On 10 April 2019, an Airbus A321 (N114NN) being operated by American Airlines on a scheduled domestic passenger flight from New York JFK to Los Angeles was about to become airborne in normal night visibility when it rolled sharply left before returning to wings level. After takeoff, a report of a dented left wingtip was received from the cabin and the crew decided to return although when asked by the controller if the situation was an emergency, a negative response followed. A landing without further event was made just under half an hour later following which part of a runway edge marker was found lodged in the left wingtip. Damage to the aircraft was substantial and it was found not to be economically repairable but there were no injuries to the 110 occupants.
An Accident Investigation was carried out by the NTSB using relevant data downloaded from the CVR and DFDR and obtained from other sources. It was noted that the 58 year-old Captain had a total of 19,654 hours flying experience which included 3,005 hours on type and 2,975 hours in command on all types. The 58 year-old First Officer had a total of 10,103 hours flying experience which included 1,824 hours on type. Both were former flight engineers.
After an on time departure from the gate with the Captain as PF, the aircraft was eventually cleared to line up and wait on runway 31L. Takeoff clearance was then given with the wind from 010° at 17 knots. Engine thrust was applied symmetrically with no nose wheel steering input and left rudder pedal inputs consistent with a small right crosswind ranging of up to 13° were made as acceleration occurred.
DFDR data showed that as the speed reached the VR, the Captain’s left rudder input suddenly began to increase from around 8° and its recorded lateral acceleration increased from 0.11g to 0.32g.
As the aircraft began to veer left in response, the Captain then applied full aft sidestick and right sidestick up to 16°. A momentary change from right sidestick to left occurred before an almost immediate change to maximum right deflection as the nose landing gear left the ground with the rudder deflected about 20° to the left.
The left roll reached 26° with left roll rate still increasing as the right main landing gear left the runway with the aircraft on a heading 25° to the left of the runway centreline. Two seconds later, the achieved left roll reached a maximum of 37° and in addition to being held at maximum nose up position, both sidesticks were moved the maximum nose-up and one second later, “a sound consistent with a wing strike” could be heard on the CVR. DFDR data showed that the left aileron moved then towards neutral without a control input consistent with the left aileron moving due to contact with the ground. The rudder pedals then reached their full right position and the left main landing gear became airborne. Over the next several seconds, with the aircraft still close to the ground, the Captain made a series of seemingly random rudder and aileron control inputs during which the First Officer was recorded as saying “I don’t know what’s goin’ on” followed by the Captain saying “what the (expletive) happened?” and the First Officer responding “I don’t know” and “the engines all…good” followed by the Captain adding “it just (expletive) rolled on me”.
During interview, the Captain stated that he had been “looking at the runway edge” and knew that he had to get the aircraft into the air and recalled making right rudder and right aileron inputs “as the aircraft had begun to roll left”. DFDR data showed that the corrective action had been taken by the Captain’s reversal of his left rudder input to full right rudder accompanied by both pilots’ sidesticks being moved to the maximum (20°) deflection right and maximum pitch up which led to an automated aural warning of dual input.
The First Officer tentatively suggested a return to land and after a brief discussion, the Captain concurred and the First Officer advised ATC of the flight’s wish to return after “experiencing a strong roll to the left”. The controller asked if the situation was an Emergency but after checking with the Captain the First Officer responded that it was not. Five minutes later, one of the cabin crew called on the interphone to say that a passenger seated near the left wing had reported that the wing “looks dented” and “doesn’t look normal” following this up five minutes later with a call saying that the left wing looked as if it had sustained “a little damage”. By this time, the climb had been stopped at FL200 and radar vectors were being provided back to the airport. The remainder of the flight was uneventful and a few minutes later, the aircraft landed after 28 minutes airborne.
An examination of the departure runway found a 170 metre-long left main landing gear tyre mark, a 65 metre-long right main landing gear tyre mark and a 98 metre-long left wingtip scrape mark. The wingtip scrape marks began beyond the runway edge marking and extended to the edge of the paved surface shoulder. Debris found embedded in the aircraft left wing (see the illustration below) was found to have been part of a ‘distance remaining’ marker located on the outer edge of the 60 metre-wide runway’s paved shoulder. Damage to the left wing outboard slat, aileron, and outboard flap as well as to the underside of the left wingtip sharklet were also found.
The left wingtip with part of a runway maker mountimg embedded in it. [Reproduced from the Official Report]
Scrape damage to underside of the left wing slat. [Reproduced from the Official Report]
Further examination of the six year old aircraft found that the left wing had a permanent upward deflection starting at about midspan which resulted in the left wingtip being about 15 cm higher than the right wingtip. This discovery led to a decision by American Airlines to retire the aircraft from service and scrap it.
Why It Happened
It was determined that the significant left roll during liftoff had caused the left wingtip to strike the paved area beyond the runway left side edge and the collision with a runway distance marker which became partially embedded in the wingtip. Engineering simulations were carried out by Airbus for the Investigation which showed that the several seconds of near-maximum left rudder “had generated a rolling moment after the gear had left the ground” which alone had caused the left roll rate recorded on the DFDR. Nothing else abnormal in the aircraft flight control was noted except for the deflection upwards of the left aileron as the wingtip scraped the ground.
These simulations were also able to show that the deviation left of centreline was entirely due to the Captain’s rudder pedal input and unconnected with the existence of a crosswind component. It was noted that a nearby automatic wind velocity recording site was recording the equivalent of a mean 14 knot crosswind with little variation and could not be associated with the event.
American Airlines examined their retained A321 OFDM takeoff data for more than 270,000 of their flights which had taken place between March 2018 and April 2019 looking only at the time interval between reaching an airspeed of 130 KCAS and reaching 50 feet agl. They found that:
- only one of these flights had recorded a higher maximum rudder deflection but more than 2,300 of them had operated with a higher crosswind component than the accident aircraft.
- the heading deviation of the accident aircraft during its takeoff roll was more than three times that of the next highest deviation.
- the duration of a rudder application greater than 50% in accident aircraft was more than that for any other such flight.
- several other flights had been exposed to larger crosswind components - up to 43 knots in gusts - but none had recorded either the maximum rudder deflection or the duration of rudder application reached by the accident flight.
- the duration of rudder input was compared with the maximum rudder deflection greater than 50% between 0.5 second before rotation and 0.5 second after liftoff. This found that the accident flight had both a greater maximum rudder deflection and a longer rudder input than any other flight.
The Investigation determined that the Probable Cause of the accident was "the Captain’s excessive left rudder pedal input during the takeoff ground roll, which caused a large heading deviation and a left roll upon rotation that resulted in the left wingtip striking the ground".
The Final Report was published on 22 July 2012. No Safety Recommendations were made.