Thursday, October 29, 2009

Automation-Induced Complacency

Image shamelessly stolen from NASA ASRS Callback #315

It would seem that you can't conduct an aviation-related blog without commenting on the saga of the peripatetic Northwest Flight 188. You know - the one where the flight crew reportedly went "Heads-Down-and-Locked" over their personal laptops while 'George' flew the airplane past the intended destination (KMSP) at FL370 and on for another 150 miles into the wilds of darkest Wisconsin. It was the Mother of all LOSA (Loss of Situational Awareness) incidents.

Now it's reported that the FAA has issued an emergency revocation order for the certificates of the Captain and First Officer involved. They can appeal the order to the NTSB but based on the circumstances reported in the press their prospects of continued employment as professional aviators appear bleak. Perhaps that's as it should be. It is certainly easy to imagine ways in which their inattention to duty could have led to catastrophe. But at the same time, I'd suggest that to a degree they were set up.

Based on what's been revealed to date, it looks like this flight crew fell victim to Automation-Induced Complacency. And this is not in any way a new problem. Consider the following quote:

"Another problem concerns the new automatic systems which are coming into service with newer aircraft and being added to older aircraft. Flightcrews become more reliant upon the functioning of sophisticated avionics systems, and their associated automation, to fly the airplane. This is increasingly so as the reliability of such equipment improves. Basic control of the aircraft and supervision of the flight's progress by instrument indications diminish as other more pressing tasks in the cockpit attract attention because of the overreliance on such automatic equipment.

Pilot's testimony indicated that dependence on the reliability and capability of the autopilot is actually greater than anticipated in its early design and its certification. This is particularly true in the cruise phase of flight....

In any event, good pilot practices and company training dictate that one pilot will monitor the progress of the aircraft at all times and under all circumstances."[emphasis added]

Those words come from an NTSB report dated 14 June 1973. The subject is the crash, on the preceding 29 December, of Eastern Airlines Flight 401 in the Florida Everglades. That was the L-1011 aboard which the entire crew was busy troubleshooting an indicator light problem while the airplane - with the autopilot inadvertantly disengaged - spiralled slowly down into the swamp.

And here we are, 36 years later, with another crew allowing themselves to be distracted to the point where no one was flying the airplane! Fortunately, this time nobody died.

"During the long ages between dawn and sunrise, I'm thankful we didn't make the Spirit of St. Louis a stable plane. The very instability which makes it difficult to fly blind or hold an accurate course at night now guards me against excessive errors. It's again a case of the plane and me compensating for each other."
-- Charles A. Lindbergh, "The Spirit of St. Louis"

Flying in the Lone Eagle's day was the polar opposite of today. The problems revolved around task saturation, i.e., there was too much to do. But aviators like Lindbergh and Doolittle and Rickenbacker nonetheless excelled. It was an environment where human beings could, with training, do well. Yet, as aviation matured, new capabilities of pilot and aircraft were needed. Necessity brought forth the autopilot.

"I pay those guys to fly, so let them fly. I'll be damned if I'll pay them to just sit there."
-- reportedly, Eddie Rickenbacker, CEO Eastern Airlines. Eastern aircraft were some of the last to be equipped with autopilots, his pilots saying if it wasn't in Captain Eddie's SPAD he won't buy it. Quoted in 'Human Factors in Multi-Crew Flight Operations' by Orlady & Orlady.

In spite of Capt. Eddie's resistance, flight systems automation has become ubiquitous and now it looks like problems are arising from crews having not enough to do. The terms in current vogue for the flight crew are PF and PM - that's Pilot Flying and Pilot Monitoring. But for long stretches in cruise there is precious little flying for the PF to do, and the PM is probably, like all humans, rather poor at monitoring.

In a 2002 paper entitled Enhancing Flight-crew Monitoring Skills Can Increase Flight Safety, Capts. Robert Sumwalt & Ronald Thomas and NASA's Dr. Key Dismukes point out that:

"...although monitoring may seem intuitive and easy, in reality, continuous and effective monitoring is not natural. In fact, it is most natural to do things as you think of them, instead of delaying them until later. However...an effective monitoring strategy involves sometimes delaying some tasks until less vulnerable periods. It is also not natural to stop doing something in the middle of the task to scan instruments (monitor). Instead, people prefer to complete the task before stopping. However, effective monitoring requires a more-or-less constant scan of instruments."

And, quoting an earlier Air Transport Assn. Automation Sub-Committee paper:

"...serious errors do not occur frequently which can lead to boredom and complacency. 'A low probability, high-criticality error is exactly the one that must be caught and corrected.'" [emphasis added]

And so we have this highly experienced flight crew on NWA188 whiling away the hours from KSAN to KMSP with 'George' doing all the work. Ostensibly, of course, they are "monitoring systems" but you can only devote so much attention to static displays and creeping moving maps. So, being human and fallible, they turn their attention to "more pressing matters" -- like the arcane new pilot scheduling software. And the share of their attention devoted to "monitoring" their progress through the night sky falls to...zero. It's just unlikely enough to be the truth.

I think of the times I've been humming along over eastern Pennsylvania with 'George' in control of N631S and the GPS in control of 'George' and me, both the PF and the PM, with no flying to do and not a lot to monitor. If there were a distraction available, I might be vulnerable. So, while I can't excuse the guys up front on NWA188, I can see how they got in trouble.

The airlines may want to look into ways to keep the flight crew engaged with the airplane during extended low-workload periods. There is, for example, the concept of Adaptive Automation which uses novel approaches to task allocation and task partitioning to be responsive to changes in workload and operator behavior. The system could deliberately cede tasks to the pilots during low-workload periods simply to maintain a minimum level of operator involvement.

For my part, I think I'll just continue to keep Rule 1 in mind: Always fly the airplane!

No comments: