A passenger on board the aircraft, said that the measured length of the airstrip was 1,000 m, with trees on the left and open grasslands on the right. He said the airstrip surface was rough to the left of the centreline for almost the entire length. The airstrip also had a slight uphill gradient in the take-off direction.

The passenger stated that he had taken a short flight with the pilot earlier the same day. He said that during the accident take-off the aircraft was further to the left of the runway centreline than on the previous occasion, and it appeared to accelerate more slowly. At a height of about 100 to 120 ft above ground level, the stall warning horn sounded, and the aircraft began to descend towards the trees to the left of the strip. The left wing then clipped a tree, and the aircraft descended to impact the ground nose first.

The investigator was unable to obtain details of the accident from the pilot in command.

The pilot, who still had an area restriction annotated on his pilot's licence, planned to ferry the aircraft to Kalgoorlie where he was going to pick up another pilot for the remainder of the flight to Melbourne. Prior to departure from Jandakot he checked the fuel tanks were full. On taxy out, with the left tank selected, the pilot observed fuel venting from the right-wing tank. As a take-off clearance had already been given, he elected to continue with the flight.

Whilst cruising at 2,500 ft the engine lost power. The pilot attempted to correct the situation by changing magneto settings, but this failed to work. He then concentrated on completing a safe forced landing. As the aircraft touched down in a farm paddock two sheep ran towards the aircraft and collided with the landing gear. The pilot lost directional control, and the landing gear collapsed.

The aircraft operations manual indicates that the right-wing fuel tank must be selected for take-off and until the contents reduce to one quarter full. If this is not done it is possible for the excess fuel, which is normally returned to the right tank, to be vented overboard; thus, reducing the range. The pilot had the left tank selected for take-off and he observed fuel venting from the right wing. it is probable that the aircraft ran out of fuel because the pilot did not operate in accordance with the operations manual instructions. The pilot had limited experience on the aircraft type and was not aware of the fuel management requirements.

FACTUAL INFORMATION

History of the flight

The B747 had been cleared to depart from Sydney for Noumea via a Runway 16 Right Jet Three departure. This departure required a track of 155 degrees after take-off to an altitude of 800 ft, then a right turn to intercept the 163-degree radial from the Sydney very high frequency omni-directional radio range beacon (VOR). The B737 had been cleared to depart from Sydney for Brisbane via a Runway 16 Left West Maitland Seven standard instrument departure (SID). This departure required a track of 155 degrees after take-off to an altitude of 600 ft, then a left turn to intercept the 126-degree radial from the Sydney VOR.

After receiving the clearance from air traffic control, the B747 crew read it back correctly.

The B747 took off from runway 16R, followed by the B737 which took off from runway 16L. When the B747 reached 800 ft altitude, the pilot initiated a left turn, putting the aircraft in conflict with the B737. The Departures North controller asked the pilot of the B747 to confirm that he was turning right to intercept the 163-degree radial, and the pilot replied that he was turning left to intercept the 126-degree radial. The controller immediately instructed the B747 to turn right onto a heading of 180 degrees and passed traffic information on the B737 to the B747 crew. The B747 crew complied with the instruction immediately.

The pilot of the B737 reported that he had the B747 in sight and that both aircraft were turning away from each other. Recorded radar information indicated a minimum separation of 0.5 NM horizontally and 200 ft vertically. The procedural separation standard of 1,000 ft vertically and the radar separation standard of 3 NM horizontally had both been breached.

The aerodrome controllers observed both aircraft become airborne, but they became diverted by other tasks and did not see the B747 commence the left turn. As a result, visual observation was not maintained, and a breakdown of separation standards occurred.

Pilot's understanding of procedures

Two days before this occurrence, a B747 crew of the same operator, with the same pilot in command and carrying out the same departure, had made the same error. However, on that occasion, the Departures North controller instructed the crew to take up headings to achieve the required departure track. The crew was not advised of its error and no conflict with other traffic occurred.

After the second occurrence, the pilot reported that he had felt some confusion due to the naming of the departure procedures but had realised his mistake when he looked at the departure chart after being advised to alter heading to 180 degrees.

