About This Website » Accident Analysis
Accident Analysis
- Overview
Introduction
From the literature, comments from the experts
we surveyed, news media, and other sources, we identified 34 aircraft accident reports
that we thought might contain evidence related to the flight deck automation issues. We
were able to obtain 20 of these reports from the US National Transportation Safety Board
and other national and international agencies that conduct accident investigations.
We reviewed these reports, looking for statements by the investigating board identifying one
or more of the flight deck automation issues as contributing to the accident. We have since
identified and reviewed 11 additional accident reports for a total of 31 reports reviewed.
The strength we assigned to the statements we identified as evidence in the reports was based on how close the investigating
board's findings were to one of our issues and on the extent to which the problem they
described contributed to the accident or incident (or at least how confident the board was
that it contributed to it). We used the following table for assigning strengths to
accident/incident evidence.
| Strength |
Contribution to
Accident |
Similarity of Board's Statement to
Issue Statement |
+5 |
probable |
equivalent to issue statement |
+4 |
possible |
equivalent to issue statement |
+3 |
probable |
similar to issue statement |
+2 |
possible |
similar to issue statement |
+1 |
probable/possible |
analyst unsure |
+1 |
analyst unsure |
any of the above |
In the table, the phrase "equivalent to issue statement" means that the
board's description of the problem they discovered in their investigation was
approximately the same as the problem suggested by the issue statement of one of the
flight deck automation issues. The phrase "similar to issue statement" means
that their description did not quite match the issue statement but that their essential
meaning seemed to be the same, or that it seemed to the analyst that the
investigating board described a problem very similar to that suggested by the issue
statement. The phrase "analyst unsure" means that the analyst thought that the
board had implied one of the issues in the accident but the way the board worded its
findings was ambiguous.
The investigating board's statement that a problem's contribution to the
accident/incident was "probable" means that the board determined that it was a
major contributing factor in the accident, while "possible" means that it was
merely a contributing factor or possibly just one of the board's hypotheses. In the same
column, "analyst unsure" means that the analyst could not determine from the
description how significant the board found the problem to be, but that the board at least
seemed to imply that it could have played a role.
For example, if we read that a board investigating an accident concluded that the
flightcrew's lack of understanding of the flight level change mode of the aircraft's
autoflight system contributed to the accident -- but was not its probable cause -- we
would record evidence for issue105 with strength +4.
We did not look for or find contradictory evidence in accident reports. An accident or
incident investigation is not an experiment and the fact that an investigating board finds
that, for example, the pilots of an accident aircraft were thoroughly knowledgeable in the
use of autoflight system and that lack of knowledge did not contribute to the accident is
not evidence that all (or even most) pilots thoroughly understand the automation they use.
Accident Report Results
The following is a list of the aircraft accidents in which we found evidence related to
the flight deck automation issues. The accidents are listed alphabetically by air carrier
and include links to the bibliographic information and evidence found in the reports:
| Investigator(s) |
Aircraft Accident |
|
|
| National Transportation and Safety Board |
Aeromexico DC10-30 accident, Luxembourg Europe, 11 Nov 79 |
EVIDENCE |
| Investigation Commission of Ministry of Transport - France |
Air France A320 accident, Mulhouse-Habsheim France, 26 Jun 88 |
EVIDENCE |
| Investigation Commission of Ministry of Transport - France |
Air Inter A320 accident, near Strasbourg France, 20 Jan 92 |
EVIDENCE |
| Council of State Appointed Investigation Commission - Finland |
Air Liberte Tunisie DC9-83 (MD83) accident, Kajaani, Finland,
3 Nov 93 |
EVIDENCE |
| Aeronautica Civil of the Republic of Colombia |
American Airlines Flight 965, B757-223 accident, near Cali,
Colombia, 20 Dec 95 |
EVIDENCE |
| National Transportation and Safety Board |
