20 Seconds to Save It: How an Impulsive Pilot Caused a Fatal Crash Revised

hello everyone, paul bertorelli from avweb, just as I was editing the video you're about to see, I received an email from a Boeing 767

pilot

friend of mine who told me a story very similar to the one we're about to cover here. . He was the pnf, that's the

pilot

who is not flying or monitoring the pilot and while flying on an approach to a major international airport he heard bad words from the first officer followed by a pitching excursion that attracted attention, he looked down for five

seconds

to make a frequency change when suddenly the plane's automation did exactly what someone told it to do but which was nevertheless unwanted when it re-entered the loop, the fo had adjusted the desired selector and flattened the things.

My friend said it took him about 20

seconds

to figure things out and by then it was all over and that kind of thing probably happens a dozen times a day or at least it happened before greed 19 stopped flying airlines. The point is that automation in airplanes is a great thing, but with Homo the sap still in the circuit can cause problems with the machines. Assisted Chaos The accident we will review here is not an automation accident, it is a lack of piloting skills. Accident aggravated by misuse of automation. This week, the NTSB held a hearing into its investigation into the

crash

of Atlas Air 3591 on a flight from Miami to Houston. on February 23, 2019. three crew members aboard that plane died atlas air is a cargo airline and is a contracted carrier for amazon the flight was carrying cargo to houston bush airport got straight to the point the summary is this the plane

crash

ed in Trinity Bay 40 miles southeast of Houston, while skirting convective weather and some turbulence, crashed into shallow bay after entering a 46-degree nose-down dive that accelerated from 225 knots to 433 knots in 30 seconds, reaching a maximum vertical speed of 30,000 feet per minute this is how the ntsb reconstructed it we will show a short animation depicting the sequence of events please note that the representations are not necessarily identical to the aircraft screens in This still frame you can see a simulated external view of the aircraft The analog airspeed indicator with digits to aid clarity The main attitude director indicator includes an airspeed tape display and an artificial horizon and here is a representation of the flight mode annunciator display, an analog altimeter also with digits added for clarity, the position of the control yoke, the thrust levers, the position of the speed brake lever, a profile graph. and selected elements of the cockpit voice recorder transcript, this animation may be annoying to some viewers who may wish to leave the screen blank for about 90 seconds.

The crew was setting up the approach procedure when the plane encountered light turbulence shortly after, the autopilot and autothrottles entering orbit. As the airplane passed approximately 6,300 feet, there were no calls from the flight crew consistent with activation of the go-around mode, the airplane stopped its descent and began a slight pitch, the thrust levers advanced and the speed brakes were applied. They withdrew, which could only be done. The captain manually responded to a routine radio call while the first officer pushed the column forward and made an expression about speed and exclaimed, "We're stalling." There was no indication that the plane had actually stalled.

It was likely that the first officer was experiencing disorientation due to the somatographic illusion in which the acceleration of the aircraft results in a false nose-high sensation, the aircraft achieved a pronounced nose-down attitude and high speed below approximately three thousand feet , the plane came out of the clouds, the controls were moved to the maximum nose up, but it was too late. Before impacting the bay, the NTSB was unrelenting in its analysis and concluded that the first officer inadvertently activated the plane's roll mode and then, due to spatial disorientation, misinterpreted the plane's attitude as approaching a stall and ordered The captain abruptly lowered his nose and was slow to respond. and by the time he managed to counter the first officer's tone, it was too late to recover the plane, the plane was never close to a stall condition.

Much of the hearing was spent discussing the first officer's weak piloting skills and how he was able to hide an unfavorable employment history from Atlas when he was hired. More on that later, but first, clarification of the terms. Transport airplanes equipped with autopilots, that is, basically all have what is called a go-around mode, even light aircraft autopilots sometimes have this capability on the 767. The switch for go-around mode is on the throttles and when commanded, the airplane gently pitches about four or five degrees and the autothrottles advance to produce a climb of 2,000 feet per minute. It is not an abrupt maneuver.

The ntsb believes the first officer accidentally crashed. back button because he probably had his hand on the speed brake control which is located on the captain's side of the cockpit pedestal, requires an awkward reach and due to altitude turbulence, the first officer's hand may have hit the button, at least that's the theory. The veteran 767 training captain, I know, told me that he doesn't see how this could happen and never saw it in decades of training, but regardless of how it happened, it's pretty clear that it happened, although the back-and-forth maneuver is not aggressive, the acceleration may have been enough. to induce a kind of spatial disorientation called a somatographic illusion you probably remember this from your instrument training and perhaps even basic fighting training the human sense of balance is controlled by the inner ear vestibular system three semicircular canals detect accelerations in pitch, yaw and swing within the vestibular labyrinth. two structures called otolith organs that detect movements in the vertical and horizontal planes, such as the movement of the head from side to side or from front to back, when an airplane tilts under visual conditions, the pilot perceives it appropriately in a similar way if the Otolith organ is subject to level acceleration.

