Columnists

November 21, 2013

Do Commercial Pilots Really “Suck” at Manual Flying?

Auixer Cockpit Look Sideways
“Pilots Rely Too Much on Automation,” trumpets a Wall Street Journal headline this week. Gizmodo.com puts it more bluntly: “Commercial Pilots Suck at Manually Flying Planes.”

A 277-page FAA report is the culprit referred to in these venerable—and not so venerable—publications (Gizmodo, after all, is mostly known for such scoops as, “How much would it cost to build the Starship Enterprise”, and “How much would it cost to to build the Death Star.”) With all due respect to the WSJ, the sensationalistic, knee-jerk media seem to be having a field day with this “bombshell.”

I think we can fairly ask: How true are these headlines, really? As a Captain for a major U.S. airline, on one of the most automated airplanes in history—the Airbus A320—I believe I can speak with authority on this issue. Indeed, I have flown the A320 class (A321 through A319) for nearly 20 years.

The exhaustive FAA report, to be discussed with industry leaders later this month, analyzed some 3,000 commercial flights worldwide, and apparently comes to the disturbing conclusion that the greatest danger in the sky today is the automated cockpit. Ironic, since the whole goal of automation in the cockpit has been to improve safety. The study appears to single out two main issues: over-reliance on automation, and inadequate training on increasingly sophisticated systems.

And sophisticated, the Airbus is. From the first few seconds after takeoff to the last few minutes before landing, nearly every pilot leaves the flying to Fifi (our affectionate name for the French-built Airbus, despite her automated male voice). Why? Because, when it comes to airliners, flying an airplane is nothing. Safely managing a flight is everything.

Courtesy: Ismael Jorda

Courtesy: Ismael Jorda

The moment I punch the FCU and leave the driving to Fifi, my mind’s tendency to tunnel vision suddenly expands to see the “Big Picture.” Rather than be distracted with the minutiae of keeping this speed, that heading and climbing to that altitude, my attention is freed up considerably by letting her do the dirty work. This situational awareness is critical to the safe handling of an airliner.

Moreover, when the sh*t hits the fan blades, I am a strong proponent of letting Fifi do the driving while my First Officer and I troubleshoot. To be sure, somebody is always flying the plane—that is, babysitting Fifi. In fact, our emergency procedures call for the Captain to run the checklists while delegating the relatively straightforward task of flying to the First Officer.

Major international A330 airline pilot, blogger and novelist Karlene Petitt, author of Flight for Control, says, “This is a new world we face—a battle between automation and proficiency. The real question is, How will we win this war without losing thousands of lives? Encouraging pilots to hand fly their planes is a great idea. Our training footprints must include rigorous sessions where pilots perform engine failures and emergencies without the autopilot, flight director and autothrust. But I have to ask, Do you want your pilot hand flying an approach to minimums after being awake for 14 hours on the back side of the clock in a state of exhaustion?” I agree. Under normal flight conditions, the autopilot greatly reduces pilot fatigue. After a long day of multiple legs, or international, back side of the clock flying, it is quite a treat—let alone safer—to let Fifi shoot the ILS through the snowstorm and autoland. Is this over-reliance on automation? I hardly think so.

Conversely, on occasion I do get myself into a situation where I have to “turn off the magic and fly.” Captain Mark L. Berry, major airline pilot, author of 13,760 Feet–My Personal Hole in the Sky, and Contributing Editor for Airways magazine, says, “When I taught the Boeing 767, one of the original ‘electric jets,’ I noticed that old-school pilots were quick to disconnect the then-modern automation and hand fly the aircraft. At our core, we are first and foremost pilots, not ‘automation engineers’.”

af447

Again, I wholeheartedly agree. Nothing like a little old-fashioned stick and rudder to straighten the kinks out. Apparently, the FAA study has found that today’s pilots are reluctant to do this, stating they “lack sufficient or in-depth knowledge and skills.” Which leads us to our second issue: inadequate training on increasingly evolving automation.Two of the first Airbus accidents, the 1988 Air France Flight 296, an air show flyby-turned-PR-disaster, and Indian Airlines Fight 605 in 1990, as well as the more recent Air France Flight 447 crash on June 1, 2009, were all attributed—rightly or wrongly—to pilot error due to inadequate training and knowledge of the Airbus systems.