While admitting his mistake, the pilot added that there were contributing factors, namely:

1. the two departures were depicted diagrammatically on the same sheet of his documentation (Atlas chart number7);
2. he had not been aware of his error two days previously; and
3. the aircraft taking off from runway 16L was on a different tower frequency and he had no information on this traffic.

Documentation

The pilot was using the Atlas chart entitled "JET 3 DEPARTURE RWY 16L/R, 34L", dated 22 June 1995. The chart was bounded by a border 27.0 cm by 18.5 cm in width (approximately A4 size) and depicted three departures plus airspace and terrain information to 40 NM by distance measuring equipment (DME). The chart was designed to be read in the landscape orientation. Despite the size of the sheet, the departure details were depicted within a circular area 10.7 cm in diameter. The details of the procedures out to 10 DME were depicted within a circular area 7 cm in diameter. There were no warnings about parallel runway operations on runways 16L and 16R to alert crews not to turn toward the other runway centreline after take-off.

The Atlas documentation included another page headed "SIDs RWY 16R...SYDNEY (Austr)". The sheet was bounded by a border 18.3 cm by 13 cm (approximately A5 size) and was designed to be read in the portrait orientation. The page contained textual descriptions of seven SIDs from Sydney runway 16R. The description of the Jet-3 departure was headed "Sydney Jet 3" and was written in letters approximately 2 mm in height. The description was contained within an area 10 cm by 1cm and was worded:

"Sydney Jet 3, Climb on 155 (degrees), Initial turn at 800, RT to intercept R163 SY, 115.4. Expect radar vectors at or before BELLA (DME 15 SY)"

Nomenclature of procedures

There were three Sydney Jet Three departures differentiated only by the runway designator, runway 16R, 16L or 34L. All three procedures were totally different, despite the similarity of their names. The pilot in command reported that he found the similarity of procedure naming confusing and conducive to error.

The use of the one title, differentiated only by the runway designator, for a number of departures was common practice. There were five (Sydney) Radar Six, three Wollongong Three, three Shellys Three, two Richmond Five, two Mudgee Five, two Katoomba Five, three West Maitland Seven and three Williamtown Seven departures.

ANALYSIS

Crew awareness

Despite the crew's correct readback of the allocated departure to Air Traffic Control, the crew members were obviously unsure of the procedure. The fact that the same pilot in command had made the same error two days earlier and that the air traffic services staff did not highlight the error to the pilot, meant that an opportunity to clarify any uncertainty in the pilot's mind was lost.

Tower monitoring

The closest proximity between the two aircraft was 0.5 NM horizontally and 200 ft vertically, thus breaching the procedural and radar separation standards. This would not normally have been a problem as the tower controllers would be providing visual separation; but on this occasion, they were concentrating upon other traffic and did not observe the B747 turn left. As there were no radar or procedural standards in place, this action resulted in no separation being applied by Air Traffic Control. However, as the B737 pilot had the B747 in sight throughout the sequence of events, no serious collision risk was present. The rapid response of the Departures North controller, who detected the error and instructed the B747 crew to turn right onto 180 degrees immediately, prevented the situation deteriorating further.

The Manual of Air Traffic Services, Chapter 4, "Separation Standards", Section 1, Para 26 describes the obligation of a controller to issue a safety alert when he/she believes that an aircraft is in an unsafe proximity to terrain, obstructions, or other aircraft. The paragraph recognises that a controller cannot immediately see the development of every situation where a safety alert must be issued, but he/she must remain vigilant for such situations and issue a safety alert when the situation is recognised.

Documentation

The Atlas chart depicted the departure procedures in a small central section of the chart but also showed a considerable quantity of additional detail that was not part of the procedures. In contrast, the Airservices and Jeppesen charts showed only the details of the procedures. The Jeppesen and Airservices charts carried warnings about parallel runway operations, but the Atlas chart did not carry this warning. Although any useability comparisons must be subjective, as crews become accustomed to the charts that they use, the Atlas chart seemed to be less convenient for crew use than either the Airservices or Jeppesen charts.

