BREAKING: Boeing 767 cargo jet operated by Atlas Air has crashed in Texas

[3WE]

True, just about 26 seconds from the point where the FO first touched the yoke to the ground.


Engines go from idle to TOGA, spool up and a rush of thrust kicks in with a lot of longitudinal acceleration involved, which pushes you back against the seat back, just as if you were tilted nose-up. That is the most basic and common somatogravuic illusion (confusing longitudinal acceleration with pitch) and is the cause, for example, of the well documented series of accidents where the pilots take off into total darkness (like onto a large body of water in a moonless night or into very low IMC), push down and crash.

The slight pitch up made by the AP when transitioning into the go-around possibly gave a rotational cue to the inner ear that they were pitching up, which was exaggerated by the brain when combined with the somatogravic illusion explained.

At this point the FO thought that they were aiming way too high and when he pushed forward on the column the plane entered a very low G state which feels like falling from the sky. The FO's brain then possibly combined his exaggerated nose-up perception with the falling sensation to fabricate the "stall" mental picture.

When the guy says “we’re stalling” it all kind of fits.

Also of note- the FIRST descent seems fast and the level off and little climb seems a tad aggressive which would make my flawed mind think towards a an accelerated, stall-prone pull up.

And, as Evan says, maybe this is combined with fatigued thinking and performance.

[Schwartz]

Outside the obvious solution of not allowing pilots who panic (i.e. start taking random actions when faced with unknowns) to fly, I get the impression there are a couple basic ergonomic problems.

It sounds like the control layout is a bit problematic in that it's easy to hit the TOGA switch unintentionally when dealing with speed flaps. I can imagine other scenarios where a sudden unexpected acceleration to maximum thrust might cause some safety issues.

The second question is if there is a clear indicator that the TOGA switch/button has been activated or do the pilots have to notice the thrust levers moving? That also could prevent a scenario like this at least allowing the other pilot to diagnose the situation quicker. It sounds like the captain didn't figure out the problem right away either.

[Evan]

It sounds like the control layout is a bit problematic in that it's easy to hit the TOGA switch unintentionally when dealing with speed flaps.

The go levers seem to be a bit vulnerable. I wonder how often they are accidentally activated. I still favor a system such as that used by airbus where the thrust levers must be manually advanced to the go-around 'click' to activate the mode and then set to the desired power.

[3WE]

TOGA

Sorry, I'm too lazy to read between the lines.

How does accidentally hitting "TOGA" lead to someone initiating a low-altitude dive?

I won't say there's zero influence here, but I can't see how it is a primary contributing cause (unless you are making it primary in the time-sense)...but still primary in the time sense does not excuse relentless, deliberate dive inputs.

If the plane is trying to go around, but you are thinking "What's it doing now?", your control input should be to keep FDH attitudes and speeds, not initiating a dive.

One other lazy question (Gabieeee)- was there some "chirp" of a stall warning in this sequence?

Still flabbergasted...relentless push overs are about as bad as relentless pull ups...Heck, we even have the relentless right roll input on another thread (albeit a more valid situation, even though it's quite the testimony to the effects of relentlessness).

[Evan]

Sorry, I'm too lazy to read...

Read posts 102 and 104. Unexpected forward acceleration can cause the somatogravic illusion of pitching upward. The PF was known to have panicked reactions to unexpected events. This plus fatigue. It's more of a question to ask why the PIC didn't intervene and take control but it happens much more quickly than I first realized. They did recover to about 15 ANU but the ground didn't care about that.

[kent olsen]

The NTSB did state that the First Officer was easily flustered as noted by his two previous employers and Atlas. 20 or 30 years ago when I was involved with hiring the FAA changed things and you could not send a recommendation for employment to someones next job offer. We furloughed and hired off and on and I used to send a recommendation along with some of my furloughed pilots until the FAA said you can't do that anymore. That was a mistake as I remember a few times when we interviewed someone with nothing from their previous employer and when I called that Chief Pilot was told that pilot was very weak. After the FAA's change we did find some pilots we couldn't get thru Initial Operating Experience.

[Evan]

The NTSB did state that the First Officer was easily flustered as noted by his two previous employers and Atlas. 20 or 30 years ago when I was involved with hiring the FAA changed things and you could not send a recommendation for employment to someones next job offer.

Why on earth not? Piloting is certainly a job that needs full disclosure of weaknesses.

[KGEG]

My dad who was a long haul truck driver from about 1995-2012 and then he got out of it and had to do 2 jobs to make ends meet until he got the one he wanted. One of them working at Office Depot unloading the truck and stocking in the morning and helping customers during the day. The job he wanted was a bus driver for our local public service and when he applied for them the second time one of his bosses at Office Depot sent the bus company a letter of recommendation of him being a hard worker and good with customers, as being a public bus driver deals with even more people so its an important skill to have in it. He got the job and the letter the boss sent helped.

This was from a office supply store to bus driver.

