AIN Blog: Shedding Light on Automation’s Dark Side

 - May 2, 2012, 4:47 PM
Debris from the missing Air France A330-200 recovered from the Atlantic Ocean arrives at the port of Recife, Brazil on June 14, 2009.
Debris from the missing Air France A330-200 recovered from the Atlantic Ocean arrives at the port of Recife, Brazil on June 14, 2009.

Like many pilots, Bill Voss is concerned about the extent to which automation has changed the role of the professional pilot today. But as president and CEO of the Flight Safety Foundation, Voss is also better placed than most to do something about the problems he perceives.

The losses of the Air France A330 and Colgan Q400 in 2009 cast doubt on the core stick-and-rudder skills of pilots flying modern commercial aircraft, some of them extremely automated. “We keep trying to pretend this snuck up on us but it had all the stealth of a freight train,” notes Voss. “Of course this was happening, and everyone knew it. But no one really talked about it until the Air France A330 crashed into the Atlantic. Now we’ve had that event, we have to have a serious conversation about stick-and-rudder skills.”

In one sense there are signs of progress already, notably at Emirates Airline, which has inserted two days of manual simulator flying into its pilots’ recurrency training. “This is an extraordinarily bold and expensive move—two days of sim is a big hunk of money—and other airlines are altering their automation policies to make sure there is more hands-on time,” says Voss. “But at the end of the day you still have the fundamental problem that the system is moving ever further away from one where pilots can fly the airplane.” Regulators expect operations in RVSM airspace to be flown on the autopilot, which takes away hands-on cruise flight from pilots in most parts of the world. “Add to that RNP or GNS procedures off the ground; continuous-descent approaches in the terminal area, which take you pretty much all the way down to final, flown coupled because they’re containment-based PBN apps; and we’re clearly going to a future where we can no longer pretend that the automation is there to help the pilot,” continues Voss. “The pilot is there as a backup to the automation. Those are two fundamentally different concepts, and they’re going to require us to fundamentally revisit our training.”

Voss regards hand flying as something that is relatively easy to recover, like riding a bike, “but what’s not easy is finding a way to keep pilots genuinely engaged in the operation of the aircraft so they know what’s happening next. You can’t monitor an automation system or step in to back it up by sitting there passively and just observing it; you have to have a clear idea of what’s supposed to happen next in the flight. If the pilots do have to intervene, they don’t necessarily have in their mind the combination of attitude and power setting necessary to sustain the aircraft for the next few minutes until things settle down. These are things that go beyond raw flying skills that we have to find a way to put back into pilots’ heads.”

The FSF is involved more in redefining recurrency training than ab initio, and the foundation is working on pulling together some of the many efforts currently under way. “But this is not a training problem,” Voss emphasizes. “The problem is that the operation of commercial aircraft has fundamentally changed. What you do every day when you fly 200 days a year has changed, and changing the curriculum of what you do two days a year during training is not going to fix it. You have to look at the entire system and find a way to reinforce the right behaviors 200 days a year, not just two. Those are the challenges we really have to strap on. It’s a fundamental change, and it’s time the world comes to grips with it.”

Comments

Lukas 's picture

Bill Voss raises some valid points but I disagree that "this is not a training problem"

The foundations of the building determine it's resilience. Those 2 days per year training are mere maintenance of that structure. Initial training is fundamental to being able to easily recover manual handling skills like riding a bike. In the case of Colgan and AF447 those bike riding skills were NOT evident!

Had the Capt or FO of AF447 been at the controls would the result have been the same?

If the Colgan Captain and Fo had under gone thorough training (or been failed after 2 attempts) would that accident have happened?

In both cases the individuals at the controls had minimal hands on flying before they joined a company with an automated airliner.

What was the experience level of the captain, when he applied for
employment with Colgan Air? 618 hours, Gulfstream Training Academy, failed initial instrument
rating, failed single engine-land rating, failed multi engine-land rating,
graded “unsatisfactory” on two simulator sessions at GTA covering:
approach to stall-landing configuration, unacceptable altitude and
airspeed control, with repeated deviations.

What training had the AF447 SO had? 250h JAR course then straight into an A320!

This is now common in the EU and experienced pilots can no longer get interviews with the likes of easyJet and Ryan Air because low timers are cheaper:
See column 2:
http://www.chirp.co.uk/Downloads/ATFB/ATFB101.pdf

Papadakis's picture

I covered 3 very different time frames in commercial part 121 operations.
In 1970 the Old Captains were truly hand flying professionals who could fly machinery and who distrusted automation.

The 1980's-a990s had a generation of Captains that watched automation evolve and they had a very healthy skepticism. They had been encouraged to hand fly as they gained seniority in the first officer seat.

The newer generation relies and believes in Automation, (they have little skepticism)and they have much less hand flying time.

The best solution is to have highly trained monitors to watch automation during weather operations and trained and capable of flying the machine.

The best advice I got from a seasoned check airmen was fly the airplane every chance you can. Remember when the autoflight system shoots craps is usually in a fog storm. Sort of an IMC murphy's law.

The software computer program engineers have little concept of the real world and and less of real operations. When and if they see anything it is in a laboratory, a flat screen, or at best a simulator.

Show comments (2)