FACTUAL DATA

1. The flight

Passing through 15,000 ft. during climb out of Canberra the left engine N1 and N2 rpm indicators began to fluctuate to 10% of engine rpm. This was accompanied by fluctuating ITT and fuel flow. At this time the engine was being controlled by the onboard computer in the Auto mode, therefore, in accordance with flight manual procedures, the pilot selected Manual mode. The engine parameters then steadied, apart from N2 which continued to fluctuate by 10% rpm. It was then noted that the oil temperature had increased to the top of the green arc on the gauge. A reduction in power had little or no effect on the temperature, however the oil pressure remained normal.

Approximately five minutes later the engine flamed out with N1 observed to be windmilling and N2 stationary. The engine was secured and the aircraft continued for an asymmetric landing at Essendon where it is based.

2. The investigation

The starter motor was removed and the engine hand turned to check for freedom of rotation. Although a slight rumble could be heard from the vicinity of the transfer gearbox the N1 appeared to turn freely. However, on removal of the cover plate from the transfer gearbox it became apparent that the horizontal bevel gear was damaged. The freedom of rotation of the N2 assembly was then checked with no apparent problems noted.

The engine was removed from the airframe and, when re-fitting the cover plate to the transfer gearbox in preparation for dispatch to the overhaul facility, a screw and washer were found loose in the bottom of the transfer gearbox. A borescope inspection of the gearbox found that the top screw, P/No. NASllOIE3-8, was missing from the cylindrical roller bearing housing in the transfer gearbox assembly p/n 3070093-4.

The missing screw was one of three that holds the cylindrical roller bearing in place. During assembly these should be torqued to 40 in lbs. A torque check prior to disassembly of the other two screws showed that they were below that torque and both needed a further quarter turn before attaining the required 40 in lbs torque.

The horizontal bevel gear P/No. 3070212-3 had impact damage on the back of the gear and also damage to the gear teeth. The vertical bevel gear P/No. 3070219-2 had sustained severe damage to the gear teeth. The transfer gearbox housing assembly P/No. 3070429-6 was foreign object damaged (FOD) over its internal surfaces, and the transducer P/No. 3070722-3 was severely damaged. Metal contamination was found throughout the gearbox components including the bearings. The engine oil pump was found FOD contaminated and completely seized.

The gearbox had operated 4,253 hours and had not been disassembled since new.

ANALYSIS

The top cylindrical roller bearing screw P/No. NASllOlE3-8 became dislodged in service resulting in substantial secondary damage to the transfer gearbox components.  Considerable metal contamination of both the transfer and accessory gearboxes occurred. The failure of the transfer gearbox resulted in the break-down of the drive to the accessory gearbox which in turn resulted in the fuel pump failing to deliver adequate fuel to the engine resulting in flame out. The considerable metal contamination as a result of the failure also resulted in the oil pump seizing.

The fluctuations in N1 and N2 rpm coupled with the fluctuating ITT and fuel flow as reported by the pilot were genuine fluctuations as a result of the transfer gearbox failing, with consequent intermittent drive to the accessory gearbox and fuel pump. The fact that the engine settled down again when manual mode was selected is considered to be a coincidence. The continuing apparent fluctuations in N2 rpm would have been due to the transducer beginning to sustain the damage which eventually rendered it unserviceable. The reason for the N2 gauge failing to read for the remainder of the flight after engine flameout, although the LP shaft was windmilling, is because of the damage to the transducer rendering it incapable of sending a signal to the indicator.

It is considered that during initial installation the bolts may have been oil lubricated. This procedure is known to result in a lower torque than that achieved in a dry installation.

The gear box installation on all jet engines is subject to heat and vibration. If the securing screws were not properly installed, and were not properly tightened to the designated torque, the operating environment would be conducive to those screws loosening and becoming displaced.

Significant factors

The following factors was considered relevant to the development of the incident.

