On 12 November 2001, the crew of an Airbus A300-600 (N14053) being operated by American Airlines on a scheduled international passenger flight from New York JFK to Santo Domingo, Dominican Republic lost control of their aircraft just 1½ minutes after takeoff during the second of two minor wake turbulence encounters and the subsequent destruction of the aircraft by ground impact forces and a post crash fire led to all 260 occupants and five others on the ground being killed.
An Investigation into the accident was carried out by the NTSB. The FDR and CVR were recovered and their relevant data were successfully downloaded. Radar surveillance data for the departures of both the A300 and the preceding 747 were also available. A video-referenced study was undertaken to compare elapsed time information taken from available video recordings with that of the radar/CVR/FDR data and to thereby calculate the position of the aircraft after the loss of FDR and radar data. Because the CVR was powered from the emergency bus bar, it continued to record data after the engines separated from the aircraft.
It was noted that the 42 year-old Captain had been employed by the airline since July 1985 and had received a type rating on the A300 in September 1988 whilst serving as a First Officer and a type rating on the Boeing 727 in December 1991. He then qualified in command on the A300 in August 1998. He had 8,050 hours total flying experience which included 1,723 hours on type in command. The 34 year-old First Officer, who was PF for the accident flight, had been employed by the airline since March 1991 and had received his A300 type rating in November 1998 after previously working on the Boeing 727. He had a total of 4,403 hours flying experience which included 1,835 hours on type. There was some evidence that the First Officer had previously demonstrated a tendency to over-control when encountering wake turbulence despite being otherwise considered generally competent by those who had flown with him.
It was established that the taxi out for departure from runway 31L was in turn behind a Japan Airlines Boeing 747-400. Shortly after the 747 had begun its takeoff, the TWR controller cautioned the A300 pilots about wake turbulence and instructed them to line up and hold. The A300 subsequently lifted off about 1 minute 40 seconds after the 747 and thereafter was cleared to turn left, fly the ‘Bridge’ climb and change frequency to New York TRACON. Radar data indicated that the aircraft began its climbing left turn at 500 feet and established on a track of 220º. The Captain checked in with the TRACON departure controller and advised passing 1300 feet for 5000 feet and was re-cleared to 13,000 feet.
Just after a minute airborne, an instruction to the A300 to turn left and proceed direct to the WAVEY intersection was given and acknowledged. At this time, FDR data indicated that there had been a mild wake turbulence encounter which had been recognised as such by the crew. FDR data showed that there had been some movement in all the primary flight controls over a period of 5 seconds as the turn to WAVEY was made. Then, approximately 10 seconds later, with the aircraft climbing at approximately 250 knots, a second mild wake turbulence encounter occurred in response to which there were a series of much more aggressive control inputs. After an initial right rudder input, the First Officer proceeded with five alternating full deflection rudder pedal inputs. These were accompanied initially by “a thump, a click and two thumps”, followed two seconds later by a “snap” and one second after that by a “loud thump” as the right rear main attachment fitting of the vertical stabiliser failed. The stabiliser itself then separated from the aircraft at the same time as the CVR recorded a “loud bang”, some 6½ seconds after the sequence of aggressive control inputs had begun. Subsequently, both engines also separated from the aircraft before impact. The ground track of the aircraft is depicted on the illustration below.
It was noted that the origin of the two short wake turbulence episodes was the 747 which had taken off the same runway as the A300 and which had then followed a similar left turning track whilst remaining a little over half a mile west of and above that of the A300 and that the available evidence was that neither was more than minor in nature. It was also noted that the Captain had not sought to intervene during the second turbulence encounter and that there was no evidence on the CVR to suggest that he had appreciated that the upset was a consequence of the First Officer’s control inputs, which given its short duration was judged ‘understandable’.
The A300 radar ground track annotated with key events (the ‘white streak’ mentioned was fuel liberated from the wing following engine separation prior to impact). [Reproduced from the Official Report]
Why it happened
No evidence was found that mechanical failure could have caused the five cyclic movements of the rudder pedals and rudder recorded immediately after the onset of the second turbulence encounter. It was also found that the only way to move the rudder in a way that could create an FDR record which would match that of the accident aircraft was if a pilot depressed the rudder pedals and it was therefore concluded that the recorded movements had been the result of the First Officer’s rudder pedal inputs. It was also concluded after extensive analysis that the vertical stabiliser “had performed in a manner that was consistent with its design and certification” and that it had detached from the fuselage in overstress after exposure to aerodynamic loads which were around twice the certified limit load design envelope and also greater than the certified ultimate load design envelope. The remainder of the Investigation therefore centred on analysis of the First Officer’s rudder pedal inputs.