Atlantic Coast Airlines Jetstream 4101 accident, Columbus, OH
USA, 7 Jan 94 |
EVIDENCE |
| Air Accident Investigation Branch, Department of Transport -
England |
British Midland B737-400 accident, Leicestershire England, 8
Jan 89 |
EVIDENCE |
| Aircraft Accident Investigation Commission - Ministry of
Transport Japan |
China Airlines A300B4-622R accident, Nagoya, Japan, 26 Apr 94 |
EVIDENCE |
| National Transportation and Safety Board |
China Airlines B747-SP-09 accident, near San Francisco, CA
USA,19 Feb 85 |
EVIDENCE |
| National Transportation and Safety Board |
Eastern L1011 accident, near Miami, FL USA, 29 Dec 72 |
EVIDENCE |
| Ministry of Civil Aviation - India |
Indian Airlines A320 accident, Bangalore, India, 14 Feb 90 |
EVIDENCE |
| Main Commission Aircraft Accident Investigation - Poland |
Lufthansa A320-211 accident, Warsaw Poland, 14 Sep 93 |
EVIDENCE |
| National Transportation and Safety Board |
Scandinavian Airlines DC10-30 accident, New York City, NY
USA, 28 Feb 84 |
EVIDENCE |
| Aircraft Accident Investigation Committee - Nepal |
Thai Airways International A310 accident, Kathmandu, Nepal,
31 Jul 92 |
EVIDENCE |
| National Transportation and Safety Board |
Trans World B707-331C accident, Jamaica, NY USA, 12 Dec 72 |
EVIDENCE |
| National Transportation and Safety Board |
World Airways DC10-30 accident, Boston, MA USA, 23 Jan 82 |
EVIDENCE |
| National Transportation and Safety Board |
Controlled Flight Into Terrain, Korean Air Flight 801, Boeing 747-300, HL7468, Nimitz Hill, Guam, August 6, 1997 |
EVIDENCE |
| National Transportation and Safety Board |
In-Flight Icing Encounter and Uncontrolled Collision with Terrain, COMAIR Flight 3272, Embraer EMB-120RT, N265CA, Monroe, Michigan, January 9, 1997 |
EVIDENCE |
| National Transportation and Safety Board |
Wheels-Up Landing, Continental Airlines Flight 1943, Douglas DC-9 N10556, Houston, Texas, February 19, 1996 |
EVIDENCE |
Accident Review Results
In addition to accident reports prepared by official investigating boards, we included
several accident reviews in our study. These were reviews conducted by qualified
individuals after the official investigations, which benefited from additional information
and the perspective offered by the individual's field of technical expertise. Although we
assigned lower strength ratings to evidence from such accident reviews, they were
nevertheless a valuable supplement to evidence from the accident reports.
The following is a list of the accident reviews in which we found evidence related to
the flight deck automation issues. The list includes links to the bibliographic
information and evidence found in the reports:
| Investigator(s) |
Short Description of Accident Review |
|
|
| Bruseberg, A. |
In order to design interaction with computer systems such as modern Flight Management Systems, we may require new interaction paradigms, since such systems become increasingly able to carry out tasks partly automatically and in parallel to the human operators who are given increasingly supervisory functions. By creating a clearer understanding of the extent to which such interaction could, or should, be viewed as a type of collaboration, we aim to formulate a new set of design principles. Cali Accident is discussed. |
EVIDENCE |
| Bruseberg, A. & Johnson, P. |
This paper illustrates how the perspective of considering the interaction between the Flight Management System and the (pilot) user as a collaborative activity can uncover design opportunities - by conducting an analysis of interaction failures that contributed to the Cali air accident. |
EVIDENCE |
| Endsley, M.R. & Strauch, B. |
Examines contributors to the loss of situation awareness by pilots in American Airlines Boeing 757 Cali, Colombia accident. |
EVIDENCE |
| Hourizi, R. & Johnson, P. |
In this paper, we show that errors, which occur within such complexity, cannot easily be described in terms of individual tasks and their component actions. We use this starting point to examine the dominant thinking in this field (Palmer, 1995, Palmer et al, 1993, Degani et al, 1996, Rushby 1999). We show this dominant position, which suggests that much of this ’automation surprise’ (Palmer 1995) results from mode error, is a fundamental misclassification of the human factors involved. We then show the existence of a deeper problem, which we identify as a knowledge gap (Johnson 1992) between operator and system. |
EVIDENCE |
| Mellor, P. |
Analysis of accident sequences in which computers have been
wholly or partly to blame |
EVIDENCE |
| Strauch, B. |
Lessons learned about automation and decision making from the Cali accident. |
EVIDENCE |
|