David Learmount/LONDON
World airline accident fatalities increased in 1994, compared with 1993, and exceeded the decade annual average. The increase is an insignificant variation in the context of annual figures during the past ten years, but the message emerging from the statistics is shocking for the insurance community, which faces estimated claims of more than $2 billion, according to London-based insurance loss-adjuster Airclaims.
The projected hull and liability losses are nearly double those for 1993 and constitute an all-time record, says Airclaims. The fact that five of the fatal passenger-service accidents - two jet-powered airliners and three turboprops - involved US domestic operations clearly boosts the expected liabilities payout.
The 1994 world total of fatalities from airline incidents rose by 25%, compared with 1993, reaching 1,385, but the number of fatal accidents dropped from 48 to 47. Fatalities also exceeded the decade annual average (1,143) by 21% and the number of accidents topped the average (42) by 12%.
That may sound bad, but, viewed among the decade peaks and troughs (see chart), it seems unremarkable. The significance of the numbers also comes into sharper focus when viewed against continual industry growth.
Also, if non-passenger-service (freight-, positioning-, ferry- or airliner-training flights) fatal accidents are omitted from the lists for the last two years, passenger-service fatal-accident totals have dropped from 35 for 1993 to only 28 in 1994.
Looking back to 1985's totals provides a highly favourable comparison (see chart), but is misleading because 1985 was the worst year in the history of air-transport safety, with 1,800 accident fatalities from 39 accidents. Since then, however, according to International Civil Aviation Organisation figures, the world commercial air-transport fleet has grown by 56%, passenger/kilometres travelled have increased by 66%, passenger numbers are up by almost 75% and aircraft movements rose by at least 33%.
Significance is usually sought in trends, but the industry is equally concerned about statistics which do not change. These can represent undesirable ills which are not responding to treatment. Initial analysis of 1994 shows that aircrew-error* appears to be a factor in 31 of the fatal accidents, or 67% of the total; weather was a factor in 16 fatal accidents (34%), and 14 flights (29%) ended with controlled flight into terrain (CFIT)*. These figures sound depressingly familiar, despite industry efforts to lower them.
The major airline-accidents of 1994 do not deliver a strong collective message, but individual studies are revealing. In the 3 January Baikal Air disaster (see accident listings), the aircrew ignored a starter-system fault warning, and 124 people died as a result. Behind this apparently simple error, however, is a litany of problems relating to confusion over which company, in the new Russian commercial environment, is responsible for bearing the cost of component failure - in this case the starter motor. Is it the airline, the manufacturer, or the aircraft design bureau? Even more importantly, according to accident investigators, lines of communication between the operator, manufacturers and the design bureaux are poor or non-existent, so designs or components which need modification do not receive it.
The 6 June China Northwest accident was caused directly by a serious maintenance error. According to the investigators' early reports, the cross-wiring of the autopilot control channels was so catastrophic in its effects that the pilots cannot be blamed for failing to act fast enough to save the aircraft. Accident records show that maintenance errors which initiate a chain of events - usually manageable - causing an incident are not uncommon, but those which are so serious that they cause an unpreventable accident are rare. The Civil Aviation Authority of China report may yet reveal how the supervision and checking system allowed this to occur.
A host of issues arise from the 26 April China Air Lines (CAL) accident at Nagoya Airport, Japan. Early assessments indicate that the primary cause is likely to be judged as pilot error* at several stages. The first mistake was in making an inappropriate flight-mode selection (if inadvertent pilot-selection of go-around mode is confirmed in the final report); the second miscalculation was in failing to recognise that the aircraft was in go-around (G-A) mode; and the final error was a lack of understanding of the consequences of attempting to override the autopilot by using physical force on the control column, rather than selecting disconnection of the autopilot.
Blaming the pilots is the easy part of the judgement, however. The first issue raised, therefore, is pilot-training effectiveness, especially in systems knowledge and understanding. The more complex issues raised by Nagoya include:
autopilot or flight-management-system (FMS) "mode-confusion". This has been shown to arise frequently in pilots - not only in Airbus aircraft;
the logic which requires Airbus Industrie aircraft stabilisers to move to counteract pilot-elevator input when the autopilot is engaged with G-A or automatic-landing modes selected seems "unnatural" and, therefore, potentially confusing, yet it is the result of certification authority requirements (Flight International, 11-17 January, P24);
finally, if a modification to the autopilot software - about which Airbus distributed a service bulletin (SB) in June 1993 - had been incorporated by CAL, would have prevented the accident, but the SB instructions had not been carried out by the airline.
FAA ACTION
The Nagoya accident, along with other examples of mode-confusion, caused the US Federal Aviation Administration to issue an airworthiness directive in October 1994 requiring US-registered aircraft to incorporate within 60 days the autopilot software modification described in the Airbus SB.
