Air Safety

As the aircraft was US registered, the incident was investigated by the US National Transportation Safety Board (N TSB). It determined that the probable cause of the accident was the captain’s preoccupation with the engine malfunction and his failure to properly monitor the flight instruments, which resulted in his losing control of the aircraft.

Contributing to the accident was the captain’s over reliance on the autopilot after the loss of thrust on the Number 4 engine.

Flight Progression

Flight 006 was flying in a polar jet stream and encountering light clear air turbulence (CAT). The captain had therefore turned the ‘fasten seat belt’ lights on.

Two automatic flight systems were engaged when the incident occurred: the performance management system (PMS) and the autopilot. When coupled, they provided pitch steering inputs into the aircraft’s pitch control system. The PMS provides thrust control above 2,500ft AGL to maintain pilot selected altitudes and speeds. When PMS is selected, the autothrottle servomotor is controlled by the PMS computer. Roll guidance to the autopilot was provided by the inertial navigation system (INS).

The Number 4 engine had a short history of earlier malfunctions, but despite rigorous testing nothing abnormal was found.

The PMS was set for the aircraft to cruise at 0. 85M (254KIAS), and when the CAT was encountered the airspeed began fluctuating between 0. 84M (251KIAS) and 0. 88M (264KIAS). Therefore the PMS began moving the throttles fore and aft to maintain the commanded cruise Mach number.

Engine pressure ratio (EPR) is the turbine discharge total pressure divided by total pressure at the compressor inlet: the higher the EPR, the greater the engine thrust output. Thus when the mach number increased to about 0. 88 the PMS retarded the throttles to about 1. 0EPR and the 747 decelerated. When 0. 84M was reached the PMS advanced the throttles and the EPR remained at 1. 05. The engine had ‘hung’ and the aircraft decelerated.

Number 4 Engine Relight Attempt

The flight engineer was directed by the captain to try to relight the engine at 41,000ft, notwithstanding the fact that the maximum relight altitude was 30,000 feet. The attempt was unsuccessful and the aircraft continued to decelerate. This time, however, the copilot had requested a clearancefrom ATC to descend to a lower altitude, but no emergency was declared. Despite at least six attempts by ATC to pass the clearance to descend, no acknowledgement was received from the aircraft.

As the aircraft decelerated past 240KIAS the captain turned the autopilot’s speed mode selector switch off, thus releasing it from the hold altitude command. This switched the autopilot to the pitch attitude hold mode whilst maintaining INS track in the autopilot roll mode without any pilot input. The captain then rotated the autopilot pitch control wheel ‘nose down’ in order to commence a descent to stop the aircraft losing further speed. When the pilot disengaged the autopilot, the aircraft immediately yawed and rolled right. The captain did not use any rudder to prevent the yaw.

Aircraft Non VMC

Almost immediately the 747 entered cloud and the captain stated that he was unable to recover to normal flight; he was uncertain of the roll attitude and was moving the control wheel to the left and to the right. However, as the aircraft accelerated during its departure from controlled flight the captain pulled the control column back and the airspeed decreased. He therefore lowered the nose and the Jumbo accelerated again. Finally, the captain and copilot between them managed to pull the control column back. The aircraft reduced speed and came out of the bottom of the cloud at 11,000ft.

Sequence of Departure From Controlled Flight

For about four minutes of flight the 7 4 7 carried out some most uncommercial manoeuvres and in the process lost around 30,000ft of altitude. The sequence of aerodynamic events was as follows:

• Time 1013:43. The roll angle of the aircraft was constant at about 3° left wing down, but the left wing down deflection of the control wheel increased as the IAS decreased. By now, the control wheel deflection had reached 22. 9°, the maximum available from the autopilot, so the aircraft started to roll slowly to the right.

• Time 1014:33. The airspeed had reduced to 225KIAS, and despite the 22° control wheel deflection the aircraft had rolled 23° right wing down and slightly (1. 8°) nose up. The airspeed slowly decreased as the aircraft continued rolling right at an increased rate. The rate of descent reached 1,200fVmin.

• Time 1014:50. The aeroplane had descended to 40,442ft at an airspeed of 221KIAS in a 64° right wing down and 4° pitch down attitude, and was now 60° off heading.

• Time between 1015:00 and 1015:23. During descent over the next 10,000ft, the aircraft’s nose down pitch angle increased to 60° and then reduced to 11°. The aircraft rolled over on its back and continued rolling right until it had reached 25° right wing low, essentially completing a 360° aileron roll.

