The Boeing B737-800 (B737) was cleared to Melbourne via the Sydney RWY 34R MARUB THREE standard instrument departure (SID) to 5,000 ft. The Boeing B767-338ER (B767) was inbound to Sydney from Auckland, NZ, and had been cleared to descend to 6,000 ft with a vector to a right downwind leg for RWY 34R. As the aircraft approached each other 12 NM east of Sydney, an infringement of the radar separation standard occurred.

The pilot in command of the B737 was the handling pilot for the sector and was manually flying the aircraft while tracking via the SID. He had recently completed retraining on the aircraft after having not flown the type for 10 years.

After take-off, the B737 entered cloud and encountered turbulence as it climbed through 3,500 ft. The pilot in command was monitoring the aircraft's weather radar and stated that he became distracted while assessing the meteorological conditions. Although the co-pilot gave the 1,000 ft to assigned altitude call at 4,000 ft, he was also observing the weather situation and did not monitor the flight instruments as the aircraft approached the assigned altitude. The B737 continued to climb above 5,000 ft and reached 5,700 ft before the pilot in command descended the aircraft back to the assigned level. During the descent the aircraft's traffic alert and collision avoidance system issued a Traffic Alert.

The departure controller issued a turn instruction to the crew of the B737 for avoidance action and an evasive turn instruction to the crew of the B767, in addition to providing traffic information on the B737. Recorded radar data indicated that lateral separation between the aircraft reduced to 2.8 NM with a vertical separation of 900 ft. The required radar separation standard was 3 NM laterally or 1,000 ft vertically.

Synopsis

A Cessna 172P (C172) aircraft, VH-KTV and a foreign registered TL Ultralight Sting aircraft, OK-GUU39, converged and collided at low altitude in the vicinity of the threshold of runway 24 right (24R) at Jandakot, WA. The occupants of both aircraft were uninjured. The TL Ultralight Sting (GUU39) was substantially damaged and the C172 sustained only minor damage.

The owner-pilot of GUU39 had imported the aircraft from Czechoslovakia and it was one of two aircraft (GUU38 and GUU39) that had recently been assembled at Jandakot. Both aircraft were being operated in accordance with special flight authorisations issued by the Australian Civil Aviation Safety Authority (CASA). The two aircraft had completed several flights in company during the days prior to the accident and were returning from Bunbury at the time of the collision. The owner-pilot was the sole occupant of GUU39.

The pilot of the C172 had hired the aircraft from a training organisation for the purpose of conducting a local flight to the training area with one passenger. The pilot was seated in the control seat on the left side of the cockpit. The C172 had joined the circuit from the training area at the time of the collision.

Jandakot tower was active until a short time before the collision and both aircraft conducted their arrival to the airport under General Aviation Aerodrome Procedures (GAAP). The GAAP control zone (CTR) was deactivated at the scheduled time of 1800 Western Standard Time (WST) and the aircraft were being operated under Mandatory Broadcast Zone (MBZ) procedures for the final stages of their flights. At the time of the collision a certified air-ground radio operator (CAGRO) was providing operational information to pilots. Although the CAGRO used the facilities of the control tower to provide this service, he was not providing an air traffic control service.

Sequence of events

Recorded information from primary radar returns, voice transmissions and information obtained during investigation interviews were used to compile the following sequence of events. Neither aircraft's transponder was transmitting encoded Mode C altitude information. Additional data was obtained from a global positioning system (GPS) receiver installed in GUU39. This GPS tracking data corresponded closely with the recorded primary radar information.

At about 1750, the pilot of GUU38 reported to the aerodrome controller (ADC) that GUU38 and GUU39 were at Shipyard, approximately 6 NM south-west of Jandakot at 1,500 ft and in receipt of terminal information Kilo, inbound. At about 1753, the pilot of GUU38 reported to the ADC that GUU38 and 39 were overhead Adventure World, approximately 3 NM west of Jandakot. The ADC instructed the crews to join left crosswind for runway 24R. This was to establish single-runway operations at the airport prior to the deactivation of the GAAP CTR at 1800 and accordingly, have aircraft in the zone complying with MBZ procedures when the airspace became non-controlled. A short time later the pilot of the C172 reported to the ADC that he was at Six South, approximately 6 NM south-southeast of Jandakot at 1,500 ft with Kilo, inbound to the circuit area.

As GUU38 and GUU39 approached the airport, the ADC provided instructions to the pilot of GUU38 to assist with entry to the nominated circuit and instructions for sequencing with an aircraft operating on a left circuit pattern for runway 24R. At about 1755, the pilot of the C172 reported at Forrestdale Lake (a tracking point, approximately 3 NM south of Jandakot) and the controller instructed the pilot to join the circuit upwind at 1,500 ft for runway 24R.

After sequencing GUU38 and providing the pilot with additional instructions to assist with his entry to the circuit, the ADC instructed the pilot to descend to 1,000 ft due to opposite direction traffic. At about 1757, the pilot of GUU39 reported downwind and the controller advised him that he was number three in the landing sequence and confirmed that he was following GUU38 ahead. The controller also instructed the pilot of GUU39 to descend to 1,000 ft due to an aircraft overflying at 1,500 ft.

As the C172 approached the airport, the ADC amended the circuit entry instructions and instructed the pilot to make a left turn and join left downwind and to follow the aircraft mid-downwind. The controller then clarified the instruction and confirmed that the pilot was required to join the circuit on left downwind for runway 24R. Recorded radar data confirmed that the pilot made a left turn and joined the circuit as instructed.