The Airservices and Jeppesen charts depicted the procedures both diagrammatically and in text on the same sheet. The Atlas chart depicted the procedures only diagrammatically on the sheet, the description in text was on a separate sheet. Arguably, this would further reduce the Atlas chart's user convenience.

None of the Jet Three departure charts contained a hachured region between the 126 and 163-degree radials as a clear warning to crews that parallel runway operations take place and that aircraft must not turn toward the extended centreline of the other runway. However, other Airservices charts depicting the standard instrument departures for jet aircraft departing north, south and west contained hachured areas with warnings.

Nomenclature of procedures

In the case of parallel runways where the nomenclature of different procedures differs by only the words "left" or "right", there may be a greater likelihood of crew confusion and error than in the case of a single runway. In addition, if parallel runway operations are taking place, the consequences of an error could be more serious than in the case of a single runway.

SIGNIFICANT FACTORS

1. The Sydney Jet Three departure chart used by the B747 crew was relatively easy to misread.
2. The B747 pilot did not follow the departure procedure as cleared by Air Traffic Control.
3. The tower controllers did not adequately monitor the B747 from its take-off to the commencement of its turn, resulting in a breakdown of separation standards.

SAFETY ACTION

As a result of this investigation, on 21 September 1995, the Bureau issued Interim Recommendation IR950203 to Airservices Australia as follows:

'IR950203

"The Bureau of Air Safety Investigation recommends that Airservices Australia review the presentation of parallel runway operations on the Sydney Jet Departure charts, to improve the indication of the dangers of turning across the departure path of the other parallel runway.

"It is also recommended that the number and naming of departure procedures should be reviewed to ensure that any likelihood of flight crew selecting the wrong procedure is reduced."

An information copy of the interim recommendation was issued to the Civil Aviation Safety Authority.

On 13 November 1995, the Civil Aviation Safety Authority responded as follows:

"I refer to your interim recommendation IR950203 concerning an incident involving Boeing 747, F-GEXA and Boeing 737, VH-TJN near Sydney Airport on 25 August 1995.

"Summary

"A revised Sydney Jet Departure Procedure plate will become effective on 7 December 1995.

"A forthcoming "Terminal Area Workshop" sponsored by Airservices Australia and attended by industry representatives will discuss the adequacy of the current SID naming convention.

"Background to Response

"Following the earlier similar incident by this operator, the Sydney Jet Departure Procedure plate was amended by the hachuring of the area between the two runways and the re-organisation of the presentation of the departure wording in the text section of the procedure. The line weight of the Caution note was also increased to make it more prominent. The revised presentation reflects the intent of the BASI interim recommendation. The revised plate will become effective on 7 December 1995. Copy of the 07 DEC 95 plate is attached.

"With reference to the BASI comments on the naming of the multiple departure instructions, the naming format used conforms with the ICAO Annex 4 naming convention. This format was introduced with the revised SID presentation some years ago which allowed the number of SID procedures to be considerably reduced. For instance, the Sydney SIDs were reduced from 72 to the current 11 procedures and the DAP page numbers from 22 to 6. At that time, this initiative was applauded by both industry and ATC.

"The foregoing notwithstanding, the subject of SID naming will be discussed at the Terminal Area Workshop on

20 - 21 November 1995. This workshop is sponsored by Airservices Australia and will be attended by representatives of the industry. The discussion should give a good indication of the adequacy of the current SID naming convention."

On 22 January 1996, Airservices Australia responded as follows:

"I refer to your interim recommendation IR950203, concerning an incident near Sydney Airport on 25 August 1995.

"In response to your recommendations, I can advise that a revised Sydney Jet Departure Procedure plate became effective on 7 December 1995. The main amendments involved hachuring of the area between the two runways and the re-organisation of the presentation of the departure wording in the text section of the procedure. In addition, the line weight of the caution note has been increased to make it more prominent.

"In respect of your comments relating to the naming of the multiple departure instructions, I am advised that the naming format complies with the ICAO Annex 4 naming convention."