• Time between 1015:23 and 1017:15. The aircraft descended from about 30,000ft to approximately 9,500ftwhen straight and level was recovered. The vertical acceleration forces reached 4. 8g at 30,552ft and 5. 1g at 19,083ft. Minimum speed reached was 54KIIAS-the aircraft has a usual stalling speed of 155KIAS. Control Regained At 9,500 Feet Most of the gyrations of the aircraft occurred in cloud and the pilot only completed effective recovery action when ‘visual’ beneath the cloud at 9,. 500 feet.

Miraculously, only two people on board were seriously injured, but the aircraft sustained major damage. The wings were bent two to three inches upwards at the tips. The left outboard aileron upper surface panel was broken. The trailing edge wedge was cracked in several places. The left and right undercarriage assemblies had separated from their attachment points. The auxiliary power unit (APU) too, had separated from its mounts and was resting on the two lower cone access doors.

Some 10-11 feet of the left tailplane, including the left elevator, had separated. The right tailplane lost 1. 75m (5ft) of the tip and a further section in the box beam area. No engines were damaged

Flight Crew Performance

The US NTSB report indicates that the copilot performed satisfactorily during the emergency and that the engineer performed in a correct and timely manner, except that during the attempted relight of Number 4 engine he did not close the bleed air valve switch before advancing the throttle.

However, the report is critical of the captain’s performance. He did not disengage the autopilot when the engine ‘hung’ as recommended in the training manual, and relied on the autopilot which he merely monitored. Had he disengaged the autopilot, he would havebeen required to perform the physical, more difficult, and more time and attention consuming tasks involved with flying the aeroplane manually. The captain then became preoccupied with the decreasing airspeed and his instrument scan broke down to the extent that he was not aware of the departure from normal flight. Finally, he became spatially disorientated and did not reorient himself until he cleared the cloud.

Although the board cited distraction and over reliance on the autopilot as causal factors, it also noted that the aircraft had been airborne for about 10 hours, had crossed several time zones and that the incident had occurred at about 0214 hours Taiwan local time, some 4 to 5 hours after the captain’s normal bed time. Thus his ability to obtain, assimilate, and analyse all the data presented to him could have been impaired by thee ffects of monotony, boredom and fatigue, even though he had had two hours sleep during the flight. However, the board did not accept these factors as facts and circumstances showed that he was alert to the situation as it developed. As an experienced 747 commander, it said, he should have known how loss of thrust of an outboard engine would affect controllability and that the autopilot control did not include rudder, so the net effect of the autopilot would be a left wing down roll which would induce sideslip, increase drag and decrease speed. In summary, the board concluded that the accident was not caused by the Number 4 engine malfunction. Indeed, an engine failure in the cruise for a 7 4 7 does not place the aircraft in immediate jeopardy, and even the airline’s operating manual does not classify such an event as an emergency. However, the malfunction did cause the captain to be distracted from his normal instrument scan, by his participation with the flight engineer in the evaluation of the engine malfunction and his desire to overcome the reduced airspeed. He did not deselect the autopilot as he was taught in training; the autopilot effectively masked the impending loss of control of the aircraft.

During his attempt to relight Number 4 engine the flight engineer did not close the air bleed valve switch before advancing the Number 4 throttle.

The automatic flight systems of the 747SP are such that the aircraft can be programmed for automated flight for the whole route. This reduces the physical workload for the pilot but increases the task of computer supervision, shifting total pilot workload from performing tasks to monitoring. The pilot is therefore only in the system monitoring mode and needs to get into the system controlling mode as soon as possible after a malfunction. There is ample evidence the board noted, from both research and accident statistics, that people are poor monitors.

Flight Crew Experience

Because the aircraft was US registered, the flight deck crew members had special purpose certificates issued by the FAA. The captain, first officer and flight engineer an had considerable flying experience on 747s.

The Captain, aged 55, had logged 15,494 hours, including 3,748 on 747s. He had passed his last two simulator proficiency checks in February and November 1984, and his last route check in April 1984. The First Officer, 53, had 7,734 hours of flying experience, of which 4,553 hours were on 747s. The Flight Engineer, 55, had flown 15,510 hours, with 4,363 hours on 747s.

There were two additional crew members (a captain and a flight engineer) but they were not on the flight deck at the time of the incident.