The crash of Atlas Air 3591 shocked the airline industry. It was caused by a first officer with a long history of poor performance that wasn’t known when he was hired. And an experienced captain in the left seat wasn’t enough to save the airplane or the three lives aboard. In this video, AVweb’s Paul Bertorelli offers a detailed dissection of the accident based on the NTSB sunshine hearing. This report was first published in 2019.

26 COMMENTS

  2. The first officer was a weak pilot. He was passed from airline to airline, job to job, with no one really saying the harsh truth–he should NEVER have been in that left seat. He should have been “washed out” for good, but that’s hard to do. It’s not like the US Air Force, where a pilot either graduates or washes out. He just got fired, left it off. his resume.

    For example, I noticed that even this video didn’t mention that he worked at CommutAir. They mentioned Mesa several times, but so what? He worked at what, seven flying jobs?

    The video should have shown a picture of him, shown his resume, and then shown his REAL “resume,” all the places he worked, how long, why did he get fired. Was it during ground school, or in the sim phase, or on IOE, or while flying the line?

    “Affirmative action.” Big time. Everyone is so scared of being labeled a “racist.” So, to REALLY criticize a Black person, as this first officer was, is hard to do. If he was white, dude would have been kicked to the curb a long time ago. But he wasn’t.

    Women, minorities–if they’re weak, they still get pilot jobs.

    This is reality.

  3. The somatographic illusion is REALLY hard to show in a Cessna 172, maybe impossible. I tried, as a CFI, no way.

    I learned about the somatographic illusion in the US Air Force as a student pilot in the T-38. Holy moley, it’s as real as cancer. The IP (instructor pilot) would have you close your eyes–no cheating!–at about 250 KIAS, then he’d go to full afterburner and accelerate to about 450 knots, and have you open your eyes and take control of the aircraft.

    While accelerating, with your eyes closed, it feels like you’re “going ballistic, Mav!” It really feels like you’re climbing in a nose-high attitude.

    When you open your eyes and see the ADI and you’re wings-level, it’s an eye-opener, for sure.

    I BET this guy never got that kinda training, so the first time HE felt that somatographic illusion, BAM, he reacted incorrectly.

    I feel like it’s really “not his fault” that he was in that right seat–he NEVER should have gotten that far. The people who put him there, all the people who “gave him a break,” were “nice guys,” etc. are to blame.

    That dude did the best he could, which wasn’t up to standards, but was “carried” by a lot of people. Dang it anyway.

  4. I did some training in a CAP Cessna 182 with a Garmin G1000 that was equipped with a “go around” button. I thought it was a recipe for disaster for an inexperienced pilot since it only set the FMS and autopilot. It didn’t automatically change the throttle or flap settings. In my opinion that would have led to a stall on a go around since it pitched the aircraft up but did not add power or reduce drag without pilot input. I shared my concerns with the instructor but he really couldn’t do anything about it since CAP had decided to equip many of the 182s in their fleet with this button.

    There wasn’t any discussion of how much training in automatic go around mode the FO had received in this accident. Most of the GA fleet doesn’t have this capability so it’s hard to train on it. It’s logical to infer that most GA instructors aren’t familiar with it either since it’s so rare. If an aircraft is equipped with this feature, then it should be mandatory that the instructor be familiar with it and train student pilots thoroughly in it’s operation.

    The NTSB focused on the pilot’s poor performance history, which is important, but they didn’t dive deeply into his training history according to what was presented in the report. I think this accident could have been prevented if the FO had been more thoroughly trained in the use of the automated go around.

  5. What I don’t understand is why there is no database at the FAA accessible to potential aviation employers to record a pilot’s full performance and record on flight tests, recurrent training at carriers, and a full resume of all aviation-related positions held.
    I don’t think this is a personal privacy issue, but a safety issue. Carriers and other aviation companies should know exactly who they’re hiring and their full record. It would help to weed out weak and incompetent pilots.
    Perhaps if something like that existed, this accident could have been avoided. This would completely avoid what Mr. Pilot above brought up (true or not).