At about 1758, the ADC issued a landing clearance to the pilot of GUU38. Recorded radar data indicates that the aircraft was about to turn onto final approach at the time this clearance was issued.

At about 1759, the pilot of the C172 reported downwind for a full-stop landing for runway 24R. The ADC advised him that the aircraft he was following was mid-base. The pilot replied with the aircraft's callsign and did not indicate to the ADC that he could not see the aircraft ahead. The recorded radar data confirmed the position of GUU39 on mid-base at this time, with GUU38 established on mid-final. The pilot of the C172 recalled seeing an aircraft that he assessed was turning onto final and identified it as the aircraft that he was assigned to follow. He recalled last seeing this aircraft about the time he made his base turn and noted that this aircraft was now on short final approach.

The last primary radar return from GUU38 was received at 1759:41, when the aircraft was approximately 150 m from the threshold of 24R and at a radar derived ground speed of about 45 kts. Twelve seconds later, the ADC commenced an all-stations broadcast, advising that the tower was closing and that MBZ procedures applied. During this transmission the ADC described the traffic disposition as `the aircraft on final GUU39, KTV is a Cessna just turning base, NMS is an Eagle on downwind and SRR to join upwind'. The controller also stated that the CAGRO was in attendance. Recorded radar data confirmed the position of GUU39 on long final approach (approximately 1.4 NM from the runway threshold and having just completed the turn from base), and the C172 on late downwind and about to turn base. During a post-accident interview the pilot of the C172 could not recall the specific aspects of this closing broadcast.

The pilot of the Eagle broadcast his position on downwind a short time after the tower closed and advised his intention to conduct a touch and go. The CAGRO responded to this broadcast and indicated that there was one aircraft ahead, although there were actually two aircraft ahead of the Eagle when this transmission was made. The pilot of the C172 could not recall the specific aspects of this transmission during a post-accident interview and could not recall such a transmission altering his perception of the traffic ahead of him in the landing sequence.

At about 1800:39, the CAGRO made a broadcast on the MBZ frequency (that was the same as the aerodrome control frequency) advising that the CAGRO service was operating, that he had copied all reported traffic and requested details from taxying aircraft. The pilot of an aircraft taxying for departure provided the CAGRO details of his intentions and the pilot of GUU38 reported that he had landed and was taxying for the southern hangars. The pilot of GUU38 subsequently recalled that he was well clear of the runway when he made that transmission, probably in the vicinity of the run-up bay on taxiway B5, about 250 m from the runway turn-off.

At about 1801:16, approximately 30 seconds before the collision, the pilot of GUU39 broadcast his position as, `All stations Jandakot, GUU39 late finals runway 24, full stop'. (The terms late final and short final are interchangeable.) This broadcast was additional to the mandatory radio calls required when operating in an MBZ. The CAGRO acknowledged this transmission and read back the aircraft callsign. The recorded radar data indicated that GUU39 was approximately 550 m from the threshold of runway 24R at this time and the track log from the aircraft's GPS indicated that the aircraft was approximately 100 ft above the elevation of the runway threshold. The pilot of the C172 recalled hearing the transmission and something about `late final', but found the call hard to understand due to static breaking through on his radio. He realised that he was also on late final approach and looked to see if there was another aircraft ahead. As he could not see another aircraft ahead he considered that it was safe to continue his approach to land. The pilot of the C172 did not broadcast his position on short final and such a broadcast was not mandatory. Radar data indicates that at this stage of the approach, GUU39 was 200 m ahead of and about 20 degrees to the right of the C172, relative to that aircraft's track. Both aircraft faded from radar coverage a short time later.

The pilot of GUU39 reported that he had just started the landing flare and estimated that he was about 10 - 15 ft above the runway, when the nose of his aircraft pitched violently upwards. The canopy above his head shattered and the aircraft's nose pitched down before striking the runway and skidding to a stop. It was not until he vacated the aircraft that he realised that he had been involved in a collision with another aircraft.

The pilot of the C172 recalled that the approach to land was normal and he had checked that the runway was clear. He was just about to commence his flare for landing when he heard a loud bang. He realised that he had collided with another aircraft and the two aircraft slid a short distance together along the runway before coming to a halt. He had not seen the other aircraft prior to the collision.

The collision occurred at about 1801:47 and corresponded with a significant reduction in groundspeed recorded on the GPS track log from GUU39. The propeller of the C172 had struck the rear fuselage of GUU39 several times and had probably also shattered the cockpit canopy. There was no post impact fire.

Separation between aircraft during base and final approach

The recorded radar data indicated that as the pilot of the C172 turned base there was a lateral displacement of about 1 NM (1,850 m) between the two aircraft, with GUU39 established on long final approach in a relative position of about a 12 o'clock. Information recorded in the GPS track log from GUU39 indicated that at this time, it was at an altitude of about 650 ft, approximately 2,150 m from the threshold of runway 24R.

As the C172 turned onto final approach, the lateral displacement between the two aircraft was about 800 m. At the time the pilot of GUU39 broadcast his position on late final, the lateral displacement between the two aircraft had further reduced to about 200 m. The recorded radar data indicated that the two aircraft continued to converge, with the ground track of the C172 appearing to converge with GUU39 from the rear left quadrant.

