LOS ANGELES—As the investigation continues into the causes of the Mar. 10 Ethiopian Airlines Boeing 737 MAX accident, sources close to the probe say flight data recorder (FDR) data firmly supports the supposition that the aircraft’s left angle-of-attack (AOA) sensor vane detached seconds after take-off and that, contrary to statements from the airline, suggests the crew did not follow all the steps for the correct procedure for a runaway stabilizer.

Detailed analysis of the FDR trace data shows that approximately six seconds after liftoff was signaled by the weight-on-wheels switch data, the data indicate the divergence in angle-of-attack (AOA) and the onset of the captain’s stick-shaker, or stall warning. Almost simultaneously, data shows the AOA sensor vane pivoted to an extreme nose-high position.

This, says one source, is a clear indication that the AOA’s external vane was sheared off—most likely by a bird impact. The vane is counter-balanced by a weight located inside the AOA sensor mounting unit, and without aerodynamic forces acting on the vane, the counterweight drops down. The AOA sensor, however, interpreted the position of the alpha vane balance as being at an extreme nose-high angle-of-attack.

With the stick shaker active, the trace indicates the crew pushed forward on the column to counteract what they believed were indications of potential approach to stall. The aircraft, now in level flight, also accelerated rapidly as its power setting remained at 94% N1 thrust used for take-off. This was followed by some manual trim inputs using the thumb switches on the control column.

Seconds after speed advisories were heard, the crew raised the flaps. With the autopilot turned off, flaps up and erroneous AOA data being fed to the flight control computer (FCC), the stage was set for the MAX’s maneuvering characteristics augmentation system (MCAS) to activate. This is indicated by approximately 8-sec of nose-down stabilizer movement, which was followed by the use of manual trim on the control column. However, with the MCAS having moved the stabilizer trim by 2.5 units, the amount of manual nose-up trim applied to counteract the movement was around 0.5 units, or roughly only 20% of the amount required to correctly re-trim the aircraft.

Because of the way the aircraft’s flight control computer P11.1 software worked, the use of manual trim also reset the MCAS timer, and 5 sec. later, its logic having not sensed any correction to an appropriate AOA, the MCAS activated again. The second input was enough to put in the full nose-down trim amount. The crew again manually counteracted with nose-up trim, this time offsetting the full amount of mis-trim applied by the latest MCAS activation.

By then, some 80% of the initial MCAS-applied nose down trim was still in place, leaving the aircraft incorrectly trimmed. The crew then activated the stabilizer trim cutoff switches, a fact the flight data recorder indicates by showing that, despite the MCAS issuing a further command, there was no corresponding stabilizer motion. The aircraft was flying at about 2,000 ft. above ground level, and climbing.

The crew apparently attempted to manually trim the aircraft, using the center-console mounted control trim wheels, but could not. The cut-out switches were then turned back on, and manual trim briefly applied twice in quick succession. This reset the MCAS and resulted in the triggering of a third nose-down trim activation lasting around 6 sec.

The source says the residual forces from the mis-trim would be locked into the control system when the stabilizer cut-off switches were thrown. This would have resulted in column forces of up to around 50 lb. when the system was switched back on.

Although this could have been reduced by manually trimming the aircraft, this did not occur, and the third MCAS activation placed the aircraft in a steep nose-down attitude. This occurred with the aircraft near its peak altitude on the flight—about 6,000 ft. The engines remained at full take-off power throughout the flight, imposing high aerodynamic loads on the elevators as the crew attempted to pull back on the columns.

Vertical acceleration data also indicates momentary negative g during which the AOA sensor on the left side unwinds. This is seen as further validation of the theory that the external part of the alpha vane was detached as the apparent change in angle indication could only be explained by the effect of negative g on the counterbalance weight, forcing it to float up inside the sensor housing. In addition, the captain’s stick shaker also comes off twice in this final phase, further reinforcing the severed vane notion.

The source indicates the crew appeared to be overwhelmed and, in a high workload environment, may not have followed the recommended procedures for re-trimming. Boeing’s stabilizer runaway checklist’s second step directs pilots to “control aircraft pitch attitude manually with control column and main electric trim as needed,” according to one U.S. airline’s manual reviewed by Aviation Week. If the runaway condition persists, the cut-out switches should be toggled, the checklist says.