  6. The captain could have saved the flight when the F/O called for Flaps 1 at 6,000’ and 30 miles from the airport. That action was completely and totally inappropriate for the aircraft’s position, altitude and ATC clearance. The captain should have disallowed it. Unfortunately, extending flaps to 1 also armed the autoflight Go-Around mode, but the captain could not have reasonably expected the outcome that it enabled.

    In the 767 (and 757), the autoflight Go-Around mode is armed under only two conditions: 1) The flaps move out of the UP position or 2) The autoflight captures the glideslope. Unless one of those conditions is met, the Go-Around switches can be repeatedly selected until the pilot’s thumb is blistered, but the Go-Around autoflight mode WILL NOT activate. Period.

    The only time extending flaps to 1 at 6,000’ and 30 miles from the airport would be appropriate is to comply with an ATC speed assignment below the aircraft’s clean maneuvering speed. That was not the case for this flight. In fact, ATC would not expect nor want an aircraft of that type to slow down that far from the airport without a speed assignment.

    This accident was caused by the first officer’s complete lack of aptitude for flying and a pilot hiring process that enabled him to hide his failures at every single flying job.

    • Hey Bill,

      You are correct about Flaps out arms GA.

      It also arms Vnav “approach mode”. Which allows the pilot to open the Speed intervene window and still be in Vnav path. I use this mode all the time. Really helps on a Rnav STAR and multiple verbal speed clearances given by ATC.

      My only thought was when an airplane does something you don’t like, why don’t more pilots turn off the autopilot and auto throttle and put the airplane where they want it?

      Just my 2 cents

      Ed

  7. For that matter the captain wasn’t ready for the left seat either. You never let things get so far. Did he just ‘veg out’, quit his scan and not notice the horizon line disappearing into the top of the AI? Did he just stand back and standby?
    It’s been my privilege to fly with four pilots I’d fly through the ‘Gates of Hell’ with. They each had their specialty. Hands down, ‘nobody does it better’ smooth as silk on instrument happened to be a black man.
    The FO in this crash had all the ratings, each passed by DPE.
    If the FAA don’t catch up with you eventually Darwin will.

  8. Irresponsive Captain’s – not the first time-
    Remember the Air France over the Atlantic – ughhhhh

  9. To Mr. Pilot (and Doug H):
    To suggest the cause of this crash was affirmative action for dark skinned pilots is really a stretch that you can’t cover.
    Next step on this stairway is that black people (and women?) can’t fly airplanes and that white man are discriminated against.
    I agree with washing out incompetent pilots and think that a national database might be a good idea. (surprised there isn’t one already) And training is key.

    • I’d ask if you are sincerely interested in the facts (and not just confirming their prior biases)to please have a look at NTSB docket DCA19MA086. There are hundreds of pages of FO’s long observed history of deficiency as a pilot. In Aug. 2017, Atlas spent an enormous effort trying to get FO to pass a 767 type rating after failing oral and checkride. By everyone’s at Atlas’ admissions (below) when asked under oath, no one at the airline could recall other pilots needing or ever receiving that extra assistance.

      I don’t know what happened. If someone feels adamantly that that it is irrationally ignorant and hateful to suggest possible preferential treatment in service of diversity in an industry with none might have put some level of pressure on the “standard” airline pilot selection process, would you mind reading the below factual short summery of the FO’s training history and offering a plausible theory for why so much was done to make an exception for this pilot?

