Listening to a recording of ValuJet Flight 592’s last moments, investigator J.F. Joseph kept hearing sounds the pilots hadn’t talked about.
“You could hear this noise clicking on and off like a pinball machine,” he said by phone from San Marcos, Texas. “That told us a lot about what was actually transpiring.”
The sound was electrical switches on the McDonnell Douglas DC-9 opening and closing as the crew tried to restore their controls, Joseph said. They’d stopped working because a fire in the cargo hold had burned through the wiring — a fact that became clear within seconds when the cockpit voice recorder picked up women in the cabin shouting “Fire, fire!”. Three minutes later, the plane plunged into the Florida Everglades; the May, 1996 crash killed all 110 people on board.
Almost 20 years later, those kinds of details will be part of the focus for officials probing the Dec. 28 crash of an AirAsia Bhd. jet carrying 162 people. Why Flight QZ8501 slammed into the sea off the Indonesian coast of Borneo island without a distress signal remains largely a mystery, with an experienced pilot at the controls of a tried and tested plane of an airline that had never had a fatal accident.
Flight recorders have “become the cornerstone of the investigation process, because the information provided is so rich,” said Geoff Dell, an associate professor specializing in air accidents at Central Queensland University in Bundaberg, Australia. “You can very quickly get a very accurate reconstruction of what the aircraft was doing.”
Flying at 32,000 feet (9,800 meters), the AirAsia pilot had asked to climb, citing clouds, Indonesian officials have said. By the time air traffic controllers responded about two minutes later, there was no reply. There were storms along the flight path, yet other planes safely traversed the area at around the same time.
Researchers working for Indonesia’s National Transportation Safety Committee have already identified one question they need answers to, according to Santoso Sayogo, a member of the investigation team.
“The tail was upside down” when the wreckage of the Airbus Group NV A320 was found on the sea floor, Sayogo said in an interview. It’s not clear whether that’s because of ocean currents or because the jet flipped as it tumbled into the water. If it’s the latter, it should show up in data tracking the plane’s angle as it fell.
Investigators will pore over each piece of debris and split into two teams for the “very labor-intensive” job of analyzing the flight data and as much as two hours of audio, Sayogo said. Those studying the flight data will start by focusing on 10 to 20 of the more than 300 information streams picked up by the device, he said.
“If a voice says, ’Watch out!’ then we will write it down, and if there’s a click sound we have to identify the click,” Sayogo said. “If the pilot or co-pilot changed an instrument in the cockpit, we will try to identify it.”
That could involve sitting in an A320 cockpit and flicking controls until they produce a noise matching the recording, or trawling through the captured flight data — which was downloaded successfully late Tuesday — to find the record of an action at the same time.
The recorders contain several dozen memory chips and are built to withstand immense shock and extended periods under water.
Investigators were able to retrieve information from the black boxes from Air France Flight 447, which crashed in June 2009, even after two years of exposure to salt water and intense pressure at the bottom of the Atlantic Ocean. Comparing the voice recording with the flight data showed the pilots seemed to be oblivious to the fact that the Airbus A330 had stalled.
Flight-data recording has advanced dramatically since 1954, when an Australian engineer, David Warren, developed the first such system in an attempt to understand a series of crashes involving the de Havilland Comet, an early commercial jet.
Until the 1990s even commercial aircraft would often measure just five types of information on their data recorders, using a needle scratching indentations into an aluminum-foil tape, according to Dell, the Australian professor. Investigators had little to go on beyond the plane’s altitude, heading, speed, acceleration, and whether the crew had switched on microphones.
Voice recorders helped flesh out this scanty information with pilots’ conversations and background noises.
That detail has often proved crucial. In the 1982 crash of an Air Florida plane at Washington Reagan Airport, investigators with the U.S. National Transportation Safety Board analyzed an engine hum in the background of the recording. The abnormal frequency helped them conclude that ice was causing the aircraft to lose thrust, ultimately sending it into the Potomac River, killing 74 of the 79 people on board.
Voice recordings arguably have become less important with digital upgrades to data recorders, enabling some models to log more than 1,000 streams of information such as fuel levels, instrument-warning lights and the position of cockpit controls.
If the flight data recorder is undamaged, it’s “almost no more difficult than shoving a USB in the side of your laptop,” Dell said by phone. “Within about an hour you would have a really, really good impression of what the airplane was doing.”
Investigators generally prepare a full transcript of the voice recordings, synchronizing every pilot comment or cockpit sound with the flight data, plotted separately as dozens of distinct graphs on a common time axis. Reading them is akin to following an orchestral score.
Inconsistent information in these graphs can be crucial, said Peter Marosszeky, a former air-crash investigator who lectures at the University of New South Wales in Australia.
In the report into the Air France crash, for example, faulty airspeed readings that led to the disaster show up clearly in diverging green and yellow lines in the printout from the data recorder.
With such a wealth of information at hand, theories about the crash can be constructed and tested to narrow in on the cause, Marosszeky said.
“An investigator will determine in their own mind roughly what will have happened, and use the flight data recorder and voice recorder to either support or refute that,” he said. “There’s very rarely a ‘Eureka!’ moment in these things.”
–With assistance from Anurag Kotoky in New Delhi, Andrea Rothman in Toulouse and Fathiya Dahrul in Jakarta.
This article was written by David Fickling, Alan Levin and Rieka Rahadiana from Bloomberg and was legally licensed through the NewsCred publisher network.