up, ALT SET. The aircraft was at about 200', so auto-throttles rolled back to idle and the PF thought he had a dual engine failure.

Just passing this along.

If this is the case, that would seem to imply the reduced thrust wouldnt have much to do with it nearly as much as the FD behavior (and unless this pilot has inside info re: reduced thrust, that reads more like a possibility than a fact to me personally). But in that context, that seems to imply that an accident like this is possible simply anytime a TDZ altitude is missed during preflight? Certainly shouldnt be missed and with enough altitude, probably not too difficult of a manual recovery but that does seem a bit hard to believe…

Also of course not typed in the 78 but casually familiar with Boeing logic; if this was a VNAV departure which it seems like most Dreamliners use, you would expect the thrust to roll back upon alt capture but would it not start to come back pretty quick to maintain speed target? Or would the A/T still being in thrust hold at that point prevent that from happening?

 

[///AMG]

If this is the case, that would seem to imply the reduced thrust wouldnt have much to do with it nearly as much as the FD behavior (and unless this pilot has inside info re: reduced thrust, that reads more like a possibility than a fact to me personally). But in that context, that seems to imply that an accident like this is possible simply anytime a TDZ altitude is missed during preflight? Certainly shouldnt be missed and with enough altitude, probably not too difficult of a manual recovery but that does seem a bit hard to believe…

Also of course not typed in the 78 but casually familiar with Boeing logic; if this was a VNAV departure which it seems like most Dreamliners use, you would expect the thrust to roll back upon alt capture but would it not start to come back pretty quick to maintain speed target? Or would the A/T still being in thrust hold at that point prevent that from happening?

Other thing is that even at ALT capture wouldn’t the AT and VNAV SPD (or whatever variant of that the 787 has) command thrust to maintain the commanded speed? I’m probably missing some logic loophole here, but that’s what would make sense to me.

 

[jtrain609]

Really good comment elsewhere from a current B-787 captain.

Crew set reduced thrust parameters incorrectly, but had enough to get off the ground.

They had 200' set instead of 2000 in ALT window.

At positive rate call, the PF called for gear up, ALT SET. The aircraft was at about 200', so auto-throttles rolled back to idle and the PF thought he had a dual engine failure.

Just passing this along.

Who the • would design the auto thrust system to work like that.

Oh, *checks notes,* Boeing.

 

[///AMG]

Who the • would design the auto thrust system to work like that.

Oh, *checks notes,* Boeing.

Well i think we can all rest assured that Boeing will be blamed for this. Too much media $$$ to be made for that opportunity to be missed

 

[Wardogg]

Haha. I deserved that. I meant what’s the source.

And shouldn’t set touchdown zone in level change or v/s, I’d agree with.

NASA callback.

CALLBACK 545, June 2025, A Snapshot of Altitude Deviations

ASRS CALLBACK Issue 545 - June 2025, A Snapshot of Altitude Deviations Problem viewing / mobile device: VIEW ONLINE Issue 545 June 2025 Assured vertical separation from terrain and other aircraft is o

myemail.constantcontact.com

 

[JDean3204]

What world do we live in that an 8000 total time as a widebody CA is considered “very experienced.”

As a 14,000 hr guy, I’m just now figuring out what makes an airplane go weeeeeee:

I’m at 5,000 and just started turning left once getting on the plane. That’s if I can find the plane amid asking people where terminal 6 is because I’m in baggage claim again

 

[JDean3204]

Other thing is that even at ALT capture wouldn’t the AT and VNAV SPD (or whatever variant of that the 787 has) command thrust to maintain the commanded speed? I’m probably missing some logic loophole here, but that’s what would make sense to me.

Correct me if I’m wrong, but even in VNAV when the altitude captures, the window opens to current speed at altitude capture and the FMA reads FMC Spd. So if the logic is like the 737 and that did happen, I would think it would open in the realm of V2+ whatever they were at when they hit the alt cap.

Now, the nose dipped slightly then pitched up, maybe the autothrottles rolled back to maintain that speed thinking they would level out at the captured altitude. The engines rolled back, causing nose to drop, PF said nope and pulled but never examined the state of the AT as the PM was giving mayday loss of engine call. But that is literally just Monday morning quarterbacking my thoughts based on the altitude being in fact 200 in the window.

I don’t know exactly how the AT would handle such a deviation from an ALT cap knowing you need the selected speed but you’re leaving the FD as you continue to climb increasing the pitch. Does it add power based solely on command speed or roll back more because it thinks you’re too high. Mind buster

 

[Baronman]

All this talk of weird ALT captures..... At that point do we say that wasn't the RAT we heard?

 

[Fixtur]

1. Never flown anything with autothrottles, but don't they automatically add power to maintain some selected airspeed, or at least keep the airplane from stalling?

2. I can't imagine any crew just letting the airplane settle into the ground without any attempt to disconnect the automation and jam the throttles to the stops.

 

[BlueMoon]

Of all the things I’ve read, this sounds most plausible. Scary if true

I just finished 767 school and they make a big deal about low level altitude captures after take off. Not because the throttles come back at ALT CAP, but because the A/T go from THR HOLD to N1 in the FMA if no vertical mode had been selected, which drives the thrust levers to the thrust limit resulting in an overspeed. (We select Vnav at 1000’)

Now I’m not sure how consistent Boeing automation logic is between fleet types as this is my first Boeing.

