Shades of AA crash at Cali
A Qantas B737-800’s close encounter with high terrain in Australia at 5:44 a.m. on July 24, 2004, could have ended in disaster, illustrating the value of the enhanced ground proximity warning system (EGPWS).
In a separate incident, the accident crew of the American Airlines B757 plane that crashed at Cali, Colombia, in 1995 did not have the extra warning provided by EGPWS and died as a result. That accident triggered the widespread deployment of EGPWS.
The Qantas incident shows clearly a line of incident causation that may be clearer in retrospect, and it has prompted changes at the carrier to avoid a repeat.
The case was investigated by the Australian Transportation Safety Board (ATSB), from which this account is derived.
As the Qantas B737 approached the nation’s capital at cruising flight level, its crew repeatedly asked for descent until, finally, once past their flight management system’s (FMS) computed descent point, en route air traffic control (ATC) admitted that no one knew the duty runway at Canberra because the airfield’s ATC (tower and approach) would not be opening for business for some time yet.
Faced with this unexpected development, the crew opted to descend using mandatory broadcast (MBZ) procedures and organize in the standard holding pattern for an anticipated Runway 35 instrument landing system (ILS) approach – once it became available. In comparison to the American Airlines B767 terrain accident at Cali, there was no language problem and no Jeppesen induced error in the navigation database. It was just that the crew had no radar surveillance and no ATC. It was also dark, and in the pre-dawn hours when circadian dysrhythmia has human performance at its natural nadir.
Additionally, the crew had spent much of the 3.5 hour flight from Perth troubleshooting an overheating flight-deck air conditioning and had been experiencing uncomfortably hot temperatures. Security rules prohibited them from opening the flight-deck door to alleviate their discomfort. The system had been troublesome for a number of recent flights (in fact over six flights since July 21, 2004) and had been accepted under Minimum Equipment List (MEL) rules. Alertness and situational awareness was compromised by both the hour and the exposure to an inhospitable heat-stressed environment.
A level of protection had been removed by the unanticipated unavailability of ATC approach radar services. When the aircraft was about 45 nautical miles to the west of Canberra, the Melbourne Center (ML) controller advised the crew that Canberra Approach Control would not be coming on line “due to staff shortages” and that the ML radar vectors would not be available below 9,000 feet for their instrument flight rules (IFR) arrival. The crew needed to quickly replan the approach to suit the new arrival scenario.
The approach
The holding pattern was anchored to the Church Creek Locator (CCK) located on the ILS localizer (LLZ) about 10.9 nautical miles south of the Canberra VOR (very high frequency omnidirectional radio range) approach – a terminal navigation aid that was located on the airfield. The published CCK holding pattern requires that aircraft holding at 5,000 feet observe a maximum indicated airspeed (IAS) of 170 knots and limit time on the outbound leg to either one minute or a distance measuring equipment (DME) limit of 14 nautical miles from Canberra VOR/DME, whichever is reached first.
As the aircraft approached CCK, the first officer, at the direction of the pilot in command, entered the holding pattern details into the Flight Management Computer (FMC). In doing so, an erroneous entry was made, which resulted in the FMC computing a holding pattern based upon CCK but with a leg length of 14 nautical miles to the south from CCK, instead of one minute or a maximum distance from Canberra DME of 14 nautical miles. This error would take the aircraft into the mountain ranges southwest of Canberra where the peaks rose to a maximum of 7,310 feet.
By also entering a leg distance of 14 nautical miles, the crew inadvertently cancelled the one minute leg-time and commanded the FMC to establish the aircraft in a holding pattern that would take it about 11 nautical miles beyond the published holding pattern limit. The crew initiated descent to the initial approach altitude of 5,000 feet after passing over CCK. As it descended, the aircraft proceeded outside the airspace specified for holding. In the descending inbound turn onto the LLZ, the aircraft passed within 2.7 nautical miles of Tinderra Peak (elevation 5,310 feet) and received an EGPWS ground proximity system warning. The crew reacted by climbing to 6,500 ft. and continued on the 35 ILS to land.