Classification of response: CLOSED - ACCEPTED

Circumstances

ANA914 was northbound on route B462 and cruising at FL310.  At 0130 UTC, the Brisbane Sector 10 controller coordinated with Moresby Control ANA914 estimating position APAVO at 0221, FL330 (the aircraft was at FL310 and maintained that level throughout the period of the occurrence). The Moresby controller read back FL330. The Sector 10 Flight Progress Strip for ANA914 showed the aircraft's level as FL310. There was no reference to FL330 on the strip. The Sector 10 controller did not include the words non-standard in the coordination, as was required by standard procedures, and the Moresby controller did not query this omission.  At approximately 0200, there was a shift change at the Sector 10 control position. There was nothing to indicate to the new controller that anything other than FL310 had been coordinated to Moresby Control.

At 0210, Moresby Control coordinated the departure of ANG3 Port Moresby for Brisbane on route B462 at 0208 on climb to FL370, estimating APAVO at 0252. The controller also said that she would provide separation with ANA914. She had planned to provide this separation by maintaining ANG3 at FL310, a level she understood to be below ANA914.  She would then climb ANG3 to the planned level, FL370, after the two aircraft had passed.

Information from the Department of Civil Aviation, Papua New Guinea, indicated that, at 0221, Moresby Control contacted ANA914 advising ANG3 as opposite direction traffic passing beneath ANA914 at an estimated time of 0226, and requesting ANA914 to report sighting and passing the traffic. ANA914 requested Moresby Control to repeat the information. This was done but the response from the ANA914 pilot indicated that the message had not been understood.  Two further attempts were made by the controller to convey the information but the response from ANA914 continued to indicate that the message had not been understood.

At 0224, the pilot of ANG3 asked ANA914 his level. ANA914 responded that he was at FL310, had traffic in sight, and was descending.  The crew of ANA914 was alerted to the confliction with ANG3 by the aircraft's TCAS equipment. ANG3 was not equipped with TCAS. However, the crew of ANG3 sighted ANA914 passing below their aircraft and estimated the vertical separation to have been less than 500 ft. The required vertical separation standard was 2000 ft.

The aircraft level (FL310) written on the flight progress strip for the aircraft had not been ticked (in accordance with normal procedures) by the Sector 10 controller to indicate that coordination had been completed. No other aircraft in the Sector 10 control area was at FL330. The basis for the controller coordinating FL330 was, therefore, not established. However, he did report what, in his view, were minor but irritating difficulties in operating the telephone link to Port Moresby to initiate the coordination. There was also some evidence of the controller suffering fatigue.

Discussion

The agreement between the air traffic control agencies of Australia and Papua New Guinea allows for non-standard levels to be used only when prior co-ordination between controllers has been carried out.  Therefore, the use of FL330 for ANA914 was contrary to this agreement as no prior co-ordination had been initiated.  The fact that the phrase non-standard was not used in the co-ordination should have provided an additional cue that an error had been made. Beyond this point, however, in the procedural environment in which the aircraft were operating, the sole remaining element in the safety net was TCAS on ANA914.

The difficulties experienced by the Port Moresby controller in communicating the opposite direction traffic to ANA914 highlight the language and comprehension problems which can arise in the control of international air traffic.  TCAS takes on added importance in such situations.

Significant Factors

1. The Sector 10 controller may have experienced some frustration operating the telephone link to Port Moresby.
2. The Sector 10 controller may have been fatigued.
3. The Sector 10 controller coordinated an incorrect level and did not use the prefix non-standard to describe that level.
4. The Port Moresby controller did not query the omission of the prefix non-standard when reading back the coordination.

Safety Action

As a result of the TCAS implications of this occurrence and as similar implications were found to be present in occurrence B9501346, the Bureau considers it is of significant safety value to reproduce interim recommendation IR950117 in this report. IR 950117 was addressed to the Civil Aviation Safety Authority on 6 June 1995.