External visual reference will confirm that there is no pitching, but under instrument conditions the brain may interpret the acceleration level as a moment of pitching, possibly fast. This is a somatographic illusion and may have led the first officer to mistakenly believe that the aircraft was approaching the stall angle. of the attack is how david lawrence of the ntsb described it because the first officer was not effectively scanning his instruments when interpreting the information they provided, he experienced sensations that led him to incorrectly conclude that the plane was stalling while the plane was not It was in stall condition the first time.

The officer's aggressive nose-down movements and his inability to disengage autopilot and autothrus were contrary to his training and the atlas air intakes, both for a stall recovery and an unintentional automation change, the First officer should have disengaged the automation and manually returned the aircraft to position. Instead, the original first officer profile continued to make manual inputs that overrode the autopilot and forced the plane into a steep dive that was unrecoverable. The first officer's apparent struggles with

impulsive

action during training scenarios at multiple employers suggest that he had an inability to remain calm during stressful situations. This trace from the aircraft's flight data recorder shows how aggressively the first officer reacted.

This blue line represents the pitch the green line is the airspeed and the purple line is the altitude the time is on the lower scale from the start of the unnoticed turn here where the The pitch increase begins to impact was approximately 37 seconds. Turning power was reduced, but not significantly. Pitch recovery begins 22 seconds after the turnaround order, when the captain began to reverse the first officer's nose-down input, but it was too late, the aircraft impacted pitch-down approximately. 16 degrees and at 433 knots, the 767 has, by the way, what is called a split-path elevator, which means that the left and right elevators operate independently of each other, but the climb command has a little less authority than the descent command, not that it would have mattered much. due to the captain's late response and the ntsb found that the captain's response was a factor in causing the accident from the audience again here is david lawrence during the descent while the captain was acting as pilot monitoring and preparing for the approach and communicating with the air traffic control's attention was diverted from monitoring the status of the aircraft and verifying that the aircraft was proceeding as planned, this delayed his recognition and response to the first officer's unexpected actions of placing the aircraft into a dive as pilot in command, of which The captain is directly responsible and is the final authority over the operation of that aircraft despite having demonstrated an effective transfer of control several minutes earlier, when the first officer had his display problem, the captain did not assume positive control away from the first officer. official and stopped the planes in a decent manner and instead started performing manual operations. control inputs simultaneously with the first officer overriding autopilot until impact when he heard the ntsb criticize the captain's failure to use proper transfer of control.

Pilots are trained to verbalize the transfer of control of the aircraft using some version of the phrase "I have." control or you are in control, in this case the captain did not do that and simply entered commands with the nose up without informing the first officer that he was assuming control; However, just a few minutes earlier, the first officer had a minor problem with the display of him and the two pilots. he exchanged control with appropriate verbal notification while the first officer corrected the problem. It is also worth noting that during the descent 40 miles from the airport, the first officer called for flaps that extended the leading edge slats but not the aft flaps, this is against atlas procedures. and they told me that it is unusual to select floors that are far from the airport, in addition, the flaps allow the round trip mode to be selected with the flaps retracted.

Round trip mode is inhibited in the audience. ntsb vice president bruce landsberg asked how much time the captain had. to intervene after the descent and recover the airplane, do we have any estimate of how long the captain might have had to intervene before this became unrecoverable and an altitude or something like that that they might have been able to successfully recover? I will leave it to Mr. English, yes, Vice President Landsberg, although we do not know precisely the point at which the plane could be recoverable, from the beginning of the anomaly in which the turn and turn mode was activated until the En the point at which the airplane had not yet deviated much from that 63,6000-foot altitude, the captain had about 20 seconds or so to first notice the inappropriate mode change and leveling, which was an opportunity to intervene before The plane will take off. to get extreme and then if you remember where the throttles were coming and going, that was about 20 seconds into the event and at that point the plane had not yet started the very steep descent and there was probably some time. after that, still a correct intervention could have changed by about 20 seconds, they probably would have been able to recover maybe a little more, but that is a good estimate, yes, okay, what happened in this accident is not difficult to understand, but why? what happened is much harder to understand the bottom line is that a qualified and trained crew flew a perfectly airworthy plane into a swampy bay but the ntsb was upfront about the first officer's lack of skill here's the human factors expert of the ntsb, dr.