Airlines, I believe, do an adequate job of training today’s modern pilot. There is no getting around the proverbial “drinking from a firehose” that airline pilots are subjected to while training on a new airplane. And, in all the aircraft I’ve flown, the Airbus has far and away the biggest firehose from which to drown. The payoff, however, is worth every bit of saturation.

So,then, is the FAA report full of hot air? Is the greatest danger in the sky today the automated cockpit or not? Well, danger is a mighty loaded term. Is it more “dangerous” to land with your fly open or closed? What the study apparently fails to mention is the vast improvement of air safety with the advent of these highly sophisticated systems. Aircraft and aviation technology have evolved to the point where human beings are by far the weakest link in the safety chain. Ironically, pilots also remain its greatest asset.

But again, pilots are human. Automation or no, we make mistakes—on every flight. Even this report admits that the vast majority of observed errors were minor, and trapped by the pilots themselves before they could become big errors. And, as I mentioned earlier, when the magic is turned on, pilots can better see the Big Picture and more easily trap those errors.

As for the report’s remedies, however, I mostly agree.The first issue boils down to complacency. Pilots can become over-reliant on automation, get bored and forget to monitor. This is a constant human factor challenge for any pilot. As for the second issue, well, more systems and procedures training is always a good idea. Recently, in the “box,” we simulated the Air France 447 situation, and believe me, things get mighty confusing, mighty fast. Those pilots were fighting erroneous and conflicting instrument indications that quickly compounded. The lesson, again: stick and rudder. Turn off the magic and fly. That training, I assure you, was worth its weight in gold.

Dr. Martin Luther King, Jr. said, “The arc of the moral universe is long, but it bends toward justice.” Well, the arc of aviation safety is long as well, but it too has inexorably bent in the proper direction—to the point where one is 73 times more likely to die in an auto accident. To that end, what is most misleading about the FAA report are the sensationalized headlines surrounding it. Written by nonflying laymen who scream, “The sky is falling!,” they are oblivious to the fact that pilots—and their automated cockpits—have helped air travel become six times safer than your own bathtub.

Eric “Cap’n Aux” Auxier is an airline pilot by day, writer by night, and kid by choice. Pilot for a major U.S. airline, he is a freelance writer, novelist and blogger. He is a regular contributor to Airways Magazine.
His second novel,
The Last Bush Pilots captured the coveted Amazon TOP 100 Breakthrough Novels, 2013. Mr. Auxier makes his home in Phoenix, Arizona.


  • Bettina Ybarvia

    Bravo! Awesome post and very revealing! I consider the FAA findings interesting. But, rather than statistical numbers, nothing compares to a first hand feedback/report from the very subject itself—-the PILOT. I appreciate the coverage on both subjects, automation and manual operation. A well balanced report and a fair detail on its highlights and vantage points. What I’ve gathered here is a humbled admittance and acceptance that automation truly has its benefits. Yet not undermining your confidence in human experience, training and knowledge of which still greatly accounts for a safe and successful operation. Call it “conceit” Mr. “Jonas” but it is what I want to hear. It gives me the peace of mind that we have competent pilots flying our skies out there.

    Thank you for this post. I am learning that I need to look deeper into any study done by the FAA before taking it as the “ULTIMATE WORD”. To be suspicious about other reports before deeming it credible or merely”sensationalism”. Mr. Jonas, I guess, SUCKED in spelling and used the word “distraction” instead of “sensationalism”. Yet to discover or trust the “auto-correct” feature of your computer program Mr. Jonas? Pure arrogance and a total lack of understanding, comprehension his part. Yet Mr. Jonas would rather aim his ire towards your credible pilot “friends” ( thank you btw, glad to know you’re not the only one) because he wanted a link to the FAA report. Who needs to google now? There are so many flaws to JONAS’ comment, but I refuse to add anymore pain to his failed dream of being a pilot. And if in case he is..Please God..I hope he is not flying my plane!!” Go read your horoscope or something. You’re like a little boy who didn’t get his “quote for the day”. Here’s one for you, YEAH BABY by Austin Powers. That’s as far as I could describe the level of your comment post. Pure bullying zero intellect!