IR 950117

The Bureau of Air Safety Investigation recommends that the Civil Aviation Authority:

1. mandate the fitment and use of an Airborne Collision Avoidance System (ACAS) in all aircraft engaged in Regular Public Transport (RPT) operations;
2. consider the requirement for the fitment and use of a suitable ACAS in other aircraft engaged in the carriage of passengers for hire or reward;
3. review the requirements for the carriage and activation of transponders with the objective of maximising the effectiveness of ACAS;
4. mandate the standard of ACAS equipment to be carried in each aircraft classification;
5. set a timetable for the introduction of ACAS equipment; and
6. ensure that air traffic services officers are given adequate and timely education and continuation training in the capabilities and operational impact of ACAS equipment.

VH-JIJ was being flown by a student pilot on her first solo navigation exercise. Approaching Moorabbin from the east on the last leg of the exercise, the aircraft entered controlled airspace at 4800 feet (lower level 4000 feet) without an airways clearance. The radar advisory service (RAS) controller had observed VH-JIJ on an easterly heading for some time previously at 4000 feet before climbing.

When VH-JIJ entered controlled airspace at 4800 feet, VH-NCE was proceeding in the opposite direction on climb to 7000 feet and approaching the same altitude as VH-JIJ. The RAS controller gave the pilot of VH-NCE a suggested heading to avoid VH-JIJ, which at that time was unidentified traffic. The pilot of VH-NCE advised visual contact with the other aircraft as the two aircraft passed. Separation was estimated to be one to two miles laterally when both aircraft were at the same level.

The pilot of VH-NCE advised that cloud in the area was scattered/broken with a base of 4000 feet. VH-JIJ was identified by transponder code as it got closer to Moorabbin.

Factors

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

• VH-JIJ entered controlled airspace without a clearance.
• The pilot of VH-JIJ was an inexperienced student pilot.

The student pilot was operating the last sector of a solo navigation exercise, tracking from Murray Bridge to Adelaide where a touch-and-go landing was made, before returning to Parafield at 1,500ft.

Parafield Tower cleared the aircraft for a straight-in approach to runway 03 L, with a clearance to land. The pilot reported on final at 1,500ft, but the aircraft was not initially sighted by the controller. It was then observed north west of the field, inside the RAAF Edinburgh Control Zone.

An Orion aircraft, conducting right hand circuits for runway 36 at Edinburgh, was on right base heading west when the other aircraft entered the control zone.

Radar plots revealed that VH-YHH passed behind the Orion at about the same level, with a horizontal separation of about 1000 metres.

The student pilot then made a right turn to enter left downwind for runway 03 L, Parafield, and continued for a landing.

The pilot stated that on Monday 21 August 1995, he was the captain of VH-JOC rostered to fly from Cairns to Weam via Daru. The aircraft departed Daru for Weam as per flight plan and arrived overhead Weam at 1215 pm local time. The pilot said he observed a Piper PA-31 Navajo, and a BN-2 Islander aircraft parked in the parking area. The pilot of the Islander had previously provided the pilot of VH-JOC with a strip report and had advised that the strip was serviceable.

The pilot stated that a normal circuit entry, approach and landing were conducted. The aircraft touched down normally and at approximately 60 kts braking was applied. The right main gear then appeared to lock up or strike something. Braking was released but the aircraft yawed to the right. The pilot said he tried to correct the yaw with rudder, but it became worse. The nosewheel appeared to dig into the runway surface and he was unable to hold the weight off the nosewheel due to the yawing effect. The gear then collapsed, and the aircraft slewed further right before coming to rest.

Subsequent examination of the right main gear by an engineer revealed that the pivot bolt had pulled through the torque links after the retaining washer had been sheared. This failure then allowed the wheel to castor outwards and resulted in loss of directional control and collapse of the left main and nose gear.

Two pilots who witnessed the accident said that there was a strong crosswind from the left, backing to a tailwind component on the landing. They said the aircraft was going sideways and looked as if it was going to ground loop prior to the failure of the torque link.