Yet its not allowed to write a letter or email and say "Hey this guy doesn't know his landing gear knob from his throttle" or "He spaces out and easily loses his bearings and can be a safety hazard". Or vice versa, "This pilot displays an A+ skill set and an A+ work ethic and has not had any safety incidents in his time here".

Yeah that makes perfect sense.

[3WE]

Read posts 102 and 104. Unexpected forward acceleration can cause the somatogravic illusion of pitching upward. The PF was known to have panicked reactions to unexpected events. This plus fatigue. It's more of a question to ask why the PIC didn't intervene and take control but it happens much more quickly than I first realized. They did recover to about 15 ANU but the ground didn't care about that.

Ok, then I do not agree that the controls need to be redesigned due to the risk of panicky pilots being lead to do steep dives due to the location of the TOGA button.

I believe it is better to train pilots on fundamentals like robust speeds and attitudes and measured control inputs, even when you get scared.

And we gotta do something about this fatigue problem.

https://www.ntsb.gov/investigations/...ts/RAR1402.pdf

[3WE]

***Yet its not allowed to write a letter or email and say "Hey this guy doesn't know his landing gear knob from his throttle" or "He spaces out and easily loses his bearings and can be a safety hazard". Or vice versa, "This pilot displays an A+ skill set and an A+ work ethic and has not had any safety incidents in his time here".

Yeah that makes perfect sense.

Careful, we are not allowed to make allegations that over-zealous legal actions sometimes lead to logic defying policy.

[Evan]

Ok, then I do not agree that the controls need to be redesigned due to the risk of panicky pilots being lead to do steep dives due to the location of the TOGA button.

I don't think it would be overkill to add some aural alert on go-around mode activation. Like, "GO-AROUND". A mere software update. The counter-argument is "competent pilots don't need that!" Right, just like they don't need "TOO LOW GEAR"...

[3WE]

I don't think it would be overkill to add some aural alert on go-around mode activation. Like, "GO-AROUND". A mere software update. The counter-argument is "competent pilots don't need that!" Right, just like they don't need "TOO LOW GEAR"...

Ok, sure...nice to confirm that it is indeed selected...Still, I’m thinking that that obscure attitude, airspeed and power might be of at least some value...but I’m sure he was well versed in acronym checklists that you like so much.

if you aren’t paying attention to attitude, will TOGA really change your thinking?

[Gabriel]

When the guy says “we’re stalling” it all kind of fits.

Also of note- the FIRST descent seems fast and the level off and little climb seems a tad aggressive which would make my flawed mind think towards a an accelerated, stall-prone pull up.

Nobody will ever know what was going on in the mind of this FO, but I don't agree.

The steady descent was about 2400 fpm which is a tad aggressive as you said (they had the spoilers extended), and it peaked at I estimate 1000 to 1500 fpm during the go-around. The peak occurred about 10 seconds after the GA started, so it is an average rate fo change of some 350 fpm per second which is equivalent to 0.2G, again, average.

The vertical acceleration (load factor) plot shows that they were flying FDH at a smooth 1G until about 5 seconds before the TOGA activation when the hit turbulence and the load factor fluctuated between 0.85 and 1.2 G, with an average of still 1G. After go-around activation the average load factor steadily increased from 1G to I say about 1.2 to 1.3Gs, hard to tell because the trend of the average is surrounded by turbulence "noise". A peak in load factor, combination of pull up and turbulence, reached about 1.6G.

Finally, the pitch maneuver itself was very smooth. You can see in the video how the yoke barely moves back (moved bu the AP at this stage), although there may have been some nose-up trim added by the AP too, no seen either in the video nor in the FDR plots available so far. But the pitch smoothly changes from 0 deg to about 5 deg ANU in about 4 to 5 seconds, so it is a quite moderate pitch rate of about 1 degree per second (a smooth take off rotation is 2.5 to 3 degrees per second).

The AoA, compared with the "steady state" AoA during the descent, started to fluctuate +/- 1.5 degrees due to turbulence some 5 seconds before the TOGA activation and peaked +3 degrees (measured from the steady state) during the go around (combination of pull-up + turbulence). Very measured variations in AoA. Furthermore, these variations are from the "steady state AoA during the descent", whic had a baseline of negative 5.5 degrees. SO the angle of attack remained effectively negative throughout the pull-up caused by the TOGA. Although, of course, the pilots don't have a means to directly measure the AoA so they would not have known all these values and variations.

On top of that, throughout the "hot" part of the pull-up the FO didn't react. He reacted seconds later.

But, since they were flying at 235 knots with some flats and flaps already extended (so they were like 100 knots faster than stall speed) and the pitch remained very much in a very moderate range with very measured and smooth changes during the pull-up, I don't see a rationale here for a mental picture of an accelerated stall.