1. During initial assembly of the gearbox the three cylindrical bearing retaining screws were not tightened to the specified torque.
2. One cylindrical bearing retaining screw loosened and became displaced.
3. The displaced screw caused distress to the gearbox components sufficient to fail the gearbox drive to the engine fuel pump.
4. The engine flamed out due lack of fuel.

GLOSSARY OF TERMS

rpm     engine speed in revolutions per minute

ITT      engine operating temperature (Interstage Turbine Temperature)

N1       engine HP core speed

N2       engine LP fan speed

FOD    foreign object damage

in lbs   inch pounds - the measure of torque applied to a screw

LP       engine low pressure (N2) stage

The pilot of VH-OGN advised that whilst on descent approaching their cleared altitude of 5000 feet via a Hopla 1 STAR for runway 16, they received a TCAS proximity traffic warning with no altitude readout. The intruding traffic was sighted and was reported to be similar to either a Cessna 182/206, tracking from left to right and about one and one half nautical miles to their left. VH-OGN was levelled at 5400 feet and banked to the right.

The other aircraft maintained its track and passed approximately half a nautical mile behind VH-OGN. The captain of VH-OGN said that while this was occurring, the TCAS changed to a "yellow traffic with aural warning". He estimated that the Cessna was between 4800 and 5000 feet.

Although the Cessna was squawking code 2000, there was no altitude readout on air traffic control screens. Attempts to contact and identify the Cessna were not successful.

The student pilot was engaged in a solo cross training flight at the time of the accident. As he was approaching the aerodrome he reported that he checked the windsock and noticed that the wind appeared to be from the north-east at approximately 10 knots. The pilot entered the circuit area and completed an approach and landing on runway 07. The aircraft touched down, on the gravel centre section, approximately 300m in from the threshold of the 1221m long strip.

The pilot observed that the aircraft was not slowing as fast as expected and he applied full braking. When it became apparent that the aircraft would not stop by the boundary fence the pilot shut the engine down and attempted to steer the aircraft to the right, on to the grass and away from the fence.

The left wing collided with a fence post, swinging the aircraft through 140 degrees and the aircraft stopped on the gravel verge of a two-lane road with the right wing blocking one lane.

As the pilot exited the aircraft, he observed that the windsock indicated a tailwind of approximately 20 knots on runway 07.

The pilot advised that during the landing roll on bitumen runway 36, the aircraft swung sharply to the right. He applied left brake to correct the swing. The aircraft began to straighten but then it veered sharply to the right again like a ground loop. The left landing gear leg collapsed, and the left wing struck and propeller struck the ground. At the time, the wind was gusting 10 to 15 knots from 010 degrees.

After the accident, the pilot thought that the left brake might have failed causing him to lose directional control. An engineer inspected the brakes and advised that there was no evidence of a brake failure. However, the engineer rebuilding the aircraft discovered a pre-existing fatigue crack through about half the left landing gear leg. It is probable the fatigued landing gear leg failed during the landing roll.

According to the engineer, the landing gear legs in VH-ICA were of the thicker metal version. Therefore, they were not subject to mandatory magnetic particle method crack testing as were the earlier version of spring legs in accordance with Airworthiness Directive AD/Cessna 188/3.

Significant Factors

This accident was not the subject of an on-scene investigation and there was insufficient evidence available from other sources to determine the factors that led to the accident.

Safety Action

The maintenance organisation rebuilding VH-ICA advised that a defect report would be submitted to the Civil Aviation Authority concerning the pre-existing crack found in the landing gear leg.

The Bureau of Air Safety Investigation advised the Civil Aviation Authority of the fatigue crack by way of the Bureau's weekly summary of occurrences.

Whilst in the circuit, the pilot noticed that the landing gear down light was not illuminated. He checked the visual drive indicator in the cabin floor and the external cowl mirrors which indicated to him that the gear was down. During the initial landing roll, the gear collapsed, and the aircraft came to a halt on the runway.