It was determined that three main factors had influenced the First Officer’s use of the rudder during the accident sequence:
- A tendency to react aggressively to wake turbulence
Evidence of his responses to previous wake turbulence encounters was obtained and two company Captains provided “noteworthy accounts” of the First Officer’s reaction to such events which suggested over-reaction had been a feature of them. It was also noted that his response to the first of the two wake turbulence encounters on the accident flight had, whilst having had no direct bearing on the accident, involved control inputs that “seemed excessive” as a response to the momentary effect that it had on the aircraft.
- The Pilot Training Regime
The training on wake turbulence, upset recovery, and rudder pedal use provided in the American Airlines’ Advanced Aircraft Manoeuvring Programme (AAMP) included ground school training which “encouraged pilots to use rudder to assist with roll control during recovery from upsets, including wake turbulence”. It was also observed that the AAMP excessive bank angle recovery exercise was based on an unrealistic scenario and was capable of causing pilots “to develop control strategies that were effective in the simulator but might be inappropriate or even dangerous” if performed in an aircraft. This was considered to have had the potential to lead to the First Officer “having an unrealistic and exaggerated view of the effects of wake turbulence, erroneously associating wake turbulence encounters with the need for aggressive roll upset recovery techniques” and, as a result, developing control strategies which would produce potentially surprising and confusing outcomes if attempted in an aircraft rather than a simulator. It was further concluded that pilot training on the effect of rudder pedal inputs made at high speeds was inadequate and that it did not clearly address how the rudder limiter system operated.
- The characteristics of the A300-600 rudder control system
It was determined during the Investigation that the type rudder control system was “more sensitive” than that on other transport-category airplanes and noted that although rudder inputs had historically been made only when aligning an aircraft with the runway during crosswind landings or retaining directional control in engine-out scenarios, rudder use was now being recommended in training such as that provided in the AAMP as a means to assisting roll control during upset recovery in some circumstances. It was also noted that the rudder limiter system on the A300-600 was of the variable stop design which means a given pedal input results in that both rudder deflection and rudder pedal travel are both limited as airspeed increases so that “a given amount of pedal input results in the same rudder deflection at all airspeeds but the pedal travel (and consequently, pedal force) required to hit the pedal stops decreases as airspeed increases”. An exemplar consequence of this design was observed as being that the rudder is “twice as responsive to a pedal displacement at 250 KCAS as it would be at 165 KCAS”.
More generally, the Investigation concluded that:
- an incorrect understanding amongst American Airlines A300 pilots of the meaning of Va and the extent to which structural protection from the effects of full deflection rudder input exists provided that an aircraft remains below this speed was likely to be widespread. The Investigation also considered that there was “no reason to believe that the misunderstanding about manoeuvring speed is limited to A300-600 pilots”.
- The shortcomings identified in the American Airlines AAMP were just as likely to be present in other airlines’ upset recovery training programmes, especially but not only in respect of simulator use in such programmes.
Formally documented Findings from the Investigation included those made directly in respect of the accident and its circumstances and with more general applicability. The latter included the following:
- To minimise the potential for aircraft-pilot coupling events, transport-category pilots would benefit from training about the role that alternating full control inputs can play in such events and training that emphasises that alternating full rudder inputs are not necessary to control a transport-category airplane.
- There is a widespread misunderstanding among pilots about the degree of structural protection that exists when full or abrupt flight control inputs are made at airspeeds below the manoeuvring speed.
- Federal Aviation Administration standards for unusual attitude training programs that take into account industry best practices and are designed to avoid inaccurate or negative training would lead to improvement and standardisation of industry training programs.
- The use of lower levels of automation, such as simulators without motion or simple computer screen displays, may be more appropriate to provide the necessary awareness training with less danger of introducing incorrect information.
The Probable Cause of the accident was determined to be “the in-flight separation of the vertical stabiliser as a result of the loads beyond ultimate design that were created by the First Officer’s unnecessary and excessive rudder pedal inputs”.
Two Contributory Factors were also identified as follows:
- the characteristics of the Airbus A300-600 rudder system design.
- some elements of the American Airlines Advanced Aircraft Manoeuvring Program.
A total of 7 new Safety Recommendations were made during the course of the Investigation as follows:
- that the Federal Aviation Administration require the manufacturers and operators of transport-category airplanes to establish and implement pilot training programs that:
- explain the structural certification requirements for the rudder and vertical stabiliser on transport-category airplanes.
- explain that a full or nearly full rudder deflection in one direction followed by a full or nearly full rudder deflection in the opposite direction, or certain combinations of sideslip angle and opposite rudder deflection can result in potentially dangerous loads on the vertical stabiliser, even at speeds below the design manoeuvring speed.
- explain that, on some aircraft, as speed increases, the maximum available rudder deflection can be obtained with comparatively light pedal forces and small pedal deflections.