The average groundspeed (radar derived, to the nearest 5 kts) of the C172 on the base leg of the circuit was calculated to be about 85 kts, 80 kts during early final and 60 kts on short final. The average groundspeed (radar derived, to the nearest 5 kts) of GUU39 on the base leg of the circuit was calculated to be about 75 kts. It was 50 kts during early final and 40 kts on short final.

Performance of aircraft on base and final approach

The pilot of GUU39 reported that the airspeed of his aircraft was about 90 kts as he turned base and that the speed was progressively reduced during the base leg. The final approach was flown at an airspeed of 60 kts with the aircraft configured with full flap.

The pilot of the C172 reported that the base leg of the circuit was flown at an airspeed of 70 kts and 65 kts on the final approach.

Meteorological and environmental information

The Jandakot Automatic Weather Station recorded a south-westerly surface wind of about 10 kts in the period leading up to the collision and visibility conditions were reported as being good. Last light at Jandakot was 1925 WST. At the time of the collision the sun was approximately 10 degrees above the western horizon and about 30 degrees to the right of the extended centreline for runway 24R.

Aircraft on final approach to runway 24R overflew a sand quarry and other areas of light-coloured terrain.

Colour scheme GUU39

The aircraft was predominantly white, with colour decals and stripes on the fuselage. The wingspan was about 8.5 m.

GUU39 pilot information

The pilot of GUU39 held a valid licence issued by the Czech Republic Light Aircraft Association. He also held an unrestricted cross-country certificate issued by the Australian Ultralight Federation and a restricted private pilot licence issued by CASA. He was entitled to operate the foreign registered GUU39 on the basis of his licence issued by the Czech Republic Light Aircraft Association.

C172 pilot information

The pilot of the C172 held a student pilot licence and had passed the general flying progress test. This permitted the student pilot to carry passengers during certain local flights. All flights conducted by a student pilot required the authorisation of a flying instructor. He held a CASA-issued Class 2 medical certificate, indicating that correction for distance vision was required. At the time of the accident, the pilot was wearing prescription contact lenses, together with polarised sunglasses.

Polarised sunglasses

Polarised sunglasses contain lenses that polarise the incoming light waves to the eye and reduce reflected glare from flat surfaces. Aviation literature identifies several limitations associated with the use of polarised lenses in aviation, including the potential for polaroid sunglasses to `mask the sparkle of light that reflects off shiny surfaces, such as another aircraft's wings or windscreen' (V. B. Nakagawara, R. W. Montgomery, K. J. Wood, Aviation Accidents and Incidents Associated With the Use of Ophthalmic Devices by Civilian Pilots, Civil Aviation Medicine Institute, Federal Aviation Authority DOT/FAA/AM-01/14 July 2001). The CASA Designated Aviation Medical Examiner's Handbook also states `Polarising sunglasses should not be used when flying. The polarising filter interacts with the cockpit transparency to produce a distorted and degraded visual field that poses a threat to air safety'.

Certified Air-Ground Radio Operator

The CAGRO operated each night, Monday to Friday, between 1800 and 2100 and provided pilots with basic operational information that may be relevant for aircraft operating at the airport. Although part of this service was to provide traffic information to pilots of aircraft operating inside the MBZ, it was not the responsibility of the CAGRO to continually monitor the relative position of aircraft or to use techniques to ensure that adequate separation was maintained between aircraft. The service did not provide any function associated with the provision of an air traffic control service and pilots remained responsible for all operations within the MBZ.

The CAGRO used the control tower facilities to provide the service. The operator on duty for the night of the accident had previously been employed as an air traffic controller by Airservices Australia (Airservices) and had extensive experience in control tower operations. He held a CASA-issued Air/Ground Radio Operator Certificate.

The CAGRO had commenced duty concurrently with deactivation of the GAAP CTR. He did not observe the two aircraft converging on final. A large concrete support pillar could restrict the view of the airspace associated with the approach to runway 24R and may have obstructed the CAGRO's view of the two converging aircraft on short final approach.

Tower operating hours

Since May 1997 Airservices had reviewed staffing levels and tower operating hours at Jandakot. Similar reviews were also performed at other GAAP airports. The Jandakot reviews had resulted in the reduction of both tower operating hours and staffing levels.

In an attempt to make tower operations more cost effective, staffing levels were further reduced during December 1998 and resulted in a reduction of controller numbers from ten to seven. This reduction in staff numbers was facilitated by a reduction in tower operating hours and various procedural changes to the conduct of flight operations at Jandakot. Prior to implementing this change, Airservices conducted a review of airport traffic levels and prepared a safety case to assess the impact of the proposed changes. This safety case was submitted to CASA and no objection was raised to the proposal to reduce staffing levels and to reduce tower operating hours.

Following the reduction in tower operating hours there was a perception among local operators that operations had become less safe. These assessments were based on their perception that the traffic density sometimes reached levels that were unsuitable for MBZ operations and also that a significant number of aircraft were failing to comply with the published MBZ procedures.

In response to concerns from the local operators and to address an issue of some pilots failing to transmit the required information on the MBZ frequency, the airport operator trialled and then implemented a permanent CAGRO service during evening week days. Data collected during the operation of this service indicated that traffic density occasionally exceeded the upper limits used to define the risk for the purpose of the safety case prepared by Airservices to assess the acceptable levels of safety during MBZ operations.

Events post-collision

Following the collision the airport operator reviewed the ongoing provision of the CAGRO function and decided to discontinue the service.