 

[jtrain609]

1. Never flown anything with autothrottles, but don't they automatically add power to maintain some selected airspeed, or at least keep the airplane from stalling?

2. I can't imagine any crew just letting the airplane settle into the ground without any attempt to disconnect the automation and jam the throttles to the stops.

It depends what mode it's in, and who designed the AT system.

 

[jtrain609]

I just finished 767 school and they make a big deal about low level altitude captures after take off. Not because the throttles come back at ALT CAP, but because the A/T go from THR HOLD to N1 in the FMA if no vertical mode had been selected, which drives the thrust levers to the thrust limit resulting in an overspeed. (We select Vnav at 1000’)

Now I’m not sure how consistent Boeing automation logic is between fleet types as this is my first Boeing.

I always got the impression that Boeing A/T logic and Embraer A/T logic were pretty similar, if not identical.

 

[BlueMoon]

I always got the impression that Boeing A/T logic and Embraer A/T logic were pretty similar, if not identical.

I think for the most part they are.

I think the logic in this case is to not have the throttles come back with a potential nose high attitude and a slow airspeed on climb out.

No idea if this is the case for the 787.

 

[Nishmaster]

For most (all?) AT implementations the control loop is just comparing acceleration towards or away from the target IAS. It doesn't do anything like "well I'm supposed to be leveled off so I'll just set the power to what would be needed for level flight".

In the case of a low level off the AT would produce whatever thrust was necessary to maintain the target speed. If the pilot pulled back on the yoke at that point, the AP would have disconnected (with accompanying warning AUTOPILOT DISC aural and CAS) and the AT would continue to maintain thrust to maintain the target speed. If the sensed IAS acceleration was away from the target and trending slower than the target, AT would just continue to keep adding N1 up to a maximum of the current thrust limit.

A TO/GA level off at or below acceleration height on takeoff sets the MCP speed to takeoff flaps placard speed minus 5kts, and continues to increase the speed command to the next flaps placard speed as flaps are retracted. VNAV does not activate as the AFCS vertical mode if armed until 400 feet AGL, likewise the thrust limit does not reduce from TO to CLB until either 400 or a flaps retraction (can be either to 5 or 1 depending on the operator or FMC inputs on the CDU TAKEOFF REF page).

Per the unreliable airspeed tables at max weight, 270KIAS, max climb thrust, deck angle would be 10 degrees with a 2800 fpm climb. At or near V2 even with a CLB-2 derate I would expect an even healthier vertical speed.

For all those reasons the "oops I set 200 instead of 2000 in the MCP alt window" theory isn't necessarily feeling like a smoking gun to me.

 

[Minuteman]

THR REF on almost all Boeing uses the primary thrust-setting parameter to match the value displayed on the EICAS (for the Frankenboeing substitute “N1” and “primary engine display” because it doesn’t have an EICAS). Its mode of operation does not change with airspeed, vertical speed, or pitch mode.

For a typical takeoff, throttles are advanced to an intermediate-low value to stabilize off-idle, when the TOGA switch is mashed. This does several things, including activating the auto flight pitch mode in TO/GA and the A/T in THR REF.

At 80 knots during takeoff further A/T commands are inhibited as HOLD mode activates. Manual movement of the throttles will still change the thrust setting, but that will not be done by the A/T. Except on the 787 if speed protection activates. If airspeed decays into the amber band (min oper speed) and the trend is headed towards the stick shaker speed; then it will exit HOLD into THR REF (presumably only adding thrust).

Generally any VNAV climb will use THR REF, except for intermediate level-off constraints where it will track a target speed. This is roughly-equivalent to Airbus Managed modes except, in addition to the FMGC and FCU, the throttle lever detents also have a say in what should be happening. It is usually most-efficient to climb with the engines operating at their maximum thermal efficiency (discounting derates for a different kind of efficiency).

It is appropriate to think that an autothrottle’s job is to track a target; either a speed or a thrust setting.

I would not expect an incorrectly-set MCP altitude to result in the behavior seen before the crash.

 

[JDean3204]

For most (all?) AT implementations the control loop is just comparing acceleration towards or away from the target IAS. It doesn't do anything like "well I'm supposed to be leveled off so I'll just set the power to what would be needed for level flight".

In the case of a low level off the AT would produce whatever thrust was necessary to maintain the target speed. If the pilot pulled back on the yoke at that point, the AP would have disconnected (with accompanying warning AUTOPILOT DISC aural and CAS) and the AT would continue to maintain thrust to maintain the target speed. If the sensed IAS acceleration was away from the target and trending slower than the target, AT would just continue to keep adding N1 up to a maximum of the current thrust limit.

A TO/GA level off at or below acceleration height on takeoff sets the MCP speed to takeoff flaps placard speed minus 5kts, and continues to increase the speed command to the next flaps placard speed as flaps are retracted. VNAV does not activate as the AFCS vertical mode if armed until 400 feet AGL, likewise the thrust limit does not reduce from TO to CLB until either 400 or a flaps retraction (can be either to 5 or 1 depending on the operator or FMC inputs on the CDU TAKEOFF REF page).

Per the unreliable airspeed tables at max weight, 270KIAS, max climb thrust, deck angle would be 10 degrees with a 2800 fpm climb. At or near V2 even with a CLB-2 derate I would expect an even healthier vertical speed.

For all those reasons the "oops I set 200 instead of 2000 in the MCP alt window" theory isn't necessarily feeling like a smoking gun to me.

Excellent post!