      ATLAS TRAINING RECORD:
      brief excerpts:
      8/1/17 – “Regression and confusion on almost all PM (pilot monitoring) duties and normal callouts and procedures. Situational awareness appears to be decreasing, always reactive, rather than proactive. Knowledge is there but not always available at the proper time.”
      8/5/17 – “Difficult time completing normal procedures. Poor QRH (quick response handbook) accuracy and systems control.”
      8/31/17 “Regression situational awareness”
      9/22/17 “Situational awareness lacking. Procedures incorrect. Limitations exceeded. Judgement poor and CRM/TEM skills poor. Didn’t work well with PM during in-flight emergency. Coordination with cabin poor. Steep turns exceeded all PTS limit. Non-precision procedures non-existent. Lack of confidence and tunnel vision were definite contributing factors”
      9/25 – Check airmen Rob Bluemel: “RECOMMENDED FOR TYPE RATING”

      MESA AIR TRAINING RECORD: 5/1/17 – Failed captain upgrade check, company decided to let him return to being FO. Left without notice.
      TRANS STATES TRAINING RECORDS: Failed checks on 4/22/14, 5/11/14, 8/15/14. Resigned after failing line check
      AIR WISC. TRAINING RECORDS: (April to August 2012): Did not complete FO initial training for CRJ “personal reasons”.
      CommutAir TRAINING RECORDS: (May to June 2011): Did not complete FO initial training for de Havilland
      DHC-8 citing “lack of progress in training” resigned

      NTSB INTERVIEWS:

      The Atlas sim check pilot Robert Bluemel who in his words, gave hours of remedial training trying to “get the FO’s confidence back” and prep him to pass on another attempt after FO busted his 7-6 checkride, was asked by NTSB if this was common: “Oh, I can probably count on one hand over the 7 years the remedials I did. It was not very — what do I say? It wasn’t very common.”

      NTSB asks William Neal, 767 instructor/DE: “Just curious, since (FO) did have to go through some remedial training to get him through the type ride. Is that typical in your experience for first officers here at Atlas, to require remedial training to get through the training program?”
      And he answered “No. No, it’s not. I’m very proud of the training program that we all put together. And if a guy needs help, we usually identify it long before he gets to the pre-type flight evaluation. So, but the vast majority of our guys just progress through, very little remedial training done as far as I ever saw.”

      Atlas Aircrew program designee Billy Peavley on FO’s check ride to the NSB:
      “He was discombobulated from the very beginning for his 3 takeoff, things were going poorly. The FAA was observing and that may have had something to do with it…yet he seemed to not be able to do much of anything correctly… I wanted to see exactly where he was. I wanted to see if it was just nerves or is there was a real issue here and it came to present itself that he actually did… it was actually a problem with his skills and it became apparent after not too much longer”
      NTSB Q: “Is that typical for what you see of a new hire for first officers here, his performance?”
      A. “No”

      NTSB Q: “Okay. This may be self-explanatory, but also in there you graded him unsatisfactory in judgment. Can you talk about that?”
      Peavley A: “His judgment was faulty. If you elect to ignore a procedure, well, your judgment’s already faulty by definition…So things would start snowballing and he made questionable decisions in that way, rather than just slowing down. ”
      NTSB Q: “Okay. And you said this was not typical of the performance of a first officer, correct? Did I get that right?
      A: “Yes, that’s correct.”

      Interestingly, the Atlas re-check DE Mark Hoover told the board that FO’s first 7-6 checkride had the FAA to observe the extensive list of basic deficiencies, but when he passed the test a few days later, it was approved by only the Atlas DE ( no FAA observer present).
      ——

      On FO’s Atlas application for the question if he’d ever failed any proficiency check, FO said he’d only had to “redo one non precision approach” on his ATP, which is true, but only one of six failures I counted prior to Atlas. More concerning is the Atlas HR person wrote internal comments on FO’s resume in ink at each line of his flying experience with the truth from his PRD report: “unsat for type rating”, “unsat”, “proficiency issues”. So Atlas knew he had a history of deficiency and was misrepresenting his past failures but still hired him.