    • Eric Auxier

      Thank you, Bettina, for your comments.

      Jonas has his right to speak and make his points, though I am still usure as to what exactly they are. That is ok. I welcome the dialogue.

      Not sure I can work in an Austin Powers quote in my future articles, but I’ll try my darndest–yeah, baby! ;-)

      Thanks again for commenting.

  • JosephGuindi

    I think an interesting question is to determine the place of greater aircraft automation in pilot training. Automation is increasing in all areas of life as a means to increase safety at the cost of lower human agency.

    Take the calculator: it effortlessly calculates without error, requires only sunlight to operate, and is extremely fast. But does that mean we should teach students to use calculators instead of learning their tables? In primary school, we were forbidden from using calculators. In high school, we were forbidden from using scientific calculators, which would prevent you from cheating during calculus. In university, we were forbidden from using matrix-enabled calculators when solving a finite element problem. All of this helped me develop a “feel” for numbers if and when I made a mistake using a calculator, so that when I’m designing a mission-critical system that I would notice in my result that I forgot the decimal point.

    I like to think that it’s important to teach how the fundamentals work. But then again, an aircraft is a hell of a lot more complicated than a calculator. Is it best to focus on doing the basics, or to focus on proper trouble-shooting skills and flight physics such that the AF447 pilots would realise they were causing their own stall and leave the stickwork to the computer?

    • Eric Auxier

      Joseph,

      This is a fantastic analogy to the basic point that automation aids, but does not supplant, the human mind in the cockpit. Thank you for sharing! Believe me, I will be “borrowing” your analogy in any future articles about automation!

      I absolutely agree, one needs (in ANY field) a “feel” for the basics of operations. I cringe every time I offer up $5.02 to pay for my $4.77 burger, and the McDonald’s cashier looks at me like I’ve given him a moon rock! I just sigh and tell him, “Type in $5.02″…wa la! $.25 change! I can never tell if the light bulb even goes off! YES, we need fundamentals before punching those automatons!

      At the risk of further inciting Jonas’ ire, here’s a link to a blog post I wrote entitled, “The Future of U.S. Aviation,” that addresses our automation dilemma. One subsection is humorously entitled, “Automation, or, ‘Nothing Can go wrong…click! Go wrong…click! Go wrong…’”

      http://capnaux.blogspot.com/2013/07/happy-4th-future-of-us-aviation.html

      Thank you so much for your refreshing insight, Joseph!

      As for your AF447 question, I shall yield the floor to our resident expert, Bill Palmer, should he choose to chime in.

      Thank you again!

      Eric

  • Keith Mendoza

    From a software engineer with 10 years experience, and a private pilot, I completely understand why the FAA pointed out that pilots needs to keep up on their manual flying. Let’s admit it, computer software are only as good as the input they receive. Any software engineer worth their salt knows that one cannot think of every possible scenario that the software must be ready to handle, so you put a “catch-all” which is to inevitably hand it back to the human operator. Considering that there are programmers out there who cannot explain how a portion of software they wrote indented to handle a specific use case suddenly just started working, it’s safe to say that if that section would stop working that there’s no “exception handler” to recover from it if some invalid input is given to it.

    Let me ask you a few questions: are autopilot capable of recovering if the aircraft somehow gets flipped upside down by the wake vortex of the aircraft that went past that’s landing on the parallel runway? Assuming that the aircraft has enough control authority to recover, what would a pilot whose never had any aerobatics exposure (no, having seen the Blue Angels or Thunderbirds perform doesn’t count) do instinctively in that situation? What about if the pilot on that aircraft has won a few IAC unlimited category aerobatics contest?