But who said that the perception had to be rational? Evidently the go-around took them by surprise and not only they didn't expect it, but once it happened they didn't recognized it, so it was very likely a "what is it doing now" situation. And we know from reports of past operators that the FO reacted irrationally and inconsistently to those situations, "he would just start to push buttons and do things just to be doing something".

Now, what really makes me think that it was not a reaction to a pull-up-induced accelerated stall is not so much how the FO actions don't correlate se well to all of the above, but how it DOES correlate with something else.

The FO doesn't push down during strongest part of the pull up maneuver. He pushes down when the engines spool up and thrust and longitudinal acceleration kick in.

From the NTSB "Aircraft Performance Study" in the public docket released months ago:

APPARENT PITCH/SPATIAL DISORIENTATION

Figure 8 contains the “apparent” pitch angle that the co-pilot might have “felt” that the airplane was at based solely on vestibular/kinesthetic perception from loads acting in the co-pilot’s body coordinate system1. A common pilot misperception during low visibility take-offs and go-arounds is that the nose of the airplane is excessively high. This is due to the large longitudinal accelerations, nx, associated with the high engine thrust settings. (A peak longitudinal acceleration of 0.27g’s was obtained at 1238:42 as shown in Figure 8.) Spatial disorientation can occur when the human body cannot distinguish between the force due to gravity at high pitch angles and the inertial force that results from accelerating the airplane. If the airplane is not flown by reference to the instruments, a natural reaction is to lower the nose in response to the increased acceleration and the misleading perception.

Below I am attaching figure 8 that show real and perceived pitch together with longitudinal acceleration. You can find this figure and others (showing load factor, pitch, AoA, etc...) in theAvHerald article, from where I took all the info I analyzed above: http://avherald.com/h?article=4c497c3c/0001&opt=0

One final thought:It amazes me that they were below 1G for almost 20 second, including 10 seconds tht they were almost permanently below zero G and reaching negative 1 G. If you are at zero G or lower you are not stalled, period. You can be flying the 767 at 60 knots and you would not be stalled if you are at zero G or lower, because you are not extracting any lift from the wings and that is because the AoA is very low. Even if, for whatever reason, you think you are stalled (and no, the stall warning which in the 767 is the stickshaker never activated), it makes no sense to be at zero or negative G and keep pushing down to recover from the stall. This is proof that the FO was acting totally irrationally and possible in panic, without really seeking an objective with his actions but just "doing something", as the previous 3 operators reported he would do in stressful situations. Finally... What was the Captain doing? He was pulling up without transfer of command, so in fact he was fighting with the FO for the control column, which caused the split elevator. I think that 2 words would have stopped all this craziness very early and saved the day: "My plane".

[Gabriel]

The second question is if there is a clear indicator that the TOGA switch/button has been activated or do the pilots have to notice the thrust levers moving? That also could prevent a scenario like this at least allowing the other pilot to diagnose the situation quicker. It sounds like the captain didn't figure out the problem right away either.

Yes, there are several cues an indications. The thrust levers move forward, the "N1 target" goes to the max, Ni, N2, EPR and Fuel Flow all start to increase, sound of engine increases, thrust increases, you can feel the acceleration, the aircraft pitches up, the speed increase, the vertical speed increases...

But the most important one... There is an instrument called PDF, PRIMARY FLIGHT DISPLAY. It was given that name for a reason, you know, it is the PRIMARY DISPLAYthat the pilots use to FLY the airplane, the one that should take the bulk of the attention of the PILOT FLYING and that the PILOT MONITORING should MONITOR.frequently. It contains the most important information the pilots need to fly the plane, some of which we mentioned above: Attitude (artificial horizon, i.e. roll and bank), airspeed, altitude, vertical speed...

It aslo contains a thing called FMA, FLIGHT MODE ANNOUNCIATOR. It has 3 boxed that show what the automation is doing, one box for the vertical channel (pitch, altitude, vertical speed), one for the horizontal channel (heading, bank) and one for the autothtrottle. The mode that each channel is doing will be shown inside each box (like VS -2000 // HDG 090 // 230KT). When one of these modes changes, the new mode will be displayed but also, for some seconds, the new mode will either flash, have a frame drown around it, or have a different color, to call the attention that the mode has changed. In this case, when Go Around was activated, all 3 boxes would have flashed GA.

[Gabriel]

If the plane is trying to go around, but you are thinking "What's it doing now?", your control input should be to keep FDH attitudes and speeds, not initiating a dive.

100% nailed it. First priority is to stabilize the flight path. Never ask "what is it doing now". Click-click, clack-clack, stabilize the flight path, and then, when you have everything under control and the plane is doing what you want, ask "why was it doing that".

The question is how do we train pilots so that they consistently do that and filter out those that are incapable of doing so?

One other lazy question (Gabieeee)- was there some "chirp" of a stall warning in this sequence?

Nope. There were several chirps, beeps, sirens... I don't know what they were, but none of them was a stall warning which in the 767 is a stickshaker which, it is confirmed now, never activated, not even for a split second.