The Boeing 767 was being processed by the Approach North controller (APPN) for a visual approach to runway 16R. A clearance was issued to descend to 2500 ft, together with radar vectoring to position the aircraft for left base. At the same time a helicopter was operating in the Middle Harbour area at altitudes between 1000 ft and 1500 ft.

The Boeing was equipped with a Traffic Collision Avoidance System, version two (TCAS II). TCAS II contributes to collision avoidance protection by issuing traffic alerting and conflict resolution advice to the flight crew.

As the Boeing descended on an extended base leg the crew informed APPN that they were responding to a TCAS Resolution Advisory (RA) and had initiated a climb. Although the controller reported that the Boeing was 1000 ft above the helicopter, the crew responded with, "Roger, have 500 ft indicated". Shortly after, the pilot of the helicopter reported sighting the Boeing and indicated there was no confliction. The Boeing was instructed to descend to 1500 ft and subsequently carried out an approach and landing without further incident.

Later analysis of radar data indicated that the Boeing had been descending at approximately 1000 fpm but discontinued its descent at 2800 ft pressure altitude and commenced to climb, consistent with receiving the RA. The helicopter had been maintaining level flight at approximately 1000-1200 ft pressure altitude. At the time the crew of the Boeing initiated the climb the vertical separation with the helicopter was 1700 ft, with a horizontal separation of 1.27 NM. The minimum required separation standard was 1000 ft vertical separation or 3 NM horizontal.

Because TCAS II software cannot anticipate that an aircraft may level off, the predicted flight paths of the Boeing and the helicopter, as determined by the TCAS II software, was most likely sufficient to activate the RA. It could not be determined why the crew of the Boeing reported the helicopter being 500 ft below their aircraft, as indicated on their TCAS II cockpit display.

The aircraft was cruising at 2,000 ft above ground level when the pilot heard an unusual noise coming from the engine. He immediately placed the aircraft into an autorotation descent. The chip detector light illuminated, the airframe began to vibrate, and the engine-out warning sounded prior to landing. The pilot was able to complete a safe but heavy emergency landing in a heavily timbered area.

Inspection indicates the engine failure was caused by the decoupling of the power module. Once separated, the fourth stage turbine wheel breached the turbine casing.

The turbine assembly and the fourth stage wheel had completed 1612 hours time in service since they were last overhauled. The turbine had 1887 and the wheel had 2032 hours to run to their next overhaul.

Examination of the power turbine components indicates that the power turbine coupling fractured in the reduced section of the coupling, aft of the region that forms the number six bearing inner race. The number six bearing components were also extensively damaged. Most of the damage was consistent with eccentric rotation of the power turbine assembly.

A blade had been lost from the fourth stage turbine wheel. The shroud ring and blades are an integral part of the wheel. Blade separation requires fractures in both the shroud ring and the blade aerofoil section. The shroud had fractures on both sides of the blade. The blade had fractured near its base. Fatigue cracking was associated with all three fractures. The failures in the shroud ring were consistent with slow fatigue crack growth under a uniform loading environment over many engine cycles (low cycle fatigue). The failure in the blade was consistent with crack growth in a variable amplitude loading environment over a lesser number of engine cycles (high cycle fatigue). No abnormalities were found which may have contributed to fatigue crack initiation.

The different nature of the fatigue cracks in the shroud and blade indicate that the shroud fractures preceded the crack in the blade. The fracture of an integrally cast shroud ring on both sides of a turbine blade would result in a changed loading condition that could lead to premature blade failure. The loss of a blade from the fourth stage wheel would create an out of balance condition. This probably led to failure of the power turbine rotor assembly.

The circumstances which caused fatigue cracking and final failure in the shroud could not be established.

As part of an endorsement training exercise the captain operated the emergency electrical Bus to simulate an electrical fire. When the Bus was selected all electrical power was removed from the aircraft. Later examination found a broken wire in a relay box. It was found that the fault was not detected during previous maintenance test/checks because the test circuit goes to the earth side of the circuit, and the breakage occurred on the power side.

The operator is currently corresponding with the aircraft manufacturer to recommended a change to the maintenance manual procedural checks that will identify any similar faults during ground tests.