      When Atlas did hire him, he repeatedly could not perform to objective standards. Everyone at Atlas training could not seem to remember about anything about his deficiencies, and told NTSB to refer to the records, and they also swore there was no external pressure.
      But from the above interviews, since no one could remember anyone else getting that level of help to pass, when they were directly asked by NTSB, everyone at Atlas admitted under oath this never happened to anyone else.

      Again, I had no idea what race gender faith etc. the FO was until it was brought up(how did anyone know this?) , but I read not just a press clips but looked at hundreds of hours of info and interviews in the NTSB docket, and I am shocked that this guy was ever put in a transport aircraft.
      Based on docs, there was no ignorance at Atlas, everything about this pilot well known to them and prior employers. Someone at Atlas was making a decision to hire and then give remedial training to this pilot when he could not succeed, and to some familiar with industry hiring practices, these favorable decisions seem very unusual.

      Those above may be jumping to conclusions, however I never heard anyone say that black people can’t fly airplanes, so that seems an unfair straw man fallacy and not civil discourse. I am not sure how women aviators came into this since all aboard according to FAA were identified male. For both groups, I know great pilots, and I wince that we departed to such broad accusations. Again, I am offering no analysis, and invite folks to spend some time reading the docket. I think everyone can agree this should have never happened and something isn’t right.
      In any event, I am interested in what you think happened in this situation.

      • I appreciate your dispassionate analysis. I had the same reaction when I read the docket and followed the hearing. Like you, I have no convincing conclusion. It may very well be an example of diversity/affirmative action gone wrong. Or it may simply be the holes lining up in a display of sheer hiring incompetence or fear of litigation.

        It’s not a coincidence that we ran this video alongside Jenny Beatty’s blog on failing checkrides. (https://www.avweb.com/insider/overcoming-multiple-checkride-failures/)

        The reality is that not everyone can be taught to fly an airplane at all, much less well. The hiring system is pretty good at weeding these people out, but it’s not perfect.

      • Here’s how I found the FO’s race, gender: I Googled “First Officer Atlas Crash,” and bam, up came his picture, more than one. Al Gore’s amazing Internet.

        • As a retired professional corporate and airline pilot, Simulator I.P. and Pilot Examiner (TCE) on jet equipment I can say that the WORST example of poor piloting I ever witnessed was a Caucasian, and the BEST examples I’ve seen were Black, Hispanic, and Caucasian.
          Race has almost nothing to offer from the perspective of pilot evaluations.
          In every clients’ training-record there is a natural tendency to avoid committing to the permanent record denigrating commentary which cannot be rectified by an opportunity for the client to subsequently demonstrate proficiency. However, if a client continues to show lack of progress I would counsel, then report/recommend to the employer either 1) additional training (which should indicate sub-standard performance and cause the employer to re-evaluate their employee), or 2) make “UNSAT” remarks on the training record which prevents certification/re-certification.
          HOWEVER, the problem with some training facilities is when DIFFERENT I.P.s subsequently train/re-train the client. If weak performance is not properly graded and the next I.P. is lax or reluctant to issue “UNSAT” grades… a weak pilot can slip thru the system.
          This is further compromised when training record grading is limited to “check” or “UNSAT” or “P” where “P” is the only grade indicating Proficient. A generous I.P. may issue a “check” (mark) indicating the maneuver was experienced/demonstrated…but no proficiency requirement is met. This is not uncommon at many training programs. Training records should include acknowledgment signatures of both facility representatives and the client.
          There should be complete access to permanent recordation of all prior training in any/all commercial operations hiring practices and oversight by the FAA-PI for each operator.
          In THIS accident, it’s despicable that this F.O. was allowed to slip through the system by changing multiple employers.
          I’m curious if this crew had ever flown together prior to this accident…or if this was their first trip together. When I operated with a new-to-me crewmember I was very observant of the others’ piloting skills before I would allow myself any distractions from “pilot monitoring” duties.
          After considering the failures of the hiring and training history of this F.O…. the next-most important question in my mind is: WHY was this captain so slow to respond? Was this their first trip together? Did he “assume” too much with regard to the F.O.s’ skills simply because he wore epaulettes? Was a jump-seater a distraction to the captain? (I.E., was there extraneous conversation/activities/interactions with the jump seat?)