Engineering advice indicated that the gear was not fully down when the aircraft landed. This had been caused by the seizure of the gear motor. It was possible for the pilot to have believed the gear was down from the view he had in the mirrors.

Prior to this occurrence, the pilot had experienced problems with the dimming function of the landing gear indicating light. He had previously landed during the day with the gear light dimmed and thought that this was again a light dimming problem.

The instructor and student were practising engine out landings, to touchdown, on a grassed area of the airport. On the accident approach, ground speed had been reduced to below five knots before the aircraft touched down. As the aircraft slid along on its skids they dug into the soft ground and the helicopter pitched forward until the main rotor blades struck the ground. The helicopter then rolled onto its right side.

The instructor was aware of the soft ground and was monitoring the approach carefully. Because the ground speed was low, the instructor was not expecting any problems with the landing and, as a result, he had insufficient time to take corrective action before the main rotor blades struck the ground.

The aircraft stalled and touched down heavily following an approach for a landing.

The student pilot was practicing a glide approach under the supervision of the instructor. The student misjudged the flare for landing and the aircraft ballooned to a height of about 25ft above the runway. The instructor applied some power as he told the student to lower the nose. The student was slow to react, and the airspeed reduced rapidly. The instructor took the controls, lowered the nose and applied full power but he was unable to prevent the aircraft from stalling and landing heavily.

Crew

In later discussion, the pilot in command stated that he was attempting to comply with the request from Canberra Approach to maintain a maximum speed descent. He believed that the aircraft would pass the 20 NM range circle indicated by the aircrafts Distance Measuring Equipment (DME), above 6,000 ft which was the lower limit of controlled airspace at that distance. He thought that it had done so as the aircrafts DME indicated 17 NM from Canberra when GPWS warning was received. The first officer, who was the handling pilot during the incident, stated that he checked the DME reading when the warning was received. At 17 DME, the lower limit of controlled airspace was 5,000 ft.

Flight data recorder information

The flight data recorder (FDR) data confirmed that the aircraft was descending through 6,000 ft when the GPWS activated. FDR data also confirmed the GPWS warnings and the 300 ft climb.

Recorded values of latitude and longitude revealed that the GPWS warnings were activated as the aircraft overflew Bulls Head, spot height 4,482 ft, approximately 20 NM on the 240-degree radial from Canberra VOR. The data confirmed that a valid Mode 2A Excessive Terrain Closure GPWS warning had occurred, and the crew had taken the appropriate actions to remove the aircraft from the GPWS activation envelope.

The FDR data showed the aircraft on descent at an airspeed of approximately 315 Kt on a heading of 078 degrees magnetic when the GPWS warning occurred. Recorded radar altimeter data indicated showed rates of closure with terrain in excess of 9,000 feet per minute (fpm). The radar altitude reduced from 2,658 to 2,044 ft in a three-second period.

ANALYSIS

GPWS parameters

The GPWS will give a Mode 2A warning if the aircraft is flying with less than 25 degrees of flap and there is an excessive rate of closure with rising terrain. If the airspeed exceeds 250 kts, the upper boundary of the radio altitude envelope is 2,450 ft with a rate of closure of 5,105 fpm and the lower boundary is 50 ft with a rate of closure of 2,063 fpm for a "TERRAIN, TERRAIN" warning. Within this envelope is the envelope triggering the "WHOOP WHOOP PULL UP..." warning.

The combination of the aircrafts rate of descent and the rising terrain on the western side of the Bulls Head resulted in a rate of closure in excess of 9,000 fpm and a radar altitude of 2,044 ft which was within the GPWS inner operating envelope.

Pilot’s intentions and results

The pilot’s intentions were to maintain the maximum speed during descent and to pass the 20 DME range circle just above 6,000 ft, continuing descent to 5,000 ft when inside the 20 DME circle. Both radar and FDR data revealed that the aircraft descended below 6,000 ft beyond the 20 DME circle and thus operated outside controlled airspace (OCTA) for a short time.