Meanwhile, the FAA is instituting a timely review of all new-generation cockpits as effective man-machine interfaces (Flight International, 4-10 January, P9). The "glass cockpit" has been in commercial operation now since the Boeing 767 entered service in 1992. The international review team, says the FAA, will look not just at the presentation of data and information, but at the human-factors effects of increased automation made possible by progressively more-capable digital FMS.
The FAA-led review will have plenty to look at. In the non-fatal incident lists on the next few pages are examples of the sensitivity - perhaps over-sensitivity - of digital flight management or flight-control systems, and confusing autopilot behaviour. Examples among the scheduled-passenger incident listings include: 3 March, Air Lanka Airbus A320; 15 May, Virgin Atlantic Boeing 747-200; 6 June, Dragonair A320; 24 September, Tarom Airbus A310. An example in the non-scheduled passenger list is the 21 June incident involving a Britannia Airways Boeing 757.
The USAir Pittsburgh accident on 8 September, for which no clues so far exist, echoes two other unexplained 737 accidents: the 30 March 1993 Thai air force 737-300 incident at Muang Khon Kaen, Thailand, and the 3 March United Airlines 737-200 crash at Colorado Springs. All involved loss of control (Flight International, 12-18 October, 1994, P40) and this has brought together an unprecedented concentration of inquiry effort. In addition to the normal US National Transportation Safety Board (NTSB) inquiry into the Pittsburgh accident, the FAA is running a critical design review of 737 control systems, due to report this month; and, finally, the NTSB is running a public hearing to review the accident and surrounding issues, such as training, and the adequacy of current regulations on flight-data recorders (FDRs) and other systems for flight analysis, such as cockpit cameras.
The FAA has said that it believes the 1994 statistics will show that the US accident rate per 100,000 departures was 0.25 for Part 121 (major) operators and 0.35 for commuters. In terms of the number of accidents (rather than fatalities caused), the agency says that 1994 was the commuter industry's safest ever.
The year was a busy one for safety resolutions and the clarification of safety standards. The FAA has been a leader throughout. The USA banned several foreign airlines from operating into its airports until the regulatory authorities in their countries had manifestly improved their safety-surveillance capability and practice. With about 60 countries' safety-policing standards still under review by the FAA, the USA has served notice to the world's nations that simply being a signatory of the Chicago Convention and thus a member state of the International Civil Aviation Organisation (ICAO) does not guarantee national airlines the right of access to the USA. The European Civil Aviation Conference nations say that they are considering applying the same sanctions as the FAA does, with the intention of redressing the manifest safety-standards imbalance between various regions of the world.
The ICAO itself, the central definer of standards, also declared that the rules requiring international openness in accident investigation and the sharing of information (Chicago Convention Annex 13) had been updated, strengthening their requirements (Flight International, 23-29 November, 1994, P4).
Meanwhile, the Flight Safety Foundation has continued with its programme of research into practical ways of reducing CFIT accidents, and has stated that the use of global-positioning-system satellite navigation instead of existing non-precision approach aids could halve the CFIT accident rate (Flight International, 9-15 November, 1994, P11). With the same aim in mind, ICAO is drawing up proposals for tighter regulations on the fitment and capabilities of ground-proximity warning systems in aircraft.
ICAO, a division of the United Nations, operates by consensus and persuasion, having no direct power to wield, but it is clear that 1994 has been a year in which notice has been served on nations which do not take flight safety seriously.
*Aircrew error could be a direct mistake, or an inappropriate pilot strategic judgement, or pilot failure to intervene correctly to save the flight; CFIT does not refer only to collisions with high ground or rising terrain, but also to collision with level ground (eg: on the approach) when the aircraft is fully under control.
Notes
Accident data have been gathered from Flight International's own research, from Airclaims, and from Lloyds Aviation. Airclaims now researches and publishes the World Accident Summary on behalf of the UK Civil Aviation Authority, and it has an exclusive agreement on co-operation and information exchange with the CIS Interstate Aviation Committee Commission for Flight Safety. The latter has improved our ability to report on incidents in the CIS.
Flight International has received some criticism for publishing details of non-fatal incidents because this type of information is not made available in most parts of the world. We intend to continue, however, because airlines tell us that there is much to be learned from those relatively few incidents that we list.
We accept that the non-fatal listing is unfairly weighted against UK and US airlines because information on their incidents is made available more readily.
Abbreviations
AA Airfield approach/early descent; ATC Air traffic control; C Climb; CVR Cockpit-voice recorder; DME Distance-measuring equipment; ER En route; FDR Flight-data recorder; FMS Flight-management system; G On ground; HP High pressure; ILS Instrument-landing system; L landing; MTOW Maximum take-off weight; RA Runway approach; TO Take-off.
Source: Flight International