William Bramble the First Officer's performance and training demonstrated that when given enough practice he was capable of performing highly procedural actions, but he became overwhelmed when he was faced with novel, complex or unexpected situations. Atlas instructors attributed his difficulties to external circumstances or low confidence, however, his long history of significant performance difficulties. It is indicative of low aeronautical aptitude to mitigate the risk of hiring pilots with inadequate characteristics. Airlines need a focussystematic, science-based approach to pilot selection that assesses relevant pilot characteristics and collects improved data on pilot performance to validate selection measures and establish appropriate cut scores from which airlines would benefit.

A collaborative effort that allows them to gather data and share best practices in this area. Staff has proposed a recommendation that the FAA establish a pilot selection data clearinghouse to help operators improve and validate pilot selection strategies. pilots the first officer's lack of fitness according to the ntsb It was very clear that he had worked for six airlines and had failed the oral bus ride check or line check at four of them. ntsb president robert sumwalt asked staff member david lawrence about this captain lawrence. I want to ask you a question here. I looked at some.

From the comments from Mesa Airlines and Atlas regarding the first officer and I want to see if you see a trend here at Mesa, the Czech aviator said that when the first officer was presented with something unexpected in the simulator, the first officer was extremely nervous and couldn't respond appropriately to the situation when he didn't know what to do, he would get extremely anxious and start pressing a lot of buttons without thinking about what he was pressing just to be doing something and an atlas. When he tore up his first check right there, the examiner told the NTSB that when the first officer realized he needed to do something, he often did something inappropriate like pressing the wrong button.

Here are two different airlines talking about the same person and they are using Almost exactly the same verbiage, would you say that this behavior that was described on each of these airlines was seen on this accident flight? We saw the same similarities and, in fact, it wasn't just Atlas, it was uh Air Wisconsin, a Czech airman from Air Wisconsin had the same, identical concerns, the Czech airman who gave him his refresher training in Mesa said the same thing as Atlas, so that there was a consistency in that this pilot tended to do things in a stressful situation, he would just do whatever and start pressing buttons. and it was consistent across all these airlines, so how did it get this far?

Hiring processes of all kinds consisted of a series of filters to rule out unsuitable candidates. Airline pilots have to walk the walk with a simulator check before hire and even after hire. They have a trial period and careful monitoring through IOE or initial operating experience, as required by the FAA. Atlas has an internal quality control system called proficiency monitoring. It is designed to identify problem pilots for remedial training if this first officer ran into trouble on Atlas. He was not placed in the proficiency monitoring program according to the NTSB, but rather his operational experience period was extended.

The biggest problem according to the ntsb is that atlas did not know about all the first officer problems at the other airlines for two reasons, one being that it was not unusual for the airline to use a designated outside agent for the initial hiring review and That agency had no experience in aviation, but was basically a records retriever and reviewer and the records available to them under what is called the Pilot Records Improvement Act were limited and allowed the first officer to hide some of her flight history. employment under the recently proposed pilot records database program, the FAA is supposed to establish a system by which the FAA allows airlines to share pilot data, the NTSB could barely contain its impatience at the delay of the FAA in implementing this thanks. you for bringing up the points you just raised about the FAA not moving in a timely manner.

I intend to offer an amendment when we get to this point to add the FAA as a contributing factor to the FAA not implementing the pilot program. records database in a sufficient, sufficient and timely manner. I intend to offer that as an amendment when we get to that point, the pilot records database has drawn quite a bit of opposition from industry trade groups, mainly related to how the FAA wants to implement that and the ntsb of They actually added FAA inaction on the PRD as a factor in this accident, but the probable cause here should be obvious: it was the first officer's inadvertent activation of circling mode followed by an inappropriate response once contributing was the decision. of the captain. failure to monitor the aircraft situation and to adequately transfer control authority to resolve it before continuing a few words about the somatographic illusion, we will probably never know if that really happened here or if the first officer was too inept to handle the plane I looked at . 10 years of general aviation accidents and somatographic illusion are listed as causal or contributing 23 times in accidents, each one

fatal

again, these are the best guesses, since there is no way to know what was going on inside the pilot's head as instructors that we try to induce. spatial disorientation and training, my method is to put the pilot under the hood with his eyes closed and looking in his lap, then do a quick island maneuver or have the pilot do it and then recover with instruments in years of doing this, I haven't even managed it once. in creating somatographic illusion or even vertigo, that's not to say that I don't think it can happen, but that the way we train it doesn't seem to make it happen anyway, the cure is a good disciplined exploration of the instrument focused on the Attitude indicator If the nose is down and the airspeed is increasing, level the wings first and then pitch the nose up to level flight If the airspeed is decreasing, lower the nose first to gain some power and then level off the wings, this is not a complicated thing, but if you never do it or if you don't do it regularly in training, it could be a smoking crater in your future for avweb.

I'm Paul Bergarelli, thanks for watching.