Traffic complexity at time of collision

In the period leading up to the collision there was one aircraft conducting circuit operations. The two Ultralight Sting aircraft had joined the traffic pattern for a landing and were followed by the C172, also for a landing. At the time of the collision there were three aircraft in the circuit and another inbound aircraft joining the traffic pattern upwind. Several aircraft were taxying for departure. Of the three aircraft operating in the circuit, all were on the final leg when the collision occurred.

The pilot hired the de Havilland DH82A Tiger Moth aircraft, registered VH-AJG, to undertake a local pleasure flight with a friend. The aircraft took off from a grassed area parallel to the Royal Australian Air Force (RAAF) base Williamtown's main runway between the eastern end of the runway and taxiway `A'. During the take-off, the aircraft was observed to veer right and strike and destroy two taxiway lights with the right wheel. The pilot continued the take-off and the aircraft departed the Williamtown circuit area at 1428 Eastern Summer Time.

Based on witness reports and radio transmissions by the pilot, the aircraft initially conducted a sightseeing flight over the coastal suburbs of Newcastle City. About 20 minutes after takeoff, the pilot broadcast that he was transiting north through the Williamtown Mandatory Broadcast Zone west of the coast. The aircraft subsequently joined the Williamtown circuit from the north at 1519. The actual flight profile and manoeuvres conducted during the flight are not known. Shortly after joining the circuit, the aircraft was observed to depart level flight and impact the ground approximately 2 km southwest of the Williamtown airport. Both occupants were fatally injured and there was no evidence of fire in flight or after the impact.

A helicopter with an instructor and student on board was in the circuit area, about 1.5 km behind the Tiger Moth. The helicopter was about 800 ft above ground level and maintaining approximately 60 kts. The helicopter crew estimated that the Tiger Moth was flying at the same speed and altitude. The helicopter's pilots reported observing the left wings fold up, the aircraft rotate and fall almost vertically in a steep nose-down attitude rotating only a couple of times before impacting the ground.

A witness on the ground reported observing the aircraft's right wings fold back, followed by the aircraft spinning or spiralling to the ground. Another witness reported seeing the right wings folding up, making the wings into an `L' shape and about a metre long silver pole flying up from the cockpit area. The aircraft then started turning to the right. None of the witnesses reported observing the aircraft changing altitude or commencing a turn prior to the loss of control.

Personnel information

The pilot held a valid Australian Special Pilot Licence and had accumulated 360.6 hours total of which 4.1 hours were on the Tiger Moth. The pilot's tail-wheel experience totalled 7.9 hours. He was endorsed on the type and flew earlier that day with an instructor. The accident flight was his first unsupervised flight on the type.

The Australian Special Pilot Licence authorises a pilot with a current overseas private or higher class of licence to undertake private flights in Australia. The pilot held a United Kingdom private licence. He held an Australian Class 2 (private) medical certificate valid until October 2005, issued by the Civil Aviation Safety Authority (CASA) on the basis of his United Kingdom medical. There was no evidence of any physiological condition affecting the pilot that may have contributed to the occurrence.

Operation and weather

The aircraft was operated by the RAAF Williamtown Flying Club located on the base and the accident flight was the aircraft's third flight of the day. The weather was reported to be fine with a light north-easterly breeze.

The aircraft and its history

The Tiger Moth was a fabric covered biplane aircraft with two open cockpits in tandem arrangement. The pilot sat in the rear cockpit. The truss type fuselage was made from steel tubes while the wings and tail surfaces were constructed from wood. The two-spar, single-bay wings were reinforced by a system of drag struts, drag and anti-drag wires. Each spar was made from a single piece of wood with a reinforcing doubler glued at the interplane strut attachment. The spars were fitted with metal fittings used to attach the wings to the fuselage and cabane centre section upper struts. The flying, landing and cabane wires, tie rods and interplane struts gave the wings the required rigidity.

The aircraft was stressed to withstand maximum loads of approximately 7.5g (acceleration due to earth gravity). Information from the manufacturer indicated that even with the reinforcing doubler delaminated and ineffective, the aircraft was designed to withstand manoeuvre loads of about 5g.

The aircraft was manufactured in 1942 and used by the RAAF until August 1947, when it was decommissioned and received civilian registration. The aircraft was substantially damaged during an accident in 1967. It was then dismantled and stored until 1988 when it was rebuilt and had since accumulated approximately 48.35 flying hours. The fuselage truss structure was repaired and an overhauled engine was installed. The aircraft was fitted with four wings and propeller of original manufacture. The previous history of the wings and the propeller could not be determined.

In November 2001, the left lower wing was damaged when the landing gear collapsed on landing in Newcastle after a ferry flight from Bankstown. A new wing was manufactured and fitted in December 2001. On 13 January 2002, the left lower wing contacted the ground while the aircraft taxied after a flight.

The aircraft was used to perform only limited aerobatic manoeuvrers such as barrel rolls, loops and stalls. A few steep turns and dives were performed during the flights on the morning of 16 February 2002.

The aircraft was maintained in accordance with the applicable and current maintenance requirements. The maintenance release was valid until 9 March 2002. Examination of maintenance documents indicated that all required maintenance had been carried out. The aircraft had no known maintenance deficiencies and was considered capable of normal operation prior to the accident.

