COVER STORY
Can courtesy
kill?
26 FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
Embarrassment about making a mistake, reluctance to
CAPTAIN’S FEARS that his company would frown on him if he conducted a missed approach is one possible factor in the fatal crash of a Gulf Air A320 into the shallow waters of the Arabian Gulf. The failure of the first officer, a shy, polite type, to take issue with the captain’s decision to execute a 360-degree orbit instead of a missed approach and go around, is another possible factor in the accident, near Bahrain International Airport on August 23, 2000. However, as with many aviation disasters, the crash of GF-072 represents a failure of the aviation system. The investigation revealed a complex series of human, organisational and management influences that combined to set the scene for the accident, which culminated in the crew’s failure to respond to insistent warnings to “pull up” from the aircraft’s ground proximity warning system (GPWS). A multinational team of accident investigators set up by the Kingdom of Bahrain in accordance with international civil aviation agreements∆ found no technical deficiencies in the aircraft or its systems. The investigators attributed the tragedy mainly to human factors at the individual and organisational level. They turned up evidence of errors and procedural violations committed by the flight crew, and of long-standing organisational and management problems that had been identified but not rectified. Gulf Air has since acted on many of the recommendations made in the Accident Investigation Board’s report on the crash and plans to implement the rest. It is establishing a new integrated safety management system, and has beefed up its safety department, while stepping up internal safety promotion. Gulf Air flight GF-072 was a scheduled service from Cairo International Airport to Bahrain International Airport (BAH). It was operated by an Airbus Industrie
A
challenge a captain’s decision and spatial disorientation are possible factors in an accident that killed 143 people. But, as John Mulcair and Rob Lee report, there is much more to the story.
PHOTO: AAP
FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
27
COVER STORY
DIETMAR SCHREIBER
A320-212. There were two pilots, six cabin crew and 135 passengers on board. When GF-072 was about one nm from touchdown and at an altitude of about 600 ft, Bahrain air traffic control approved the crew’s request for a 360-degree left-hand orbit. On completion of this manoeuvre, the aircraft was about parallel to but beyond the extended centreline of the landing runway, and the crew initiated a missed approach. With a radar vector offered by ATC, GF-072 overflew the runway in a shallow climb to about 1,000 ft. The aeroplane crashed into the sea about four kilometres north-east of the airport at about 1930, Bahrain local time. The debris field was 700 m long and 800 m wide. Most of the aeroplane was recovered, along with all significant structural components, flight control surfaces and both engines. There was no evidence of pre-crash failure or fire damage. The crew GF-072’s last flight began with the arrival of its crew at the gate 25 minutes before the scheduled departure time of 1600. The airliner was under the command of a 37-year-old captain who had joined Gulf Air in 1979 as a cadet flight engineer. He had later retrained as a pilot, flying on the Boeing 767 and Airbus A320 as a first officer, and, since 1996, as a supervisory first officer. He was promoted to captain on the Airbus A320 on June 17, 2000. He had logged total pilot time of 4416 pilot hours, 86 of which were as pilot in command on the A320. Gulf Air pilots who had flown with him described him as responsible, knowledge-
able, open to suggestions, happy, very helpful, professional and sharp. They differed on whether he was overconfident. Gulf Air hired the first officer, aged 25, as a cadet on July 4, 1999, after he attended its ab-initio training program. He was promoted to A320 first officer on April 20, 2000. He had accrued a total of 608 pilot hours, made up of 200 hours in training, and 408 hoursrs as an A320 line pilot with Gulf Air. Gulf Air pilots who had flown with the first officer described him as timid, meek, mild, polite, shy and reserved in social situations, and keen to learn. While most felt his reserved nature would not stop his speaking up during flight operations, others felt he might have been too reserved to challenge a captain. The crash After an uneventful flight from Cairo, the aircraft was prepared for a visual approach and landing on Runway 12 at Bahrain. At the time, Runway 12 had no instrument landing system. The weather was fine, and the night was clear and dark with no moon. The conversation and sounds in the cockpit for the 30 minutes before the accident were recorded on the cockpit voice recorder (CVR). At 1926:37, the captain stated: “OK, visual with airfield”. Seconds later, the flight data recorder (FDR) showed that the autopilot and flight director were disengaged. At 1926:49 and about 2.9 nm from the runway, the aircraft descended through 1,000ft. At 1926:51, with GF-072 about 2.8 nm from the runway, at an altitude of 976 ft