It was only after the crew had completed a belated descent/approach checklist that they sensed that the holding pattern “did not seem right.” At that time, the pilot in command selected “Terrain” mode on the navigation display and the copilot checked the DME distance. Very shortly afterward, the pilot in command initiated the holding pattern’s inbound turn back toward CCK. This was just prior to the terrain alert from the EGPWS.
Factors in creating the circumstance
At the time of the occurrence, the aerodrome and approach air traffic services were normally available from 0530 to 2200 local. Outside those hours, the aerodrome reverted to a non-controlled mandatory broadcast zone, requiring that pilots provide their own separation from other aircraft and from terrain. On the day of the occurrence, the rostered air traffic controller did not report for duty at 0515 as assigned. Consequently, the Canberra Approach Control service was not available until approximately 40 minutes after the scheduled opening time of 0530 – when a relief controller was located. A manning failure contingency plan promulgated by Airservices Australia three days prior to the incident was not implemented on the morning of the occurrence – presumably because of a failure of initiative.
The minimum sector altitude within 10 nautical miles of the Canberra VOR navigation aid was 5,100 feet, but due to the mountainous terrain, particularly south of Canberra, was 7,400 feet within 25 nautical miles from the VOR, in the sector between the 080� radial and the 250� radial.
The MEL of the flight deck air conditioning system allowed continued flight operation despite abnormally hot conditions. Consequently (and subsequent to the incident) the aircraft was flown to Sydney where maintenance engineers were again unable to rectify the defect. The engineers therefore made an entry in the aircraft’s technical log advising crews to “keep the trim air pressure regulating and shutoff valve in the closed position.” That had the effect of isolating the section of the air conditioning system associated with the faulty temperature controller, thereby preventing the entry of 60� C (140� F) hot air to the flight deck ducts.
The crew misinterpreted the holding pattern DME limit distance on the referenced instrument approach chart. The crew did not detect that the DME distance referenced on the chart was based on the Canberra DME. Contributing to that was the instrument approach chart not containing the specific Canberra DME identifier in the CCK holding pattern limits. The ATSB report contained no information as to whether either pilot had flown the 35 ILS approach (and its hold) in the simulator – presumably a good idea for routes regularly flown by individuals.
Data recovery
Data was recovered from the EGPWS non-volatile memory and radar data. EGPWS recorded data covers a duration of 30 seconds; 20 seconds before an event and 10 seconds after. However flight data recorder (FDR), cockpit voice recorder (CVR) and quick access recorder (QAR) data were unavailable. No information relating to the occurrence flight was available from these recorders because the occurrence flight data had been overwritten by the time ATSB investigators were notified and began their investigation. The aircraft’s quick access recorder was found to be unserviceable and had not recorded the occurrence flight. The incident obviously eluded the airline’s Flight Operations Quality Assurance (FOQA) program also.
Fixes put in place
The following has happened since the incident:
- Jeppesen Sanderson Inc. advised the ATSB that they intend to include the DME identifier in the holding pattern limit notes on relevant charts.
- Three days after the occurrence, Airservices issued a new local order that instituted clearer instructions for implementing the ATC manning failure contingency plan.
- Qantas crews’ attention was drawn to the requirements and guidelines concerning the submission of Air Safety Incident Reports, which were detailed in Chapter 3 of their Flight Administration Manual (FAM).
- Qantas now requires the removal from service of both the FDR and the QAR when it is believed that there will be a need for incident data, and
- Qantas raised the minimum holding altitude at CCK by 1,000 feet and issued the following Flight Standing Order clarification to 737 flight crews:
B737 FMC HOLD PAGE
A recent incident has indicated that there may be some confusion relating to the function of the Leg Distance (LEG DIST) prompt of the B737 FMC Hold page.
The Leg Distance prompt allows entry of the actual length of the inbound Leg of a holding pattern in nautical miles. It does not refer to a DME limit as depicted on a charted holding pattern. Additionally, beware that a Leg Distance entry will override a Leg Time value.
The incident demonstrated the value of EGPWS in breaking an accident chain and the perils of accepting a debilitating MEL. It also perhaps indicates that attention is required of simulator instructors in checking knowledge of FMC “gotchas” and in providing more challenging (and less prosaic) hazards encounter training.
(The complete ATSB report is available at
http://www.atsb.gov.au/aviation/occurs/occurs_detail.cfm?ID=659 )