Accident site and wreckage examination

The impact site was an area of dense undergrowth, tall grass and 10 to 12 m high trees. The ground was soft and waterlogged. The wreckage was contained within a small area at the foot of a tall tree, its distribution indicated that the aircraft impacted in a steep nose-down attitude while rotating to the left. The aircraft and its four wings were extensively damaged. The wing spars were splintered and broken at numerous locations. The right wings struck a tree during impact and their damage was significantly more severe than the damage sustained by the left wings.

With the exception of both propeller blades and parts of the right wing rear spar interplane struts, all aircraft components and extremities were accounted for at the accident site. The propeller boss that remained attached to the engine shaft was the only part of the propeller recovered. Propeller contact marks on the tree indicated that the propeller was rotating at the time of impact. A part of the missing interplane strut was found some six months after the accident approximately 73 m from the accident site.

Examination of the engine, systems and flight controls did not reveal any pre-impact defect that would have prevented them from normal operation. The left front and rear flying wires and the left cabane wire were severed in overload. There was no fuel remaining in the ruptured fuel tanks, but fuel was evident at the accident site.

Sections of the wing spars were examined by a timber specialist who concluded that the wood was of high quality and in good condition. It was free of decay and there was no evidence to indicate that a failure of the wood was a factor in the accident. Some blue colouring of wood was present in one laminate of the propeller boss, but there was no evidence to link this discolouration with failure of the propeller. The specialist reported that many of the wood failures, on the three wings of the original manufacture, were brash rather than splintering, indicating some embrittling associated with the age of the wood.

Adhesive failure

The specialist also reported that some laminated members on the three wings of the original manufacture and the propeller failed at the glue line rather than in the wood. The failure was due to the glue line being devoid of the adhering adhesive. The adhesive was identified as Casein. The specialist reported that the observed failure was typical of that of Casein that was exposed to attack by micro-organisms and that the attack was evident all over the glued area of the joint except on small areas close to the edges of the glued components. No delamination was observed on the new wing fitted in December 2001. This wing was manufactured using modern synthetic resin adhesive.

Casein is a milk-based glue that was particularly popular around the 1940s, when the three wings and the propeller were most likely manufactured. Since it contains protein, it could be subject to attack by micro-organisms and weakened if the moisture content of the wood and the adhesive is allowed to increase above a certain level. The Casein glued joints, however, do not degrade instantaneously when wet. The amount of degradation is proportional to the time the joint is allowed to remain moist.

The specialist advised that the attacks by micro-organisms occurs when the wood and adhesive moisture content is approximately 18% or greater. A moisture content of 18% could be achieved if the wood was exposed to a relative humidity of 85% or greater. While such humidity is experienced in tropical Australia, the average moisture content of the wood and the adhesive is not likely to reach this level due to lower values during various times of the day. The evidence, however, indicated that the glued joints were probably subjected to a number of periods when the moisture content was high, allowing micro-organisms to attack the adhesive. Such periods were further evidenced by the presence of corrosion around the nails, bolts and screws securing the joints.

However, the specialist concluded that there is little or no evidence indicating that any single glue failure may have resulted in a catastrophic failure of a major structural element of the aircraft or the propeller.

Propeller

The numbers stamped on the propeller boss were consistent with the propeller having been manufactured in the 1940s. The wooden propeller consisted of a number of laminates. After the propeller was manufactured, the centre part of each blade was wrapped with a layer of `Irish linen'. The entire propeller was then coated with a relatively thick coat of dark coloured cellulose-based paint.

A Boeing 747-48HE registered VH-OEB (OEB) was en-route from Los Angeles, USA to Auckland, New Zealand was maintaining flight level (FL) 330 as assigned by Tahiti air traffic control (ATC). A Boeing 747-4H6 registered VH-OED (OED) was en route from Auckland to Los Angeles and was also maintaining FL330. The crew of OEB reported that they observed, on their traffic alert and collision avoidance system (TCAS), another aircraft that was on a reciprocal track at the same level (OED). The crew of OEB turned their aircraft right 15 degrees and descended to FL325. The crew of OED later reported that they observed, on their TCAS, another aircraft that was on a reciprocal track at the same level (OEB), and climbed their aircraft to FL333. A third aircraft, a Boeing 747, was en-route from Los Angeles to Auckland at FL340. The crew of OED also observed an indication of that aircraft on their TCAS.

The vertical separation standard was 1,000 ft. The vertical distance between OED and OEB reduced to 800 ft, and to 700 ft between OED and the third aircraft. There was an infringement of separation standards.

The crews of OEB and OED were communicating with Tahiti ATC via both Controller-Pilot Data Link Communications (CPDLC) and high frequency (HF) radio. CPDLC was a 'means of communications between a controller and pilot using data link for [Air Traffic Control] communication' (ICAO Doc 4444 ATM/501 14.1.1). Messages were compiled and initiated either by the crew of the aircraft or by ATC and were, in this case, pre-formatted. The use of pre-formatted messages was 'intended to reduce the possibility of misinterpretation and ambiguity' (ICAO Doc 4444 ATM/501 14.3.4).

