Flight data indicates that the automated control system was responsible for Ethiopia’s crash

Manual control is the first step in a typical flight. The pilot and copilot will manually steer the aircraft onto the taxiway. They will then configure the flaps and actively control the aircraft while it accelerates along the runway. Finally, they will pull back the control yoke and lift the plane into the air. FlightRadar24.com made the altitude and speed data of Ethiopian Airlines Flight 302 available to the public.

The initial climb after takeoff also appeared to be going as normal. Normally, the pilot retracts the landing gear while maintaining a steady speed. The plane may accelerate until the speed is high enough to safely retract the flaps, which are extended at low speeds to increase lift. This usually happens within the first minute of takeoff. The autopilot will be activated once the aircraft has reached 1,000 feet.

At this point, the computer takes control. My analysis of the data indicates that things went wrong on Ethiopian Airlines Flight 302. Modern autopilot systems give the computer full control of the aircraft, including the throttles, rudders, elevators, and ailerons.

Simulating an expected flight

Using modeling tools developed by my team, I recreated a fictional flight profile to simulate Ethiopian Airlines 737 departure based on handbook procedures for an identical aircraft carrying a comparable amount of weight. The simulation’s timing, altitudes, and key speeds all match my best estimation of what a trained pilot should do.

By comparing this data with the actual flight data, I was able to see the differences between the ideal performance predicted and the actual motions during the lost flight. My simulation matches the speed of the aircraft during its takeoff roll and recreates the first few miles in airborne flight. The official manual does not specifically call for or prohibit the pilot from accelerating the aircraft gently during the initial climb.

Flight paths diverge sharply between a typical and actual flight only when the aircraft reaches an altitude of 1000 feet above the runway.

The pilot should have activated the autopilot immediately after retracting the flaps, allowing the computer to control a constant climb. The ill-fated plane began to accelerate and dive, gaining speed and losing altitude until it hit the ground just a few miles from the airport.

A plane can crash for several reasons. One possibility is that the engine might malfunction. The telemetry data does not indicate the loss in acceleration caused by an engine failure. A part of the fuselage or wings, or even the tail, could have broken or collapsed. The data does not show any change in speed or rate of climb that such a loss of stability would cause.

The crash was not caused by pilot error. I have studied overreactions by pilots in developing emergencies and do not see any evidence that this occurred before the initial dive. The pilots appear to be flying a typical takeoff. The pilots did not report any other mechanical failure to the control tower. It is not clear if they overreacted to an emergency or overcompensated. The radar track does not show any evidence that the pilots overcompensated.