and a speed of 207 knots, the captain said: “Have to be established by 500 feet.” Flaps two were selected. As the approach to Runway 12 continued, the captain said at 1927:06, and again at 1927:13, “….we’re not going to make it”. At 1927:23, he instructed the first officer to “tell him to do a 360 (degree) left (orbit)”. Bahrain tower approved the request. The left turn was initiated about 0.9 nm from the runway, at an altitude of 584 ft and an airspeed of 177 knots. During the left turn, the flap configuration went from flaps two to flaps three and then to flaps full. At 1928:17, the captain called for the landing checklist. At 1928:28, with the Airbus about halfway through the left turn, the first officer advised that the landing checklist was complete. After completing about three-quarters of the 360-degree turn, the aircraft rolled wings level. The Airbus’ altitude during the left turn ranged from 965 ft to 332 ft, while its bank angle reached a maximum of about 36 degrees. At 1928:57, after being cleared again by Bahrain tower to land on Runway 12, the captain stated: “We overshot it.” The aircraft began to turn left again, followed by changes consistent with an increase in engine thrust. At 1929:07, the captain said: “Tell him going around.” The FDR indicated an increase to maximum take-off/go around (TOGA) engine thrust. Bahrain tower provided radar vectors, with instructions to “fly heading three zero zero (300 degrees), climb (to) 2,500”.
The A320 that crashed into the Arabian Gulf in 2000, pictured a year before the accident.
28 FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
COVER STORY Perceived pitch
Actual pitch
Perceived pitch versus actual pitch Just before the captain pushed the sidestick forward (t=1929:43). Source: Accident investigation report Gulf Air Flight GF-072. See http://www.bahrainairport.com/caa/gf072.html.
The auto thrust remained active throughout the approach until TOGA was selected. The flaps were moved to position three and the landing gear was selected up. The gear remained retracted until the end of the recording. At 1929:41, with the aircraft at an altitude of 1,054 ft and an airspeed of 191 knots, and having just crossed over Runway 12, the CVR recorded the beginning of 14 seconds sounding of the repetitive chime of the aural master warning consistent with a flap overspeed followed by the first officer saying, “speed, overspeed limit …” About two seconds after the master warning began, and with the aircraft still accelerating under TOGA power, the FDR data indicated movement of the captain’s side stick, which was held forward of the neutral position for some 11 seconds, with a maximum forward deflection of about 9.7 degrees. During this time the aircraft’s pitch attitude decreased from about five degrees nose up to about 15.5 degrees nose down. The recorded vertical acceleration decreased from about +1.0 to about +0.5 G, while
airspeed increased from about 193 to about 234 knots. About 1929:51, with the aircraft descending through 1,004 ft at an airspeed of 221 knots, a single aural voice warning of “sink rate” from the GPWS was recorded, followed by the repetitive GPWS aural warning “whoop whoop, pull up”, which continued until the end of the CVR recording. At 1929:52, the captain requested, flaps up. About 1929:54, the CVR indicated that the master warning ceased for about one second but began again and lasted about three seconds. Two seconds after the GPWS warnings began, the captain’s side stick was moved aft of the neutral position, with a maximum aft deflection of some 11.7 degrees. These data showed this nose-up command was not maintained and subsequent movements never exceeded 50 per cent of full-aft availability, and the aircraft continued to descend. FDR data indicated no movement of the first officer’s side stick throughout the approach and accident sequence.