The FAA should also require revisions to airplane and pilot operating manuals that reflect and reinforce this information. In addition, the FAA should ensure that this training does not compromise the substance or effectiveness of existing training regarding proper rudder use, such as during engine failure shortly after takeoff or during strong or gusty crosswind takeoffs or landings. [A-02-01]
- that the Federal Aviation Administration carefully review all existing and proposed guidance and training provided to pilots of transport-category airplanes concerning special manoeuvres intended to address unusual or emergency situations and, if necessary, require modifications to ensure that flight crews are not trained to use the rudder in a way that could result in dangerous combinations of sideslip angle and rudder position or other flight parameters. [A-02-02]
- that the Federal Aviation Administration require all manufacturers of transport-category airplanes to review and, if necessary, revise their maintenance manual inspection criteria for severe turbulence and extreme in-flight manoeuvres to ensure that loads resulting from positive and negative vertical accelerations, as well as lateral accelerations, are adequately addressed. [A-03-41]
- that the Federal Aviation Administration require all manufacturers of transport-category airplanes to establish and validate maximum threshold values for positive and negative vertical and lateral ‘g’ accelerations beyond which direct manufacturer oversight and intervention is required as a condition for returning the airplane to service. [A-03-42]
- that the Federal Aviation Administration require all operators of airplanes that have experienced accelerations exceeding the threshold values established as a result of Safety Recommendation A-03-42 (or that the operator has reason to believe might have exceeded those thresholds), as determined from FDR and other available data, to notify the FAA immediately of such high loading events and provide all related loads assessment and inspection results. [A-03-43]
- that the Federal Aviation Administration require manufacturers of transport-category airplanes to immediately notify the appropriate certification authority of any event involving accelerations exceeding the threshold values (or that the manufacturer has reason to believe might have exceeded those thresholds) necessitating the intervention of the manufacturer, and provide all related loads assessment and inspection results. [A-03-44]
- that the Federal Aviation Administration require that within 2 years, all Airbus A300-600/A310 and Boeing 747-400 airplanes and any other aircraft that may be identified as recording filtered data be retrofitted with a flight data recorder system capable of recording values that meet the accuracy requirements through the full dynamic range of each parameter at a frequency sufficient to determine a complete, accurate, and unambiguous time history of parameter activity, with emphasis on capturing each parameter’s dynamic motion at the maximum rate possible, including reversals of direction at the maximum rate possible. [A-03-50]
A total of 8 new Safety Recommendations were made at the conclusion of the Investigation as follows:
- that the Federal Aviation Administration modify 14 Code of Federal Regulations Part 25 to include a certification standard that will ensure safe handling qualities in the yaw axis throughout the flight envelope, including limits for rudder pedal sensitivity. [A-04-56]
- that the Federal Aviation Administration should, after the yaw axis certification standard recommended in Safety Recommendation A-04-56 has been established, review the designs of existing airplanes to determine if they meet the standard. For existing airplane designs that do not meet the standard, the FAA should determine if the airplanes would be adequately protected from the adverse effects of a potential aircraft-pilot coupling (APC) after rudder inputs at all airspeeds. If adequate protection does not exist, the FAA should require modifications, as necessary, to provide the airplanes with increased protection from the adverse effects of a potential APC after rudder inputs at high airspeeds. [A-04-57]
- that the Federal Aviation Administration review the options for modifying the Airbus A300-600 and the Airbus A310 to provide increased protection from potentially hazardous rudder pedal inputs at high airspeeds and, on the basis of this review, require modifications to the A300-600 and A310 to provide increased protection from potentially hazardous rudder pedal inputs at high airspeeds. [A-04-58]
- that the Federal Aviation Administration develop and disseminate guidance to transport-category pilots that emphasises that multiple full deflection, alternating flight control inputs should not be necessary to control a transport-category airplane and that such inputs might be indicative of an adverse aircraft-pilot coupling event and thus should be avoided. [A-04-59]
- that the Federal Aviation Administration amend all relevant regulatory and advisory materials to clarify that operating at or below manoeuvring speed does not provide structural protection against multiple full control inputs in one axis or full control inputs in more than one axis at the same time. [A-04-60]
- that the Federal Aviation Administration adopt and disseminate written guidance for use in developing and accepting upset recovery programs; such guidance could take the form of an advisory circular and should reflect the industry’s best practices and be designed to avoid inaccurate or negative training. [A-04-61]
- that the Federal Aviation Administration along with developing the guidance recommended in Safety Recommendation A-04-61, evaluate issues concerning the level of automation appropriate to teaching upset training, and develop and disseminate guidance that will promote standardization and minimize the danger of inappropriate simulator training. [A-04-62]
- that the Direction Général de l’Aviation Civile France review the options for modifying the Airbus A300-600 and the Airbus A310 to provide increased protection from potentially hazardous rudder pedal inputs at high airspeeds and, on the basis of this review, require modifications to the A300-600 and A310 to provide increased protection from potentially hazardous rudder pedal inputs at high airspeeds. [A-04-63]
The Final Report of the Investigation was adopted on 26 October 2004 and subsequently published.