Accident · ssh / airDisasterReports
Air France Flight 447 Crash Investigation Explained
Air France Flight 447 disappeared over the Atlantic in 2009. Nearly two years later, its flight recorders revealed how a temporary airspeed problem developed into a fatal high-altitude stall.
At about 02:10 UTC on 1 June 2009, Air France Flight 447 was cruising over the Atlantic at roughly 35,000 feet when its autopilot disconnected.
There had been no engine explosion or structural breakup. The Airbus A330-203, flying from Rio de Janeiro to Paris with 216 passengers and 12 crew members, had entered a chain of events that began with something far less dramatic: its airspeed measurements temporarily stopped agreeing.
Four minutes later, the aircraft hit the ocean.
Understanding why took almost two years of searching beneath the Atlantic.
The Trouble Began With Airspeed
Flight AF447 was crossing the Intertropical Convergence Zone, where large convective clouds are common.
France's aviation accident-investigation authority, the BEA, concluded that ice crystals temporarily obstructed the aircraft's Pitot probes, producing inconsistent measured airspeeds.
Pitot probes are pressure-sensing devices used as part of the system that determines an aircraft's speed through the air.
When the A330's computers detected unreliable speed information, the autopilot and autothrust disconnected and the flight-control system changed to a less-protective mode known as alternate law.
The pilots suddenly had to fly manually at high altitude while deciding which indications could be trusted.
The sensor problem itself was temporary.
What followed proved fatal.
The Aircraft Climbed Into a Stall
The pilot flying made nose-up control inputs. The aircraft climbed above its previous cruise altitude while losing speed, and a stall warning sounded.
An aerodynamic stall does not mean the engines have stopped.
It means the wing's angle of attack—the angle between the wing and the incoming airflow—has become too high for the airflow to remain properly attached. Lift falls and drag rises.
Recovery requires reducing that angle, normally by lowering the nose enough to let the wing fly normally again.
That did not happen consistently on AF447.
The BEA's final investigation found that the crew did not correctly connect the unreliable airspeed indications with the appropriate procedure, did not adequately recognise the deviation from the intended flight path, failed to identify the approach to the stall and ultimately failed to diagnose that the aircraft remained stalled.
Why Was the Stall So Difficult to Recognise?
This is what makes AF447 more instructive than a simple story about a failed sensor.
When the Pitot readings became inconsistent, automation dropped away, warnings appeared and displays changed. The pilots had to rapidly rebuild their understanding of what the aircraft was doing.
The BEA highlighted startle, confusion and weakened task-sharing among the crew. It also identified shortcomings in high-altitude manual-flying training and training for speed-indication anomalies as contributing factors.
Flight 447 carried three pilots because of the length of the journey. The captain had been taking a scheduled rest break when the problem began, leaving two co-pilots at the controls.
He returned during the emergency, but another experienced pilot entering the cockpit did not instantly solve the central problem: the crew still did not form a correct shared understanding that the aircraft was in a sustained stall.
That distinction is important.
The final investigation did not reduce the accident to one pilot making one wrong movement. It described an interaction between unreliable indications, automation changes, manual control inputs, training, human factors and cockpit coordination.
Then Flight 447 Disappeared
AF447 was far from land and outside conventional radar coverage.
Its automated ACARS system transmitted a sequence of maintenance messages between roughly 02:10 and 02:14 UTC. The messages included indications associated with inconsistent measured speeds and the loss of several flight-assistance functions.
Then the transmissions stopped.
Floating wreckage and bodies were recovered in June 2009, but the main wreckage and flight recorders remained missing.
Without the cockpit voice recorder and flight data recorder, investigators had technical clues but could not confidently reconstruct the final minutes.
Several underwater searches failed.
The Black Boxes Were 3,900 Metres Underwater
A new search effort in 2011 finally located the main wreckage on the Atlantic seabed at a depth of about 3,900 metres.
On 1 May, search teams located and recovered the flight data recorder. The cockpit voice recorder was recovered two days later.
The BEA later reported that the recorders could be read even after spending almost two years underwater.
They transformed the investigation.
Investigators could now reconstruct the aircraft's control inputs, warnings, attitude and crew responses during the final minutes.
The mystery was no longer simply what the airplane had done.
The harder question was why a serviceable modern airliner had remained stalled all the way to the ocean.
What Investigators Ultimately Found
The BEA's final report, published in July 2012, described a sequence rather than one isolated cause.
First came temporary inconsistent airspeed following obstruction of the Pitot probes by ice crystals.
That led to autopilot disconnection and a change in flight-control law.
Control inputs then destabilised the flight path.
The crew did not correctly associate the airspeed problem with the appropriate procedure, did not recognise the developing stall in time and did not subsequently diagnose the sustained stall. The control response needed to sufficiently reduce the aircraft's angle of attack was therefore not applied.
But the investigation also looked beyond the pilots.
The BEA examined training, aircraft-system behaviour, warning design, operational feedback, crew coordination and certification assumptions. Its investigation ultimately produced 41 safety recommendations.
AF447 became a major case study in what can happen when highly automated flying suddenly becomes manual flying under uncertainty.
The Search Changed Aviation Too
The consequences extended beyond stall training.
Finding AF447 required five search phases over roughly two years. Later BEA work on oceanic aircraft location says the search cost more than €30 million.
The difficulty helped drive international work on better aircraft tracking, underwater locator equipment and methods of protecting or transmitting flight data.
Flight 447 therefore left two different lessons.
One was aerodynamic:
however sophisticated an aircraft becomes, its wing can still stall.
The other was technological:
investigators should not need years to locate the information explaining an aircraft's final minutes.
AF447 was not lost because one sophisticated system simply failed.
A temporary sensor disturbance developed into a breakdown of aircraft-state awareness at the worst possible moment.
That is what made the accident so unsettling—and why its investigation mattered far beyond one flight.
Conversation
Comments
Sign in to join the conversation.