At 1929:59, the captain requested, “Flaps all the way” and the first officer responded, “ Zero!” This was the last comment from the crew recorded on the CVR, which stopped recording at 1930:02. The FDR data showed continuous movement of the flap position toward the zero position after the captain’s “Flaps up” command. The last flap position recorded on the FDR was about two degrees of extension. The investigation To find out what visual cues the pilots had, investigators retraced the flight path of GF-072 in a helicopter. The flight was carried out at night, in meteorological and visual conditions similar to those on the night of the accident. The reconstructed flight path was recorded on video. The cockpit view calculations of the field of view from the A320 cockpit, supported by the video reconstruction of the flight path, indicated that all external visual cues were lost about 1629:41 as the last lights on the ground passed under the nose of the aircraft. The forward sidestick input by the captain started at 1629:45, when the aircraft was accelerating into complete darkness. Somatogravic illusion The crew would have been vulnerable to a kind of spatial disorientation known as the somatogravic illusion. The absence of visual cues combined with rapid forward acceleration and the force of gravity create a powerful pitch up sensation. In such cases, particularly on dark night takeoffs, pilots often respond by lowering the nose. In some cases, the aircraft descends and hits the ground, usually at a shallow angle of impact. The US Naval Aerospace Medical Research Laboratory used the FDR data from GF-072 in a perceptual study. At the time of the captain’s forward sidestick input at 1929:45, he would have experienced a pitch-up sensation of about 12 degrees, the study showed. The application of forward sidestick input by the captain for 11 seconds resulted in the aircraft pitching down to an angle of 15 degrees, which is the maximum pitch-down angle allowed by the A320 flight control system. This would have almost cancelled out the perceived pitch-up sensation. In the absence of any external visual cues, and with its attention probably focused on the flap overspeed, the crew probably believed it was in near-level flight. The cockpit instruments would have been displaying the true pitch attitude of the aircraft. However, the captain, as pilot flying, did not use this source of information, suggesting that he did not perceive the attitude information from his Primary Flight Display.
FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
29
ACCIDENT INVESTIGATION REPORT GULF AIR FLIGHT GF-072
COVER STORY
Other research after the accident included studies to determine the effects of certain variables on altitude loss during GPWS recovery, simulations of the approach, orbit and go-around of GF-072 at BAH, and a series of flight tests. Variables examined in the GPWS recovery study were the amount of the pilot’s pitch-up command, the time between the GPWS warning and the pilot’s reaction to it, and the duration of the pitch command input. An A320 fixed-base engineering simulator at Airbus Industrie’s facilities at Toulouse, France was used to simulate the approach, orbit and go-around of GF-072. The simulator also allowed investigators to fly the approach to Runway 12 and to observe cockpit warnings during flap overspeed and GPWS warnings. During one of the simulator sessions, the 360-degree turn and go-around manoeuvres were performed to approximate the flight path and the sequence and timing of events recorded on the FDR recovered from the aircraft.
