2 October 1996, minor revision 4 October

The Final Report on the accident to Flight AA965 on Dec 20 1995 has been prepared by the Colombian Oficina de Control y Seguridad Aerea, and was released through the US National Transportation Safety Board on Friday September 27, 1996. I have not yet obtained my copy of the final report. Pertinent information from the report will appear here within the next few weeks.

An Associated Press article by Randolph E. Schmid on 27 September reported the probable cause as being

• Failure of the flight crew to properly execute their approach;
• The crew's lack of situational awareness
• The crew's failure to revert to basic radio navigation when they became confused about their location

Aviation Week and Space Technology (1) says that the report states that the crew

• did not adequately plan and execute the approach, including "inadequate use of automation" (quote from the report);
• failed to discontinue the approach despite numerous clues alerting them it was inadvisable to continue;
• experienced a lack of situational awareness with regard to vertical navigation, proximity to terrain, and the relative location of navaids;
• neglected to revert to basic radio navigation at a time when use of the flight management system "became confusing and demanded an excessive workload in a critical phase of flight"

An intact FMC was recovered from the scene of the accident, and was examined by the NTSB Systems Group at Honeywell in Phoenix AZ on April 29 and 30. The report was placed in the public docket in August. An Aviation Week and Space Technology report (2) recounts that the memory was installed in a functioning FMC in a laboratory and read out. The FMC was programmed to fly direct Cali VOR (identifier 'CLO'), then reprogrammed to fly to 'R', which is the identifier for the ROZO NDB, but also the identifier for the ROMEO VOR, near Bogota, some 150nm away from ROZO. Finally, the FMC was reprogrammed to fly direct the Tulua VOR (identifier 'ULQ'), which was behind the aircraft at the time. However, the aircraft did not head towards ULQ, although this was the active leg at the time of the accident, because lateral navigation mode was disengaged before the crash. At the time of impact, the FMC interface was in a `scratchpad' mode, in which 'R' had again been entered, but was not yet part of the active flight plan.

The identifier 'R' for ROZO appears on the paper arrival and approach plates. However, in the FMC database, it is listed as 'ROZO', and 'R' designates the ROMEO VOR near Bogota. This is the result of an ARINC 424 specification, created to resolve the problem of multiple navaids in an area having the same identifier. (Remember that AA965 had to fly from the US to Colombia, therefore navaids in multiple countries had to be in the database. Clearly such a conflict-resolution specification is necessary.)

According to the Boeing submission to the docket, the cockpit area microphone picked up a sound as of rustling pages as the aircraft approached Tulua VOR from the north (between 49 and 44 DME from CLO, at between 2136:43 and 2137:25 on DFDR time). At about 41 DME, Execute Direct to 'R' was set in the FMC (time 2137:40) and the aircraft began the left turn towards ROMEO, which ended at about 2139.07 (for a total of 87 seconds unremarked by the crew). The crew then engaged Heading Select and the fatal right turn began. American Airlines, the Allied Pilots Association and Boeing all made submissions to the docket, which give "similar descriptions of key crew actions during the last 7 min. of the flight. The descriptions are a synthesis of the information from the digital flight data recorder (DFDR), the cockpit voice recorder (CVR) and the FMC readout. None of these devices gives a complete story, and judgement is required to piece together the scenario." (2).

American Airlines proposed the following probable causes:

• Inadequacies of the FMC's navigational database, and failure of those responsible to ensure that the database matched conventional published and charted information (the database did not have terminal files as recommended by ARINC 424-13);
• The flight crew's failure to perceive the FMC-initiated turn away from the intended routing;
• The approach controller's alleged inadequate English language abilities and inattention during a critical phase of the approach. (AA alleges the controller was distracted by a nonpertinent phone call during a critical radio transmission, and that he should have noticed that the reported DME distance was not decreasing, and thus that the aircraft was on a wrong heading.)

American also proposed contributing causes, including:

• lack of radar coverage
• increased task overload caused by the unexpected change in assigned runway

AA's submission seems to me to broadly agree with that of the final report, except that the controller's performance is not (reported to be) cited amongst the probably causes in the final report.

One should keep in mind that many parties to the final report have interests, namely the manufacturers, pilots' associations, the airline, and the Colombian government. The NTSB is disinterested. This situation pertains more or less with all accident investigations. One may obtain some human insight into how this may affect an investigation from (3).

References

(2): Michael A. Dornheimer, Recovered FMC Memory Puts New Spin on Cali Accident, Aviation Week and Space Technology, September 9, 1996, pp58-61. Back

(3): Jonathan Harr The Crash Detectives, The New Yorker, August 5, 1996, pp34-55. Back

Back to top

Back to the `Incidents and Accidents' Compendium

Copyright © 1999 Peter B. Ladkin, 1999-02-08
	Last modification on 1999-06-15 by Michael Blume