The crew of OEB had requested climb from FL320 to FL340 but that request was denied. About ten minutes later the crew of OED requested climb from FL330 to FL350. Tahiti ATC asked the crew of OED, via CPDLC, when they could reach FL350 and then denied the request for climb. The French Bureau d'Enquetes et d' Analyses pour la Securite de l'Aviation Civile (BEA) produced a report on the occurrence. The BEA reported that the crew of OED then contacted Tahiti ATC via HF radio and advised that they could reach FL350 by time 1140 universal coordinated time. The controller responded via HF radio and instructed the crew of OED to maintain FL330. The crew of OEB then requested, via CPDLC, climb to FL330. The CPDLC response provided to the crew of OEB was 'climb to and maintain FL330 due to traffic' even though FL330 was not available. The message was selected by the controller from the menu of pre-formatted messages available in the system.

The South Pacific Operations Manual (SPOM Part 5.5) stated that 'when a clearance request is denied, the controller shall use the element "UNABLE" in the uplink message'. The SPOM detailed the procedures and requirements applicable in the South Pacific flight information regions (FIRs) for CPDLC equipped aircraft and applied within the Tahiti FIR. The SPOM (Part 5.1) also stated that `generally, when a CPDLC aircraft is operating within a CPDLC FIR, CPDLC will be the primary means of communication'.

Subsequent to the occurrence OEB returned to FL320 and OED returned to FL330. The crews then reported to Tahiti ATC at those respective levels. The controller had not intended to assign FL330 to the crew of OEB and did not realise that they had been assigned FL330, or that they had climbed to FL330 and subsequently returned to FL320. When the crews reported at FL330 and FL320 respectively, after the occurrence, that information was consistent with the information the controller had recorded on the flight progress strips. The controller was not aware that there had been an infringement of separation standards.

The BEA reported that the controller believed there were possibly two reasons why a climb instruction had unintentionally been assigned to the crew of OEB:

1. In reply to the request by OEB for climb to FL330, the controller pre-selected the wrong pre-formatted CPDLC message and sent the message without checking it, or

2. The controller confused the two aircraft because of their similar callsigns.

The BEA reported that four controllers had been rostered for the period between 1900 hours and 0700 hours (Papeete local time) and were rostered to cover the aerodrome control position, the approach control position and the area control position. The event occurred at 0050 Papeete local time. The controller involved in the occurrence was alone in the tower at the time of the occurrence and was performing all three functions. That controller considered that the workload was high due to poor quality HF radio, increased coordination with other centres in relation to aircraft using 'flexible routes' and difficulty validating CPDLC messages with OED.

The controllers at Tahiti had been trained in France but that training had not included the use of CPDLC. Initial training on the use of CPDLC was incorporated into a one-week training program in Papeete that included CPDLC with other local training requirements. Ongoing CPDLC training was incorporated into on-the-job training which could take controllers around 57 weeks to complete. Controllers reported that the initial training was essential but they had not been exposed to the system sufficiently during training to master all aspects of the system.

The controller involved in the occurrence had been working Tahiti Oceanic Controlled Airspace for approximately three and a half years and was qualified on the three ATC positions being managed at the time of the occurrence.

The pilot of a Cessna 206 (C206), departed from Badu Island, Qld at about 1210 Eastern Standard Time (EST) on a positioning flight to Horn Island, Qld in accordance with the visual flight rules (VFR). The aircraft had an estimated fuel endurance of 270 minutes. The pilot, who was the sole occupant, had been tasked to conduct a charter flight from Horn Island at 1330 with passengers who were reported as arriving on a scheduled flight from Cairns.

At about 1221 the pilot broadcast on the Torres Mandatory Broadcast Zone (MBZ) frequency that he was over Wednesday Island and tracking for a 3 NM final approach to runway 26 at Horn Island. A short time later he broadcast that he was holding until the weather over the runway cleared. At about 1238 the pilots of two aircraft in the Bamaga area reported hearing a MAYDAY broadcast from the pilot of the C206 on the MBZ frequency. The pilot did not describe the nature of the emergency. Further efforts by the pilots operating in the Bamaga area to contact the pilot of the C206 were unsuccessful and they advised air traffic services of the MAYDAY they had heard on the MBZ frequency.

An air and sea search was commenced. Later that day floating debris, identified as belonging to the C206, was located. The recovered items included the left main wheel and landing gear leg, the nose wheel and part of the nose gear landing leg and a seat. The following day divers located the aircraft approximately 3 NM east of Horn Island in 7 m of water but the pilot was not found.

The aircraft was recovered from the sea floor where it had been submerged for just over two days. It had been extensively damaged by impact forces. The nature of the recovery process resulted in further disruption of the wreckage. The outer left wing, left aileron and the engine cowls were not recovered. Salt-water corrosion had affected many of the aircraft components. The flaps were found in the retracted position and damage to the propeller blades was consistent with low engine power at the time of impact. Damage was consistent with the aircraft having struck the water at a moderate to high speed in a nose-down, left wing low attitude. The collision with the water was not survivable.

Examination of the damaged gyroscopic flight instruments did not reveal any indication of pre-impact malfunction. The vacuum pump was found in good condition and capable of normal operation. Although the aircraft was equipped with appropriate instrumentation for flight in instrument meteorological conditions (IMC) it was maintained to the VFR standard, as appropriate to the category of operation. Maintenance requirements were certified as having been performed and no evidence was found to indicate that the aircraft was other than serviceable prior to the flight. Analysis of recorded audio data determined that engine operation was normal during the pilot's radio transmissions.