The team considered several scenarios: • The pilots were instructed to recover with full aft stick movement at the onset of the GPWS alert. The simulator recovered with about 300 ft altitude loss. • Half back stick was applied instead of full back stick. The delay between the GPWS warning and the stick command was approximately four seconds. The simulator recovered with about 650 ft altitude loss. • The co-pilot performed a recovery after he had verified that the captain had taken no action to recover from the GPWS alert. The co-pilot depressed the priority button on his side stick, announced his control override, and applied full aft side stick input. The simulator recovered with about 400ft of altitude loss. • The 360-degree turn was performed but the pilots were instructed to make no further control inputs after selection of TOGA power. The simulator trimmed nose down to counter the noseup effect due to the thrust
30 FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
increase and to maintain +1.0G, the target when the side stick is in the neutral position in Normal Law. The pitch remained positive and the aircraft climbed slowly. • The 360-degree turn was initiated to match the flight path and sequence and timing of events recorded on the FDR. But instead of rolling wings level upon reaching a heading of about 211 degrees magnetic, as the captain of GF-072 had done, the 360-degree turn was continued at a moderate bank angle at the pilot’s discretion to align with Runway 12, and the approach and landing were continued. The pilots were able to successfully land on Runway 12 from the 360-degree turn. In this final scenario, the pilots noted that the approach was not stabilised and little time was available to successfully complete the final approach and landing. On September 27, 2000 a flight demonstration in an A320 test aircraft observed various conditions similar to the flight profile flown on August 23, 2000. It was
COVER STORY flown in daylight in visual meteorological conditions. Additional tests were performed to simulate the 360-degree orbit of the accident flight, but continuing to turn at the end of the orbit instead of rolling out. Several scenarios were flown, with a similar flap configuration sequence to that in GF-072, or with full flaps. The pilots were able to align the aircraft with the runway and perform low approaches down to 50ft where a go-around was performed. With no evidence that the accident was caused by aircraft technical problems, the investigation focused on human factors. Investigators analysed the role and performance of individuals as components of a system. The considered systemic factors, such as training deficiencies, inadequate procedures, faulty documentation, lack of currency, poor equipment design, poor supervision, failure on the part of the company to take action on previous violations and commercial pressures to take shortcuts. On the night of the accident, there was no evidence the approach briefing in keeping with standard operating procedures (SOPs) had been carried out aboard GF-072. Although the aircraft was established on the VOR-radial of 301 degrees at the FAF (final approach fix), other parameters were far from the required standard for a stabilised approach: the speed was 223 knots instead of 136 knots; the flap position was one instead of full, and the altitude was 1,662 ft rather than 1,500. Unless the speed was reduced, the captain could not have selected landing flaps to full. Excessive speed was one reason for not achieving the required stabilised approach configuration. Although the captain used speed brakes three times from 1922:49 to 1926:13, he could not achieve the required approach configuration before reaching the FAF. The aircraft’s speed of 223 knots at the FAF was 87 knots greater than the target speed. However, rather than initiating a missed approach, the captain decided to continue with the approach, during which the speed remained excessive. Investigators suggested the reason for the excessive speed could be the planning of the descent, or the omission of the descent clearance from the descent profile. The GF-072 simulation and flight tests showed that, based on the aircraft configuration, speed and altitude at the FAF, a successful landing could have been achieved, especially if the speed brakes had been
deployed continuously. To do so would have involved manoeuvring with a steep approach angle and rapid deceleration, however, and this would have unsettled the passengers. The captain did not stabilise the approach on the correct path at 500 ft in the required landing configuration, as required by company SOPs. When he apparently concluded that the landing could not be made, the captain elected to carry out a “Three Six Zero to the left”. This was non-standard procedure. Following the accident, Gulf Air issued a fleet instruction that: “Once an aircraft is established and descending on the final approach to the runway of intended landing, 360-degree turns and other manoeuvres for descent profile adjustments are not permitted.” The investigators concluded that the circumstances in the cockpit and the behaviour of the captain indicated that he probably experienced information overload. Departure from SOPs Even though GPWS voice warnings to “pull up” sounded every second from 1929:51, neither flight crew member responded according to SOPs. Instead, the captain concentrated on dealing with the flap over-speed which, at that stage, was not a critical emergency situation endangering the aircraft. The investigators said the accident could have been prevented if the pilot flying had adhered to SOPs. Departures from SOPs, particularly during the approach and final phases of flight, included: • During the descent and the first approach, the aircraft had significantly higher speeds than standard. • During the first approach, standard “approach configurations” were not achieved, and the approach was not stabilised on the correct approach path by 500ft. • When the captain perceived that he was “not going to make it” on the first approach, standard go-around and missed approach procedures were not initiated. Instead, the captain executed a 360-degree orbit close to the runway at low altitude with considerable variations in altitude, bank angle and ‘g’ force. • A rotation to 15 degrees pitch up was not carried out during the go around after the orbit. • Neither the captain nor the first officer responded to hard GPWS warnings. • In the approach and final phases of flight,there were several deviations of the aircraft from the standard flight parameters and profile.