An aircraft operating at an altitude which is the published lower limit of controlled airspace is operating OCTA. Another aircraft, about which the air traffic services held no information, could have been operating OCTA at the altitude reached by the B737.

As both pilots reported that the aircrafts DME indicated 17 NM from Canberra when the GPWS warning was received, the DME equipment may have been faulty.

SIGNIFICANT FACTORS

1. While the pilot was attempting to maintain a maximum speed descent for air traffic control sequencing purposes, the aircraft descended below the lower limit of controlled airspace.
2. The combination of the aircrafts rate of descent and rising terrain placed the aircraft within the GPWS warning envelope.

Boeing B737-476, VH-TJK, was being radar vectored onto the Sydney 16R localiser from the right, to follow a B747-438 already established on the localiser. The pilot in command was flying the aircraft with autopilot "A" engaged, heading mode selected, and VOR/localiser capture mode armed. The aircraft turned onto the localiser, descending through an altitude of 2,500 ft, some 2.3 dots below the glideslope, about 10 NM from the landing threshold. As the B737 began to intercept the localiser track, maintaining an airspeed of about 218 kts, the bank angle was progressively increased to 29.5 degrees before the aircraft abruptly rolled further to the right. This uncommanded event coincided with initiation of trailing edge flap and leading edge slat extension from the stowed position. The aircraft reached a maximum bank angle of 62.9 degrees before the roll was stopped, and the wings levelled by the handling pilot. A missed approach was carried out, after which the aircraft was landed normally without further incident. At the time of the occurrence the surface wind was 210 degrees at 10-15 kts, with scattered cloud at 1,800 ft and light rain in the area.

VH-TJK immediately underwent an extensive ground inspection program, followed by a flight test, in accordance with recommendations from the manufacturer. No defects were found which could have contributed to the occurrence.

An examination was carried out of recorded radar data, and of information derived from the flight data recorders of

VH-TJK and the preceding B747. It was found that the B747 had intercepted the 16R localiser at an altitude of 3,000 ft, approximately 11 NM from the landing threshold. The aircraft maintained 3,000 ft until intercepting the glideslope at about 9.5 NM and carried out what appeared to be a normal ILS approach to runway 16R. There was no reported turbulence.

VH-TJK intercepted the localiser at about 10 NM from the landing threshold, some 500 ft lower and about 115 seconds later than the B747 had been at the same lateral position. Recorded wind data from the inertial reference systems of both aircraft indicated the wind direction varied between 155 and 165 degrees, at a velocity of 25-35 kts. As the localiser track is 155 degrees, this meant there would have been little or no lateral displacement of any wake vortices produced by the B747.

Research has shown that for a typical jet transport aircraft, the wake descends behind the generating aircraft at approximately 300-500 ft/min for about 30 seconds. The descent rate decreases and eventually approaches zero at

500-900 ft below the flight path. The decay process of the wake is complex and is strongly influenced by atmospheric conditions (Boeing Airliner/Jan.-Mar. 1995). UK CAA wake turbulence studies have also shown that B747 and B757 aircraft produce higher Category A incident rates than other aircraft, where Category A incidents correspond to the development of uncommanded roll angles 30 degrees or greater to following affected aircraft. The same studies showed that aircraft with the highest incident rates of encountering wake turbulence behind B747 and B757 aircraft were BAC-111, B737 and DC9 types.

CAA wake turbulence radar separation standards are described in the Australian Manual of Air Traffic Services and are based on three categories determined from the maximum certified take-off mass of the aircraft. B747 aircraft are categorised as heavy, while B737 aircraft fall into the medium category. The standard, which is based on distance, shall be applied when an aircraft is crossing behind, or operating within 0.5 NM laterally of another aircraft's flight path at the same level or less than 1,000 ft below. In this case, a medium behind a heavy required a minimum separation of 5 NM. Examination of the recorded radar data showed that the separation between VH-TJK and VH-OJC was approximately 5 NM at the time of the occurrence.

