Aircraft Accidents and Lessons Unlearned XLII: TAM Flight 3054

On July 17, 2007, Táxi Aéreo Marilla (TAM) Linhas Aéreas flight 3054 (TAM3054), an Airbus A-320, registration PR-MBK, crashed while landing on Runway 35L at São Paulo/Congonhas airport (airport identifier: SBSP). After touching down, the aircraft did not slow; it veered to the left, overran the southwest side of the runway, crossed over Washington Luis Avenue and struck both a cargo building and a fuel service station at ninety-six knots. It caught fire; the aircraft was destroyed. The investigation was conducted by Brazil’s Centro de Investigação e Prevenção de Acidentes Aeronáuticos (CENIPA), [translated] the Aeronautical Accidents Investigation and Prevention Center. No archived evidence could be located on the investigation agency’s website to review; all information presented here is taken straight from the report: RF A-67/CENIPA/2009.

There are problems analyzing accidents written in a language that is not native to the reader. It was important that the Findings/Recommendations had received the proper interpretations to other languages, such as English. How does anyone benefit from the analysis if the translation was poor? Even the manufacturer terminology can baffle the investigator, cascading into more confusion. Consider the old game, Telephone Line, where a message spoken to the first person in line is drastically different when it reaches the fifteenth person. Confusion can result from simple terminology; the General Electric CF6-50 engine had a Constant Speed Drive that drove an engine generator at a consistent speed. On a later model, the CF6-80 engine, the device was called the Integrated Drive Generator – same purpose, different name. When an investigator is unfamiliar with the terminology differences from Boeing logic to Airbus logic, the end report becomes gibberish.

On page 52 of the TAM report, such an uncertainty exists, “… Aeronautical Accident Prevention Program (PPAA) of the company [TAM] for the year 2007, those accredited professionals were not considered for the development of accident prevention actions.” The Operations and Maintenance departments, by definition, subscribe to accident prevention in every action they perform; an accident prevention program would be redundant. What was an accident prevention program? Were investigators experienced in airline cultures? Other problems could arise, such as an overreliance on the manufacturers and the airline to fill in the blanks for them. How likely were manufacturers and airline to expose their own weaknesses? The A320 was/is a popular airliner around the world; the importance of a lost opportunity to learn cannot be stressed enough, especially if lost in translation.

A curious report problem: the number of recommendations. This report had fifty-two Conclusions and fifty Recommendations– an incredible number of recommendations for a single-aircraft accident. In the accident report, Quantity ≠ Quality. Recommendation numbers are not proportional, are not indicative of safety value. Some recommendations were valid, insightful; unfortunately, good information was lost in the commotion.

There were three focuses for the final report: Runway Integrity, Training and Mechanical Anomalies. Of the twelve possible (probable) causes, there were five the report considered ‘contributors’ to the accident: Training, Cockpit Coordination, Management Planning, Little Experience of the Pilot and Management Oversight. Any reference to the ‘pilot’ was ambiguous. Were both pilots considered inexperienced, or just one? Runway integrity should have been considered a contributing factor and the factual information about this should have been exploited. The report spent resources analyzing the runway’s condition, but the report did not list it as a contributor nor as an undetermined factor.

The aircraft was operating with a deferred #2 engine thrust reverser. This meant the right engine could not be used to stop the aircraft once it touched down on the runway. Thrust reversers are not required to stop the aircraft, but this scenario, exacerbated by the wet runway, created a landing challenge. The #1 engine thrust reverser would be deployed on landing, thus introducing a yaw effect, where the aircraft will pivot left in the direction of the deployed reverser. On the right engine, the thrust lever (TL) was set out of configuration (to Climb), which provided forward thrust to further drive the aircraft left.

Per the flight data recorder (FDR), at touchdown, the number one TL was at ‘IDLE’ while the number two TL was at ‘CL’ (Climb). The report stated on page 67, “If one lever stays at the “CL” position during landing, it deactivates the actuation of the ground spoilers, significantly reducing the aircraft braking capability (between 45% to 50%)” Did the pilots mistakenly misconfigure the aircraft for landing, essentially deactivating the autobrakes and ground spoilers? Why were the TLs not moved together on landing, why the split? Did they pay attention to the Minimum Equipment List (MEL) procedures for the deferred reverser? How did this reflect in their training? Was this pursued with TAM post-accident?

Maintenance was not listed as a contributor, an unfortunate oversight. The investigators failed to interview members of Maintenance, especially at Porto Alegre, the airport TAM3054 had departed from. The #2 reverser had been deferred since July 13, 2007, mechanically locked out to prevent inadvertent deployment. Any pilot concerns in Porto Alegre may have been discussed with the mechanic who launched the flight, including pilot/MEL procedure compliance issues.

Both the cockpit voice recorder (CVR) and the FDR confirmed the ground spoilers did not deploy. Autobrakes also did not function on touchdown in SBSP. These issues should have been explored with Maintenance. The mechanic could have provided insight into why the pilots’ actions led to the spoiler problems. Were the MEL procedures understood before launching the accident flight? What about anti-skid? Page 42 stated, “The Anti-skid system, in turn, functioned normally, preventing the blocking [locking?] of the main gear wheels during the braking.” PR-MBK was launched out of Porto Alegre with an allegedly functional anti-skid system … or was it?

According to information provided by the operator, several of PR-MBK’s last maintenance records from Porto Alegre were destroyed in the accident; they were being transported to Congonhas for entry into Maintenance’s database. Why were there no copies kept in Porto Alegre? Were TAM maintenance log pages normally duplicated and, if not, why not? Did an investigator pursue this point?

One good point made, though lost in the recommendations, concerned training on page 95. “The theoretical qualification of their pilots was founded on the exclusive use of computer interactive courses (CBT) which allowed a massive training but did not ensure the quality of the training received.” Anyone who has received CBT for maintenance or flight, understands the problems associated with this method, that the computer is the sole instructor during an important phase of the training. CBT is a poor instructor in that it cannot answer questions as thoroughly as a trained instructor. Quality depends on a distraction-free, uninterrupted learning environment, one that does not add to the confusion.

Concerning the second in command (SIC) pilot’s experience, per the report, page 10, “The SIC had recently been hired by the company as a captain [When?]. He did not have previous experience in the A319/320 airplanes and did his A-320 training (already as a captain) at the very company.” If correctly translated, the SIC – first officer – had a captain’s rating with no previous experience before TAM, in the A-320. The SIC had 14,760 total flight hours, 237 flight hours on the Airbus A-320.

The report’s Findings pointed to credible administrative and cultural issues with TAM that were not pursued, while drifting into unproven findings, such as the Captain’s ‘mild headache’ (page 95), saying it, “… may have influenced his [captain’s] cognitive and psychomotor capabilities during the final moments …” A migraine would have been debilitating. A mild headache?

Recommendation 172/A/07 suggested, “… a warning system to allow the crew to identify a wrong setting of the thrust levers …” Another alarm or warning? The fallacy of overregulation and/or overengineering, taking the responsibility from the pilot and giving it to the aircraft. How does removing the pilots’ responsibilities to aircraft system monitoring benefit safety? How does adding another alarm help?

The TAM3054 accident report was confusing. Worse, it wasted resources on irrelevance and ignored issues that could have benefited safety. It was not that CENIPA missed the important safety mistakes, it was that CENIPA did not shine a bright enough light on the very important problem of inadequate pilot training, coordination and flight management.