Editorial: Maturing Safety Safely

Someone once remarked that the way to tell if a technology is mature is to observe whether each new solution creates an additional set of problems. Take automobile airbags, for example. They were developed to address a specific safety issue, which is that large numbers of drivers and passengers were being injured or killed in survivable accidents because they would not wear their seatbelts. The great thing about an airbag, advocates pointed out, was that it was a “passive restraint system”—drivers and passengers didn’t have to do anything to be protected, which was not the case with seatbelts and shoulder harnesses. Generally speaking, airbags have performed well and it is unarguable that they have saved several thousand lives.

However, it is also the case that they have been directly responsible for the deaths of a small but not statistically insignificant number of people in accidents that were not otherwise life-threatening. Furthermore, some percentage of drivers and passengers have been hurt or killed in crashes that were survivable had they been buckled up, but they believed the presence of the passive restraint system obviated the need to wear a seatbelt, which is a superior safety device in almost all accident situations. Automotive safety technology, we would argue, is mature.

Now consider the accident that is on everyone’s mind: Air France 447. At the end of May, BEA released its initial findings based on analysis of the FDR and CVR. These indicate that aircraft and engine systems apparently functioned as intended during the final minutes of flight. The autopilot and autothrottles disengaged—as they were designed to do—following a brief but meaningful discrepancy in airspeed indications, which is assumed to be owing to pitot tube icing.  That initiated a sequence of cockpit actions that led to the aircraft crashing less than five minutes later, having entered a stall from which it never was recovered. As in the Colgan Air accident that also occurred in 1999, it seems that the AF cockpit crew was faced with a serious but non-emergency situation, yet it quickly developed into a crisis with the loss of the aircraft and all aboard.

Modern aircraft are a monument to human design and intelligence. One need only look at the aviation accident trend, particularly over the past decade, to see the enormous positive impact highly automated aircraft have had on airline safety as they have achieved critical mass in the fleet. According to IATA, the industry accident rate in 2010 was the best ever, with one accident for every 1.6 million flights, a 42% improvement since 2000.

But it also has been known for some time that automation brings its own challenges and that within the long trend of improvement has arisen a subset of accidents in which issues of what some call “man/machine interface” have been crucial. “What is it doing now?” has become a common question among pilots since introduction of the flight management computer, and the recommended solution for pilots uncertain of the answer has been to “turn off the computer” and rely on their training and flying skills. This assumes that the crew has the skills and training to hand-fly an aircraft in all situations, including at night, in turbulence and in clouds—something that recent accidents and incidents have called into question.

While the industry waits on BEA’s final report, much can be done. Regulators must give more attention to the problem of pitot tube icing. We also would suggest that autopilot logic is due for a review, particularly what circumstances/events should result in uncommanded disengagement and whether some kind of alert sequence is desirable.

If BEA finds that AF447 was recoverable through most of the stall, attention must turn to training for stall recovery, which already is under review owing to the Colgan accident. As ATWcontributing writer Jerome Greer Chandler reported recently (ATW,4/11, p. 37), current stall training could be more accurately described as “approach to stall training.” Furthermore, conventional wisdom is to minimize altitude loss, with little or no emphasis on controlling angle of attack. With loss-of-control accidents now the leading killer in aviation, according to the Flight Safety Foundation, it is clear this is an area overdue for some fresh thinking.

Beyond all this, however, we think that the industry needs to recognize that the problems and challenges of a mature industry are different. Fundamentally: How do pilots maintain hand-flying skills if they are not required to use them regularly and if designers create aircraft in which hand-flying is no longer desirable or even practical in non-emergency situations?