American 757 Crashes

[Flight International, 3-9 January 1996, p5]

An American Airlines Boeing 757 which left Miami for Cali, Colombia, on 20 December crashed into mountains at night, killing all but four of the 167 people on board. The aircraft was on its descent into Cali from the north, which requires a step-letdown procedure using VHF-omni-range/distance-measuring-equipment navigation beacons. The aircraft hit a 12,000ft (3,660m) mountain near the town of Buga, at about 9,000ft and around 18km (10nm) to the east of the letdown track. Cali is in a steep-sided, north/south-orientated valley, with its runway aligned for the terrain (01/19), and the letdown procedure almost straight. The Colombian Civil Aviation Authority says that all navigation beacons were serviceable, the flight-data recorder and cockpit-voice recorder readouts implied no technical problems, and that the crew reacted to a ground-proximity warning-system alert, but too late to save the aircraft. There is no indication as to why the 757 was off track.

Air safety takes a dive

by David Learmount [excerpts, Flight International, 10-16 January 1996, p8]

[..] The American 757 crashed into a 12,000ft (3,600m) mountain near Cali, Colombia, killing all 160 people on board. The accident has already been attributed by the US and Colombian authorities to aircrew error. Now US Federal Aviation Administration Administrator David Hinson has ordered a review of American Airlines' pilot training and cockpit procedures. Colombian accident investigators confirm that the 757 crew failed to carry out a pre-descent approach briefing or checks. Cali Airport is in a steep-sided north/south-orientated valley.

The co-pilot first called Cali air traffic control (ATC) when the flight was 63 miles north of the Cali VHF omni-range/distance measuring equipment (VOR/DME) navigation beacon, descending to 20,000ft. He received clearance to descend to 15,000ft and proceed direct to the Cali beacon. This is south of Cali airport and is the key beacon in a VOR/DME step-letdown to the runway (01/19). The northbound runway direction (01) is normally used because it has an instrument-landing system. The 757's autopilot was engaged and the flight-management system (FMS) selected to the lateral navigation mode. ATC has no effective surveillance radar once aircraft reach mountaintop level - about 15,000ft.

Because the airport wind was calm, the controller offered the crew an approach to the southbound runway (19). The crew accepted, and was cleared for "the VOR/DME 19 Rozo One arrival", and told to report reaching Tulua VOR. Tulua, 63km (34nm) north of Cali Airport, marks the start of the Rozo One VOR/DME approach; Rozo is a non-directional navigation beacon on the approach 5km north of Cali Airport.

The aircraft, however, had already passed the Tulua beacon, but the pilots apparently did not realise this.

The crew requested clearance direct to Rozo, and then "...do the Rozo arrival". ATC had said, after some further exchanges, "...affirmative, direct Rozo One and then runway 19", but repeated that the crew had to report over Tulua first.

ATC advised that, after Tulua, the approach consisted of reaching 5,000ft by 21nm DME [from the Cali beacon].

Meanwhile the crew continued descent through the cleared 15,000ft, with airbrakes selected to increase the rate of descent. When they finally entered Tulua on the FMS, the aircraft began a fatal 90s left turn (initially eastward) to return to the beacon.

Then the pilots selected heading mode on the FMS and turned the aircraft right. The aircraft hit the mountain with the ground-proximity-warning system ordering "pull-up". The airbrakes were still deployed.

Off target

excerpts from the 1995 Airline Safety Review by David Learmount [Flight International, 17-23 January 1996, pp24-34]

[..] the biggest shock of 1995 was the American Airlines Boeing 757 CFIT crash in December near Cali, Colombia. This involved an airline with one of the best safety records in the world and an aircraft type, which, in its 13-year operational history, had not suffered a fatal accident before. Yet the initial reports of the Colombian Civil Aviation Authority and the US National Tranportation Safety Board, based on readouts from the 757's cockpit-voice recorder (CVR) and flight-data recorder, have stated categorically that the crew failed to carry out pre-descent briefings or checks. This would be a breach of standard operating procedures under any circumstances, but, when operating at night into an airport surrounded by mountains, the omission seems extraordinary to line pilots.

As with so many accidents, the chain of events leading to the Cali crash was started by an apparently innocuous happening. The crew were expecting a particular approach procedure on to the northbound runway (01), with which they were familiar, because it is the main runway and the only one with an instrument-landing system for the final approach. Because the wind at Cali was calm and the American flight was approaching from the north, Cali air-traffic control (ATC) offered the crew the option of carrying out the VHF omni-range/distance measuring equipment let-down to reciprocal runway 19, using a specified arrival procedure.

The crew accepted the proposed arrival, and then set about looking for the appropriate charts while still descending with airbrakes extended. From the limited CVR transcripts published by the Colombian CAA, it would appear that the crew were not familiar with the approach they had accepted and did not have the charts to hand, yet they continued their descent with disastrous results (Flight International, 10-16 January, P8). If the crew had not been offered the alternative approach, the accident would not have happened, but there is no suggestion from the official information so far released that the ATC could be criticised for offering the choice.

During the investigation, it will become apparent whether the much-discussed human-factors effects of having ultra-smart flight-management systems (FMS) and clear lateral navigation displays (NDs) will be shown up in the clear relief which only disaster seems to bring. The crew were certainly relying on their ND, through use of their FMS, to provide them with the answers they sought when confused by the new let-down procedure offered by ATC. Meanwhile, the FMS was entrusted with the aircraft's flightpath and, "obedient but dumb", as one US pilot commentator remarked, turned the aircraft toward a beacon which the pilots believed was still ahead of them, but which, in fact, they had passed. That turn initially took the aircraft eastward toward the mountains, and the pilots acquiesced for 90s, according to the interim report, before they decided to take charge of the aircraft's heading themselves. It was too late.

Modern-screen NDs have the potential to provide better situational awareness for crews than the instruments they replace and, combined with the accurate flightpaths and performance prediction which FMS can provide, are definite safety benefits. The US Federal Aviation Administration, together with other agencies, is continuing to study the overall effects of "glass cockpits" on aircrew performance. The Cali incident will give the inquiry some food for thought.

The fact that the American crew had, however, begun their descent without the appropriate checks and procedures, has clearly been a factor in FAA Administrator David Hinson's statement that American's training procedures are to be investigated. Another quoted factor had been the accident at Bradley International Airport, Connecticut, USA, where an American Airlines McDonnell Douglas MD-83 escaped disaster by a hairsbreadth when it flew much too low on the approach. The aircraft hit trees on a ridge, the engines lost power as they ingested debris, and the pilot managed just to get the aircraft into the airfield. No-one was hurt.

The FSF says that the moral of the story is that it does not matter how good an airline's safety record is, CFIT can still happen unless constant awareness of its multiple causes is a specific part of the training programme's aim. [..]

Back to top

Back to the `Incidents and Accidents' Compendium

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