The pilot was reported to have obtained a forecast from Airservices electronic briefing facility using the company computer terminal at Badu Island. That forecast predicted north-westerly stream weather characteristic of conditions normally experienced during the wet season in the Torres Strait area between October and April. The terminal area forecast for Horn Island, valid from 0600 to 1800, indicated light showers of rain and a visibility greater than 10 km. Throughout the forecast period, a temporary deterioration in conditions (up to 60 minutes) was forecast due to thunderstorms, with visibility reduced to 2000 m in rain and a cloud base of 800 ft. Approximately 40 minutes before he departed Badu Island, the pilot received a telephone call from the senior base pilot. The senior base pilot had flown from Badu Island to Horn Island that morning and advised him of the actual weather conditions he had encountered, including 20 minutes holding east of Horn Island while awaiting a rain shower to move from over the aerodrome.

Witnesses reported that the weather conditions at Horn Island aerodrome between 1230 and 1245 were less than visual meteorological conditions. Heavy rain had reduced visibility to less than 100 m. One witness reported seeing lightning to the north of the aerodrome. A Bureau of Meteorology assessment of weather conditions for the area east of Horn Island at the time of the occurrence indicated that the generally low overcast cloud contained embedded thunderstorm cells, with associated heavy rain and a cloud base less than 1,000 ft. The recorded rainfall intensity was heaviest between 1230 and 1300. Weather conditions at Horn Island at the time of the occurrence were described by witnesses as being the most severe seen that season.

The pilot held a Commercial Pilot (Aeroplane) Licence and a valid Class 1 medical certificate. He had obtained his commercial pilot licence in September 2000 and qualified for a command multi-engine instrument rating in April 2001. The pilot's logbook was not found. A compilation of flight time records showed that at the time of the occurrence the pilot had between 270 and 290 hours total flying experience that included approximately 45 hours on type. The company did not require the pilot to maintain IFR currency there was no evidence that the pilot had met recency requirements for instrument flight. The pilot was not reported to be suffering from any physiological condition that may have affected his capability as a pilot. He had been off duty for the two days prior to commencing duty on the morning of the accident.

In July 2001 the pilot moved to the Torres Strait area and gained some occasional flying experience with another operator. In November 2001 the pilot commenced employment with the operator he was working for at the time of the accident as a VFR charter pilot on C182 and C206 aircraft. On 19 November 2001, before commencing operational duties, the pilot had flown a familiarisation flight under supervision of the senior base pilot. On 27 November he had demonstrated proficiency on the C182 in normal and emergency procedures to an approved company check pilot. That flight of one and a half hours also incorporated a short area familiarisation. The pilot was then certified as competent to conduct company charter flights. On 1 December 2001 he flew the C206 while acting in-command under the supervision of the senior base pilot.

The flight was being conducted under the VFR, at an altitude that required the aircraft to remain clear of cloud, and with a minimum flight visibility of 5,000 m. The operator's operations manual instructed pilots to consider uplifting additional fuel for diverting or holding when the forecast indicated elements of weather below the minimum required for the flight. The pilot had departed with ample fuel reserves for holding or diverting. Company pilots reported that diversions and holding, due to rain showers and associated poor visibility, were not unusual during the wet season. The pilot had broadcast his intention to hold until weather conditions improved.

Although the pilot of the C206 had flown in IMC during his training he did not have any instrument flight recency and had very little exposure to tropical wet season weather conditions and its characteristic heavy rain shower activity.

The circumstances of the occurrence were consistent with a loss of control at low level and at an altitude from which recovery was not considered possible. Due to the limited information available to the investigation, the reason for the loss of control could not be determined. However, the circumstances were consistent with VFR flight into IMC.

The pilot had planned to conduct a charter flight, with three passengers, from Essendon to Latrobe Valley, Vic. in a twin engine Beech Duchess aircraft. The pilot reported that he arrived at the Essendon airport about 90 minutes prior to the scheduled departure, carried out the daily inspection on the aircraft and submitted an instrument flight rules flight notification. He checked the aircraft maintenance release and the company fuel log that included the aircraft's last flight four days earlier. The pilot reported that he checked the two fuel quantity gauge readings which indicated a half-full tank and a slightly less than half full tank. He then visually checked the contents of the tanks relative to the '30 US gallon' metal tabs that are visible through the filler opening. He estimated that the fuel tanks contained a total of about 200 litres, but did not confirm this, as a fuel tank dipstick was not provided for that aircraft. The pilot's flight plan indicated that 128 litres of fuel, including reserves, would be required for the flight.

During the climb to the planned altitude of 7,000 ft, the right engine's power reduced. The pilot concluded that, from the engine tachometer reading of 1,500 revolutions per minute, and the manifold pressure indications, the right engine had partially failed. He carried out engine failure confirmation checks, but as the propeller pitch lever was very stiff, was unable to place it in the feather position. The pilot later reported that, during manipulation of the pitch lever, the aircraft had yawed significantly. Therefore, he decided to reset the right engine controls to a cruise setting because partial power was preferable to no power.

