Accident investigations yield useful information. But is all this information actually being fed back to the system and acted upon?
The primary objective of accident investigations is to determine the causal factors and to use that information to prevent that type of accident from occurring again. However, the same types of accidents still occur.
Through no fault of the accident investigation agencies around the world, the industry is not doing a very good job of assimilating their findings into effective training examples in the classroom.
Although the sequence of factors leading up to an accident may be complex, the final triggering mechanism itself often is simple — such as taking off with ice on the wings or intentionally descending below landing minimums when a go-around should be conducted. In most cases, these triggering events can be ascribed to fundamental decision errors by the crew.
These are what I label thematic accidents. Four such thematic accidents, with almost identical probable causes, occurred over a 21-year period. The probable causes are extracted verbatim from the official accident reports.
The first of these four accidents occurred in 1987, the most recent in 2008. Each was attributable to deficiencies in checklist usage, adherence to standard operating procedures and cockpit discipline. Each crew failed to set the flaps/slats for takeoff and, in each case, the takeoff configuration warning system was inoperative for unknown reasons. If the warning systems had been functional, these accidents could have been prevented.
This shows how much trust we bestow on a defense that should warn of impending danger. Unfortunately, in each of the accidents, that defense was not available. Additionally, in the Northwest Flight 255 (1987) and Delta Air Lines Flight 1141 (1988) accidents, there were flagrant violations of the “sterile cockpit” rule. The Northwest pilots were chatting about non-flight-related items during taxi (in lieu of executing the proper “Taxi” and “Before Takeoff” checklists). In the Delta accident, the pilots and a flight attendant riding in the jumpseat were discussing the dating habits of flight attendants and — in reference to being recorded by the cockpit voice recorder — how they needed to leave something for their wives and children to listen to in case they died.
Why, after the first accident in 1987, did we not learn enough to prevent the same type of accident? In fact, it was just one year later that the almost identical Delta accident happened. It could be argued that, despite the shock factor of Northwest 255, the full investigation into that crash was still not complete. Then, it appears, from 1988–2005, there was a “latent period” for this type of accident.
Was it because of lessons learned? Maybe the significance of the Northwest and Delta accidents finally got the attention of global airlines — or maybe not; in 2005, the same accident occurred again (Mandala Airlines Flight 091), and again in 2008, with the crash of Spanair Flight 5022.
The Spanair accident occurred although there had been three almost identical accidents to learn from over the previous 21-year period. This was just one of numerous recurring accident themes that could have been chosen.1
True, major accidents of the past have been catalysts for important safety initiatives such as ground-proximity warning systems, smoke detectors and automatic fire extinguishers in lavatories and cargo holds, on-board wind shear detection equipment and crew resource management (CRM). But, while these initiatives have made a remarkable improvement in safety, we still need to shore up the human performance aspects of flight operations. Each of the aforementioned accidents was caused by a lapse in human performance.
The following recommendations are offered to overcome the apparent gap between the rich data available from accident reports and the effective assimilation of those data. The recommendations focus on the recurring accident theme highlighted in this article.
Air Traffic Control
Military air traffic controllers have long used the “check gear down” reminder for pilots of landing aircraft. This has prevented a number of gear-up accidents. The same type of reminder should be considered for civil aviation, particularly airline operations. Why not make it a requirement for air traffic controllers to add the phrase “check configuration” when the pilots receive their takeoff clearance? I would bet that this simple, additional safety net would have prevented most of the accidents mentioned.
Flight Attendant Awareness
Flight attendant training should include an increased awareness of misconfiguration issues. Because flight attendants still are walking through the aisles during the pre-takeoff cabin check, and the aircraft by this time should have flaps extended for takeoff, they are in an excellent position to detect a misconfiguration. However, it should be made clear to the flight attendants that not all aircraft require flaps to be extended for takeoff. Ensure that the information is aircraft-specific.
Focused Flight Crew Training
Although some links in the accident chain can be traced to the organizational level, the responsibility for prevention of these types of accidents still lies squarely on the flight crews, as they are the last line of defense. Thus, an approach consisting of more focused flight crew training and awareness is appropriate. All four accident examples occurred due to deficiencies in human performance — centered primarily on handling interruptions, sterile cockpit procedures and checklist usage — involving unprofessional behavior and lack of discipline.
Some of these deficiencies are externally propagated, or beyond the pilot’s control, such as interruptions, the effects of which can be addressed with good threat-and-error management skills. Other deficiencies may be internally propagated, for example, when crews violate the sterile cockpit rule. In this case, the pilots have full volition, and thus control, of their behaviors. Additional focus should be aimed at these types of internally propagated behaviors.
More Effective Use of Accident Reports
My final recommendation is to enhance learning by making more, and better, use of the rich data available from accident reports. Thematic accidents should receive special attention. This can be accomplished by using relevant case studies and crafting a learning module that not only stimulates the pilots’ attention, but also enhances retention. I have seen, and heard of, too many CRM courses that simply rehash the Tenerife runway disaster and/or the American Airlines crash near Cali, Colombia.
While in no way diminishing the importance of learning valuable lessons from these accidents, I believe that they have been studied to excess. We need to be more forward-thinking and focus on current accidents whose causes are more elusive. I am confident that CRM and threat-and-error management trainers can craft more effective learning modules that produce better retention and transfer to the real world. I wrote “learning modules” rather than “training modules” because the emphasis is on learning from other crews’ errors and misfortunes. We are not simply training to prevent accidents; we want to develop better critical thinking and error-avoidance skills.
Robert Baron, Ph.D., is the president and chief consultant of The Aviation Consulting Group. He has assisted a multitude of aviation organizations in the development of their human factors, safety management systems, CRM, and line-oriented safety audit training programs. Baron is also an adjunct assistant professor at Embry-Riddle Aeronautical University and teaches courses on aviation safety and human factors subjects.
The author gratefully acknowledges Kimberly Szathmary for her review and input.
Note
- Space does not allow for a discussion of the research related to the pre-departure and taxi phases of flight. See, however, the work of R. Key Dismukes and his colleagues at NASA Ames Research Center’s Flight Cognition Laboratory, humansystems.arc.nasa.gov/flightcognition. Dismukes et al. have conducted extensive studies related to, among others, checklist usage, interruptions, concurrent task demands, and prospective memory, each being highly relevant to all the accidents presented in this article. An increased understanding of these factors is imperative in preventing further accidents of these types.