          • Affirmative action. Diversity. Fine ideas, right up until you (Atlas, in this case) lets a guy this weak, this bad, through training. Lets him get into a cockpit and fly. Lets him control an airplane. Oh, you might go on and on about how Caucausian pilots are bad, minority pilots are great. Well, maybe. But there certainly are MORE white dudes flying, aren’t there, so more of them will suck, number-wise. I’m just saying if you “carry” a weak pilot, whatever race or gender, it’s wrong as can be. Objective training standards, is what I’m saying, not “special treatment.” For guys like John McCain. Look up his accident in training. Read about how he crashed a Navy aircraft into the water during training. Fell asleep. I wonder why he graduated? Oh, affirmative action, sort of.

  10. jimbarber2125

    For whatever reason, the FO was given many “second chances”(remedials) in his training with Atlas, and his record with previous employers indicated a lack of aptitude for being an ATP level pilot. The main problem, as I see it, was that the FO covered up previous performance difficulties with other employers when he lied on his application to Atlas. This would have resulted in his termination had he been found out. For details see the Operational Factors section of the docket of the NTSB Accident Report. The interview of Denise Borrelli, Director of Human Resources, Atlas Air, on pages 534 through 595, is of particular interest. Occasionally, as quoted (NTSB Member Michael Graham) by Vice Chairman (NTSB) Landsberg, it was sometimes necessary to inform someone that for their own safety and that of the public, they needed to find a different career field.

  11. Typical light airplane don’t have enough performance to allow instructors to demonstrate somatographic illusions. They simply can’t accelerate fast enough to generate the illusion.

    What they can demonstrate is the insidious nature of the “leans” . All you have to do is get the student to close their eyes and describe what they think the airplane is doing. The instructor then smoothly and slowly rolls the airplane into a medium turn. Eventually the student will say the airplane is now flying straight and level. When they open their eyes they will be momentarily disoriented. It is a powerful demonstration of why you have to disregard your senses and trust the instruments.

    With respect to the demonstrated incompetence of the accident pilot, I would suggest the questioning should start with the instructor who signed them off for their commercial pilot flight test. In my opinion it would have been obvious that this person did not have the qualities necessary to be a successful professional pilot. The reality however is that anyone can eventually pass the flight test if they spend enough money. Flight schools and instructors generally won’t tell students they are unlikely to succeed as long as they are spending money at the school.

  12. Atlas delivered to the Captain a marginally qualified or wholly unqualified crew member that the Captain was ill-experienced to deal with. The Captain likely devoted critical time to first assessing and analyzing mechanical factors rather than believing human factors were the cause. All that said, I continue to believe that if standard and emergency “call-outs” (cockpit discipline) were rigidly adhered to, this accident may not have occurred.

  13. No one has mentioned the possibility of suicide. Look at FedEx flight flight 705. Just saying could that have been a potential factor.

  14. This is just one example, that if it weren’t for the accident, we wouldn’t even be reading about this. There are many more out there who shouldn’t be flying. Its not for everyone.

  15. When you put diversity and affirmative action ahead of skill and merit you always get less than superior top notch performance. All you have to do is look at Kamala. Affirmative action full circle.

  16. One thing that’s barely been mentioned here is the highly unusual practice of overpowering the autopilot. The FO shoved the stick forward and never disconnected the autopilot. Nor did he disconnect the autothrottle, which in a go-around is giving significant (not full) thrust. The yoke pressure required to overpower the autopilot would have been significant, but I suppose in times of pressure, the adrenaline rush could’ve made it feel lighter. Also… when you’re going fast (which ideally, one of the two pilots would have noticed), it’s logical to pull the throttles back to idle. It doesn’t seem that was done either.

LEAVE A REPLY