    I ask these questions because if I were the software engineer tasked with handling upset recovery for an autopilot that would be the types of questions that I would be asking. I only know to ask these questions because I happen to have gone to a flight school where spin recovery training is mandatory before you can solo. I also have been fortunate to have had some aerobatics training, so I know how to execute a loop, barrel roll, wing-over, and spins. If the software engineer that doesn’t even know what “control authority” means how can you guarantee that they will even ask the right questions?

    • Eric Auxier

      Keith,
      I apologize for the tardy reply. The thread was “fizzling,” so I have not checked up on it in some time.

      I can only answer for my aircraft, the A320. However, the basic concept is, as you pointed out earlier, that “catch-all”, or “fail-safe” that is built in. At least initially, we have had “upset recovery” training as well. We do have it occasionally in Recurrent training, but not often. Perhaps an area we could beef up?

      In a nutshell, the autopilot does indeed kick off (past specific parameters, such as 75° bank, etc.), handing it back to–what did you call it?–the “exception handler” to recover (a new great name for the auotpilot-babysitting flight crew!)

      As for the design of this, I am wondering just how much “code” really goes into it. Rather than training a software engineer in spin recovery, it seems there are simpler, mechanical (i.e., laws of physics-defined) parameters built in. In Normal Law (normal flight in the Airbus), the stick controls, believe it or not, g-force. Again, when it exceeds straightforward parameters–speed, attitude, etc.–things start kicking off (grossly oversimplifying here), and you have a more “direct” stick-to-control relationship–in other words, more control authority to better recover from an upset. This same direct control phases in for the flare and landing.

      I always loved teaching my students spins, but have yet to execute loops, barrel rolls and wing-overs. Aerobatics is most definitely on my Bucket List!

      Thanks so much for the comments and questions, Keith!

  • ewrcap

    Thank you for the fine article. Somehow your name reminds me of another great writer with a French name, St. Exupéry!

    I wanted to add that the AF 447 accident was not really related to the other failures to monitor such as Asiana 214. It was more a matter of lack of situational awareness. Long before fly-by-wire and highly automated flight control systems there were several crashes due to pressure instruments. NWA 6231 in 1973, Birgenair 301 and Aeroperú 603 in 1996 all resulted in fatal loss of control.

    I disagree with the assumption that training for recovery from low speed situations is more important that training crews to constantly monitor and verify. Teaching a student pilot to do great spin recoveries is useless if he fails to monitor his airspeed and attitude while buzzing his friends house. In the case of Asiana, the pilot ultimately DID attempt a stall recovery. The plane hit just as max thrust kicked in. It was just too late. Once again, not an inability to execute a stall recovery, but a failure of all three crew to monitor airspeed.

    • Eric Auxier

      ewrcap
      Thank you for your comments and input!

      I will not turn down an opportunity to be associated with the great Antoine de St. Exupéry!

      I agree with you on the AF 447. Perhaps Captain Palmer can chime in here, but I do have my suspicions that the erroneous data was possibly overriding pilot input. It’s simply hard to believe experienced pilots would ride a stalled airplane all the way down without trying a simple stall recovery.

      Again I agree, monitoring and verifying are a constant human factor challenge to instill and continue to enforce, for hours of boredom on the line. As captains, we have some of that responsibility to do so.

      Thanks for bringing up some great points!

  • http://www.rapp.org/ Ron Rapp

    If there was one thing I’d like to see more of in airline/corporate recurrent training, it’d be a discussion of our decision making processes about when to use the automation and when to turn it off and hand-fly.

    The days of hand-flying these kinds of airplanes all the time are long gone, but we also must maintain sufficient manual flying skill for those times when the automation either fails or doesn’t do what we want it to. What I’m realizing is that this balance is not easy to maintain. We’re left to our own devices (no pun intended) and from what I’ve seen, we often underestimate how our skills have degraded.

    –Ron

    • Eric Auxier

      Ron,
      100% agree! And I think those discussions are coming. Nice to see you here!

  • Peter Son

    Your information about automation testing is really
    interesting. Also I want to know the latest new techniques which are
    implemented in software. Can you update it in your website?

    Software
    testing training