“The investigators regarded two questions as critical: Why did the captain violate the SOPs, and why was there no challenge or comment from the first officer?” • During the approach and final phases of flight, the first officer did not call out or draw the captain’s attention to several deviations from the standard flight parameters and profile. Big questions The investigators regarded two questions as critical: Why did the captain violate the SOPs, and why was there no challenge or comment from the first officer? The captain’s sudden decision to execute an orbit was apparently aimed at avoiding the need for a standard missed approach procedure. A missed approach is a perfectly routine safety procedure, although in practice it is relatively rare. However, there could be reasons why a captain might be reluctant to carry out such a procedure. At the time of the accident, a go-around required the submission to the company of an air safety report describing the circumstances. Although Gulf Air said its policy was not to take action against pilots who had conducted missed approaches, the investigation found that some pilots at the time believed, rightly or wrongly, that company management would view such actions unfavourably. As a post-accident safety initiative, Gulf Air issued a fleet instruction, stating: “All pilots are further assured that no disciplinary action whatsoever will be taken against any crew that elects to carry out a go around for safety-related reasons, including inability, for whatever reason, to stabilise an approach by the applicable minimum height”. Another factor could be that captains might have feared losing the respect of relatively junior first officers if they executed missed approaches. Investigators said the CVR showed that the first officer performed his routine role of communicating with ATC, reading the checklist and carrying out the checks.
FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
31
COVER STORY
PHOTO: AAP
However, it also revealed that he played little effective part in flight deck management and decision making. He did not raise any issues with the captain or question his decisions, even though the captain performed non-standard procedures and manoeuvres. Crew resource management Evidence from the training records of the first officer indicated that he was seen as “shy” and “unassertive”, and that his operational performance overall was marginal. However, investigators also observed that at no stage during the approach did the captain consult the first officer on any operational decisions. The first officer was a valu-
able, but untapped, resource available to the captain, and crew resource management was virtually non-existent in the cockpit of GF-072. Although Gulf Air had been required by Sultanate of Oman regulations to provide a formal CRM training programme since June 1999, the original company CRM programme, established in 1992 and active until early 1997, appeared to have been discontinued with a change of management. The acting manager of human factors at the time of the accident said that his predecessor had resigned in frustration over his attempts to re-establish the program. Another factor contributing to the departure from SOPs could be that a company
Grim search Wreckage from Gulf Air Flight 072 is recovered from the Arabian Gulf.
32 FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
“...airlines with positive safety cultures, strongly motivated towards compliance with the regulations, are in the interests of the regulator.” might not emphasise strongly enough the importance of, the reasons for and the need to adhere to SOPs. And although Gulf Air had a flight data monitoring and analysis system in place, the system was not functioning satisfactorily at the time of the accident. Such systems can help identify the level of compliance with SOPs by detecting events including unstabilised approaches or times when an aircraft had exceeded specific pre-programmed parameters, such as airspeed, in a particular configuration. CFIT training CFIT (controlled flight into terrain) accidents account for the highest proportion of fatalities in commercial aviation. The CFIT training in the A320 fleet in Gulf Air was severely limited at the time of the accident. Airbus Industrie’s A320 normal course syllabus includes a GPWS pull-up demonstration. However, there was no similar syllabus for Gulf Air and no requirement to execute such a demonstration for its A320 fleet. Nor did Gulf Air’s A320 training program emphasise GPWS response training. The Airbus training program requires an instant, instinctive side stick response when a hard GPWS warning occurs. Organisational deficiencies: The investigators found that from 1998 to the time of the accident, the manager of flight safety was the only person in his department, and he did not report directly to the highest executive level within the company. They labelled this a serious organisational deficiency. They also noted that for many years Gulf Air had not participated in the regular sixmonthly meetings of the International Air Transport Association’s safety committee, at which the latest safety information is shared freely and confidentially between airlines, manufacturers and safety specialists. This had greatly restricted Gulf Air’s awareness of developments in areas such as accident investigation case studies, safety and risk management, training and safety information.