The recorded trailing edge flap and leading edge slat data for VH-TJK indicated that the uncommanded roll occurred whilst the devices were extending from their fully stowed positions. It was not possible to determine from the recorded data whether this was responsible for the uncommanded roll or contributed to its severity. However, ground inspections and subsequent flight testing did not reveal any defect in the operation of those systems.

It is therefore considered that the onset of the uncommanded roll resulted from an encounter with wake turbulence generated by the preceding B747.

Significant Factors

1. Atmospheric conditions were conducive to the slow decay of wingtip vortices generated by the preceding B747.
2. VH-TJK intercepted the localiser some 500 ft below the altitude of the preceding B747.
3. The maximum longitudinal spacing between VH-TJK and the precedingB747 was 5 NM.

At 0349.34, the pilot of VH-EXC reported to Flight Service (FS) that he was over St Helens at 7000 feet, commencing descent and advised he would report again by 0410 or on missed approach.  His intention was to make a practice NDB approach.  In response FS advised him that there was no IFR traffic.

At 0357.00, the pilot of VH-HFD reported that he had departed St Helens at 57, was tracking 243 degrees (from the NDB) and was on climb to 8000 feet. (VH-HFD was running 20 minutes early on flight planned estimate for departure St Helens).   FS asked him if he had copied VH-EXC (as traffic) to which he replied negative.  FS then advised VH-HFD that VH-EXC was in the St Helens area for an NDB approach.  The two pilots then made contact with each other. By this time VH-HFD had left 5000 feet on climb and VH-EXC was at 2000 feet, tracking outbound in the NDB approach. Conditions were IMC and it was apparent that the two aircraft had gone through each other's levels, possibly in close proximity.

From timings on the transcript and information provided by the pilots, it was calculated that VH-HFD would have commenced its take-off roll at about 0352. Engines would have been started two to three minutes prior to then at about 0349.  The pilot of VH-HFD said that he made taxy broadcasts on both the FS and Common Traffic Advisory Frequency (CTAF).  Both aircraft were dual VHF equipped, and both pilots said they were monitoring both CTAF and FS frequencies.  The pilot of VH-EXC did not hear either of the taxy calls reportedly made by the pilot of VH-HFD.  The pilot of VH-HFD did not hear the call made by the pilot of VH-EXC on arrival over St Helens at 7000 feet.  On timings, it is possible that that broadcast was made prior to the pilot of VH-HFD commencing a listening watch on the two frequencies.

On arrival over St Helens at 7000 feet, VH-EXC entered the holding pattern to lose altitude.  The outbound leg of the NDB approach was commenced at 4500 feet probably at about 0355.  The outbound track is 107 degrees.  The pilot of VH-HFD departed from runway 08 at about 0352 and maintained runway heading until reaching 2000 feet where he made a right turn to track back to the NDB while continuing to climb.  It is obvious from these timings that the two aircraft passed in close proximity both vertically and laterally while in cloud and on close to reciprocal tracks.

Significant Factors

The following factors were considered relevant to the development of the incident:

Neither pilot heard advisory broadcasts made by the other pilot.  It is possible that the pilot of VH-HFD may not have commenced a listening watch at the time the pilot of VH-EXC made his call on arrival over St Helens but the pilot of VH-EXC should have heard the calls from VH-HFD.

Melbourne FS cannot receive calls made on the ground at St Helens so were not aware that VH-HFD was taxying and were therefore unable to pass traffic information to VH-EXC.  In addition, VH-HFD was running 20 minutes early on flight plan so FS could not pass traffic information in anticipation of a planned departure time.

The pilot of VH-HFD did not establish contact with Melbourne FS as soon as possible after becoming airborne. Instead, he waited until he had climbed to 5000 feet and established the aircraft on the departure track.  This denied both FS and the pilot of VH-EXC the opportunity to become aware of the whereabouts of VH-HFD.