How much automation is enough? When is the line crossed between having something that’s really useful and doing so much that, like the Wizard of Oz with his hot air balloon, we have no idea how it works? What role does training play?
In the wake of the Asiana accident where the crew flew what appeared to be a perfectly functioning B777 into a seawall, inquiring minds are now wondering if the auto-throttle system was perhaps a bit overdesigned. My only experience with auto-throttles have been in simulators—Gulfstream IV and the Airbus 330. It was pretty easy. Push a mode button here, select a speed there, and then monitor the system for anomalies. “Monitor” is a key word.
The ease of flying a fully automated approach makes you wonder just how hard this could be? In primary instrument training there are roughly five or six instruments to be scanned to keep the localizer and glideslope (or was it glideslop?) somewhere near center scale. It’s a lot of plates to keep spinning simultaneously. When first introduced to the horizontal situation indicator (HSI) and the single-cue flight director, they significantly lowered my estimation of how hard it was to fly bigger aircraft.
One of the NTSB’s Asiana findings was that auto-throttle modes were confusing. Former NTSB Chair Debbie Hersman noted, “This crew was extremely experienced. They had a lot of hours, but they just didn’t have the ability to understand what was happening in the critical few seconds before the crash. It wasn’t just one person in the cockpit that didn’t understand. There were three experienced people in the cockpit that didn’t understand what was going on.” Really? Didn’t have the ability to see that the aircraft was not stabilized early in the approach and failed to take action when slowing below Vref? Hmmm.
I know nothing about the B777 systems which may be complex, but with several hundred thousand (or more) successful automated landings completed over the aircraft’s 20 plus year history, ya gotta wonder just how bad the system design was. Ms. Hersman and I both agree that engineers often overthink and overdesign systems. We diverge somewhat on the statement regarding this crew’s capability. Experience and competence are two different attributes.
The other two airline accidents that bear at least some similarity with somewhat clueless crews are the Colgan Q400 accident in Buffalo and Air France 447, the Airbus 330 that stalled over the South Atlantic. In all cases, a highly automated aircraft wrested control away from the crew. (Shades of Space Odyssey and the HAL 9000 computer… “I’m sorry, Dave. I’m afraid I can’t do that.” )
Lest you think I’m just picking on air carrier pilots, many more GA pilots demonstrate a lack of clue far more frequently. Automation doesn’t seem to trap us often although it’s becoming more prevalent. It’s more often judgment. Lack of fuel, too much weather, too much wind, too little altitude, not enough runway—you know the list.