COVER STORY The regulator: The investigators also examined the relationship between Gulf Air and its regulator, the Sultanate of Oman’s Directorate General of Civil Aviation and Meteorology (DGCAM). A review of correspondence between DGCAM and Gulf Air revealed letters citing non-compliance with civil aviation regulations (CARs). In some areas, Gulf Air did not rectify problems identified by DGCAM. The company lacked several programs required by CARs. And it did not meet regulations in areas including crew resource management, quality management, safety awareness and other areas of crew training. An evaluation of Gulf Air carried out by the International Civil Aviation Organisation for DGCAM in October 1998 turned up evidence of delayed or non-compliance with regulatory requirements. The ICAO review concluded that, except for isolated incidents, most of the infractions could be traced to inadequate supervisory oversight within Gulf Air, rather than a deliberate disregard for the regulations. DGCAM was well aware of this situation, and had made many unsuccessful efforts to correct it, including imposing various sanc-
tions on the airline. Despite this, Gulf Air did not implement many changes sought by DGCAM. A review of relevant information and documentation covering the three years preceding the accident indicated that, despite intensive efforts, DGCAM could not get Gulf Air to comply with some critical regulatory requirements. The investigators said regulatory authorities and airlines had complementary roles in maintaining the safety of the aviation system. Strong and effective regulators are in the interests of airlines because they provide an independent means of quality control in airline operations. Conversely, airlines with positive safety cultures, strongly motivated towards compliance with the regulations, are in the interests of the regulator. At the time of the accident, this was not the case with the DGCAM and Gulf Air. The regulator needs to check that airline resources, structures and processes necessary to ensure regulatory compliance are adequate, the investigators said. It also needs the political support of the government to fulfill its safety role. This broader issue was the subject of a specific recommendation in the GF-072 investigation report.
Meanwhile, James Hogan, Gulf Air’s president and chief executive, says a lot has changed since the accident. The airline has enhanced regular fleet instructions and improved crew training, he says. The airline now electronically analyses flight data to ensure adherence to standard operating procedures, while all Gulf Air crew must be trained intensively in CRM, says Hogan, who took up his position at the helm of the airline after the accident. Gulf Air has incorporated into its flight crew training program modules driving home the risks posed by spatial disorientation, a problem also addressed in the first issue of the company’s upgraded safety magazine. And during the accident investigation, Gulf Air reviewed its A320 flight training program. This led to the reorganisation of the company’s operations division, a move Hogan says ensures a high level of pilot training. John Mulcair is a journalist based in Sydney. Rob Lee is an international aviation safety consultant and former director of the Australian Bureau of Air Safety Investigation. He was a consultant to the Kingdom of Bahrain Gulf Air Bahrain investigation team.
ite bs ls. we cia ur pe eo sS Se Xma
for
ers Gif t Vouch ble a also avail
“Bringing the Pilot Shop to the Pilot”
Australia’s leading mail order & web based supplier of pilots’ equipment & accessories. Authorised dealers for quality aviation products including:
Icom Radios
Garmin Portable GPS Systems
David Clark Headsets
Fastfind Personal Location Beacons
•Headsets & Accessories •Garmin Portable GPS Systems •Icom Radios •Ray-Ban Sunglasses •Leatherman Tools •Casio Watches •EPIRBs & PLBs •Aviation Books & Publications •Navigation Aids & Accessories •Aviation Videos & DVDs •Pilots Clothing •Windsocks
To see our entire range visit www.sk ysupply.com.au or call 1800 636 707 SS 3438 IFC
FLIGHT SAFETY AUSTRALIA NOVEMBER–DECEMBER 2003
33