The 185 delegates from 52 states who participated in the 1944 Chicago Convention that created the International Civil Aviation Organization (ICAO) fully intended for the aviation community to investigate, learn and adapt from accidents so the deadly past would not be repeated.
Thanks to their vision in 1944 — and to the efforts of so many more — today’s global airline accident rate is at its lowest ever. The International Air Transport Association (IATA) reports that last year the global rate was one accident for every 1.6 million flights, a 42 percent improvement since 2000.
Three key areas helped get us to that low rate: data, technology and design.
Those Chicago Convention pioneers recognized the importance of data. From the beginning, ICAO was to be a center for the collection, study, and distribution of information on all significant aircraft accidents. This focus was essential. After all, between 1946 and 1950, on average, U.S. carriers had a major aviation accident every 16 days.
Early foil flight recorders, followed by their second- and third-generation descendants, contributed significantly to today’s outstanding global safety record. You could fill a book with lists of accidents solved thanks to information obtained from data recorders. Much of that information led directly to technological improvements, such as the enhanced ground proximity warning system (EGPWS), which have contributed immensely to aviation safety. EGPWS has all but solved controlled flight into terrain (CFIT) accidents. And, with Doppler radar and so much more, aircraft now fly more safely in all kinds of weather conditions. The traffic alert and collision avoidance system (TCAS) has helped prevent midair collisions.
As for design, in civil aviation’s first century, the community learned a tremendous amount — the hard way — about aircraft design issues from a number of accidents, including the McDonnell Douglas DC-10 with its poorly designed cargo door latches and the Boeing 737 and metal fatigue.
As we ended that first century, we saw further design improvements on the workhorses of the airline industry — remedies for the rudder design issues in the 737 and the flammability of the Boeing 747 center fuel tank. Today, we find fewer and fewer equipment and design failures.
While there is greater safety in civil aviation’s second century, there are greater challenges in investigating accidents and assuring safety. One reason is that while modern technology has made aircraft more efficient, they are also far more complex.
Old “steam” gauges have been replaced by electronic displays. Hand flying has been supplanted by increasing automation. Many flight controls now rely on electronic actuators rather than control cables. Also, there are more and more composite structures.
While these all provide advantages, they require adjusting how accident investigators acquire evidence and information. The evidence and failure signatures relied upon in yesterday’s investigations are not always available today.
For example, in 2001, when the National Transportation Safety Board (NTSB) pulled the vertical fin of the Airbus A300 out of New York’s Jamaica Bay, it took a long time to figure out how the failure began and why. This is because the vertical fin was a largely composite structure and the typical overstress signatures that were available with metal were not present.
The good news is that investigators have access to more data sources. Today’s flight recorders collect thousands of parameters. Investigators are also able to retrieve information from non-volatile memory sources, which can be recovered from electronic components, including digital engine controls, flight control and maintenance computers, and more.
Even when these devices are severely damaged, chip-level data extraction has successfully contributed to accident investigations. Data are also transmitted from onboard reporting systems, such as the aircraft communications addressing and reporting system (ACARS), which can provide investigators with critical real-time information. We are also seeing an immense amount of video data from surveillance cameras and personal cameras, as well as information from global positioning system (GPS) devices and electronic flight bags.
Yet, even with all the data sources, investigators continue to deal with the most complicated piece of equipment in aviation — the human — for which there is no data recorder.
Human factors accidents are harder to investigate, because often there is little evidence to document the decision-making process that led to the accident. Unlike airplanes that come off the assembly line designed to be exactly the same and perform to predictable and repeatable specifications, human beings are not always predictable.
There’s only so much data on the cockpit voice recorder (CVR), often the most scrutinized piece of equipment on an accident airplane. Investigators listen for inflections in the pilots’ voices — yawns, straining on the controls, and many other subtle changes in speech — to determine why pilots responded the way they did or did not respond as expected.
One of the most frustrating things investigators encounter is listening to a CVR and hearing a pilot say, “Look at that!” It can take years of painstaking effort to finally determine what “that” was and its relevance to the accident.
Adding to the complexity of accident investigation in aviation’s second century is aviation’s increasing globalization. There is no longer a clear distinction between domestic and international accidents. Accidents involving U.S. operators and U.S. equipment can and do occur anywhere in the world. Likewise, accidents may happen in the United States, but involve a foreign-operated or foreign-manufactured aircraft.
As we plan ahead, to prevent accidents in aviation’s second century, we must recognize the increasing importance of working together. The accident investigation framework provided by ICAO Annex 13 is crucial since it provides the foundation — the protocols, the rights and responsibilities — for the states to work together.
With globalization, accident investigation will depend far more on data and cooperation than in the past. While time honored tin-kicking will never go away, it is increasingly being joined by sophisticated data analysis.
The investigation of the Jan. 17, 2008, crash landing of a British Airways Boeing 777 illustrates the 21st century model of accident investigation and the importance of data and cooperation. This flight, which originated in Beijing, was on short-final approach at 720 ft above ground level when the right engine and then the left engine stopped responding to auto-throttle. Through outstanding airmanship, over busy roadways and dense population, the pilots brought the plane to land just beyond the perimeter fence at Heathrow. The U.K.’s Air Accidents Investigation Branch (AAIB) led the investigation, which the NTSB joined as an accredited representative.
The flight data recorder (FDR), CVR and quick access recorder were recovered; there were some 1,400 parameters on the data recorders. The pilots gave extensive interviews. None of this told the team precisely why both engines failed. Nor did tests of the fuel, of fuel water content, examining where the airplane was last serviced, and more. Everything came up blank.
Yet, with a rich store of data, the team reviewed thousands of similar flights. One key finding was that the accident plane flew longer at a low fuel flow in cold temperatures than other flights. Temperatures on the accident flight’s routing reached as cold as minus 74 degrees C.
This, in turn, led to scrutiny of fuel delivery to the engines. Lab tests looked at the effect of extreme cold temperatures and long idle times. Of particular interest was the fuel-oil heat exchanger, which uses cold fuel to take heat away from the oil and leads to the engine running cooler, especially the bearings.
The investigative team performed tests running a fuel system mockup for hours with cold fuel. They saw ice crystals collect on the face of the fuel-oil heat exchanger. If the engine throttle was applied, the newly formed ice broke up. But, with no throttle applied, the ice continued to form.
It turned out that this perfect flight — with minimal throttle usage to conserve fuel — led to slushy ice forming within the fuel system. When throttle was applied during the later stages of approach, the accumulated ice traveled to the fuel-oil heat exchanger and restricted the fuel flow.
Corrections included interim procedures that were followed by a redesign of the fuel-oil heat exchanger. Safety was served, which was enabled by data and cooperation.
In this era of dynamic growth and greater complexity, collecting and analyzing data are more important than ever. Accident investigators need all the data available to put together the big picture of what happened.
I applaud the agreement reached last year at the ICAO 37th Assembly to foster data sharing through the creation of the Global Safety Information Exchange. This information can be vital to learning what really happened and determining what can be done to improve safety. The recent General Assembly initiated an important dialogue about data sources. This is essential in setting standards of protection for the use of data in accident investigations.
Looking ahead, no matter how proud we are of the strong safety record the aviation community has achieved, we must not be complacent. We must make a constant commitment to further improve aviation safety by using data and further improving international cooperation.
This article is adapted from remarks made by NTSB Chairman Deborah A.P. Hersman at the 8th Annual Kotaite Lecture to the Montreal Branch of the Royal Aeronautical Society on Dec. 8, 2011.