The aircraft was unable to maintain altitude so the pilot decided to return to Essendon. He requested an air traffic clearance from the Melbourne Approach air traffic controller (ATC) when the aircraft was about 50 NM east of Essendon. After receiving a clearance, and as he turned onto a westerly heading, the aircraft descended into cloud. At the pilot's request, ATC provided headings for the pilot to track further to the south to avoid the higher terrain on the direct track to Essendon. Lowest safe altitude (LSALT) is a published or pilot calculated minimum altitude that ensures terrain clearance during flight in instrument flight conditions. Flight below an LSALT altitude is only permissible during visual meteorological conditions or while conducting a published instrument approach. When the aircraft descended below the LSALT, ATC advised that a diversion to the closer Moorabbin airport, which was to the south east of Essendon and to the south of the aircraft, was available. That option would have allowed the aircraft to track over lower terrain and would have minimised the track distance over the Melbourne suburbs. However, the pilot decided to return to Essendon, where the operator's maintenance facilities were located and the passengers could be transferred to another company aircraft. After descending through the next LSALT step while in cloud, the aircraft descended into visual conditions about 21 NM east of Essendon at about 2,500 ft. The aircraft continued to descend until it stabilised in almost level flight at about 1,500 ft. The pilot then tracked direct to Essendon and carried out a visual approach and landing.

The pilot had bypassed two other suitable airfields, Lilydale and Coldstream, approximately 10-15 NM to the right of his track. The Civil Aviation Safety Authority (CASA) Civil Aviation Orders (CAO) 20.6 permitted the pilot of an aircraft with a failed engine to fly past a suitable aerodrome if another suitable aerodrome was available nearby and the pilot assessed that the aircraft could be flown safely to that aerodrome.

Company engineering inspection of the aircraft found that the right fuel tank, that was supplying the right engine when it lost power, contained no fuel. The right fuel quantity gauge transmitter unit was corroded and seized in a position that resulted in the gauge always indicating half-full. It was possible to feather the right propeller, although the pitch control was stiff.

The control cable and the fuel tank sender unit were subsequently replaced. The pilot commented that he had conducted a feather check as part of the pre-takeoff checks and although the right pitch lever was stiff to operate, he was satisfied that the propeller feathering mechanism was operating satisfactorily. The pilot later commented that the engine had failed due to fuel starvation and that he had not recognised the symptoms of a piston engine failure. He reported that he did not notice the reduced fuel pressure to the right engine until after the aircraft descended into visual conditions.

The aircraft's pilots operating handbook cautioned pilots against attempting to determine the inoperative engine by reference to the tachometers or the manifold pressure gauges and stated that those instruments often indicated near normal readings after an in-flight engine failure.

The pilot had recently resumed employment with the operator. He began his career with the operator and had flown as a first officer on F-27 turboprop aircraft for four years. He then obtained employment with a regional airline for two years as a first officer on turboprop aircraft, and had just completed line training as first officer on a jet aircraft when that airline suspended operations.

The day before the incident, the pilot had completed a 12-hour tour of duty, including 8.1 hours of flight time completing a co-pilot endorsement on a business jet. The two days prior to that had been spent on ground duties. He had 6-8 hours rest overnight at home and had risen early on the day of the occurrence. The pilot reported that he was tired on the day of the occurrence and that he had felt similarly for some time. He had been on duty for 16 consecutive days or a total of 159.4 hours duty time, primarily in a capacity unrelated to his employment as a pilot. Those additional duties were reflected in the pilot's recorded duty times.

The pilot had logged about 3,600 hours total flight time, including 600 hours in command on piston-engine aircraft. Those command hours consisted of about 200 hours twin-engine, of which 25 were in Duchess aircraft. All his other flying had been in turboprop and jet aircraft. During the three months preceding the incident, the pilot had flown approximately 70 hours, but had only flown the Duchess for three hours during that time.

The pilot's work/rest history for the four weeks prior to the incident was examined using a computerised fatigue algorithm developed by the Centre for Sleep Research, University of South Australia. The results indicated that the pilot was probably experiencing moderate levels of fatigue in the week leading up to, and on the day of the incident.

During cruise flight, the crew of the Metro III aircraft noticed a burning smell and smoke in the cockpit. The crew discovered that the problem was associated with avionics bus number 2, and isolated the bus. Following this action, the smoke and burning smell dissipated and the flight was successfully completed.

The investigation revealed that a short circuit had occurred in the right avionics bus circuit breaker panel and that the aircraft was being operated with a minimum equipment list item open for a faulty press-to-talk (PTT) for the co-pilot's audio system. Damage was sustained to insulation of the wiring and the bus link in the co-pilot's audio loom, including the PTT wires. The damage had resulted in a short circuit between the right avionics bus circuit breaker bus supply link, and the co-pilot's audio wiring loom. The wiring loom was physically secured to the circuit breaker busbar.

The damage to the insulation of the link wire was in the form of cuts, possibly from manoeuvring the right avionics bus circuit breaker access panel past sharp edges on the panel housing during routine maintenance. The insulation damage ranged in depth and in some areas exposed or damaged the copper conductor. There was also evidence of heat moulding from physical contact between the bus link and the co-pilot's audio wiring loom, due to excess current draw from the faulty PTT.

The routing of the PTT wiring loom ran along the circuit breaker busbar from the right avionics circuit breaker bus. Chafing against this bus may have caused the original PTT defect, and combined with the damage to the bus link wire, produced a short circuit from the bus link wire through to the earth wires in the co-pilot's audio wiring loom. The short circuit bypassed individual circuit breakers allowing a current draw of up to 100 amperes through the right essential bus 100 ampere circuit breaker switch.

During power reductions for landing, the engine compressor air scheduling to the left engine was incorrect. That resulted in disrupted airflow throughout the engine and subsequent stalling. The discrepancy in the air scheduling of the engine was due to the incorrect performance of the engine vane bleed control unit. This anomaly could also have resulted in excessive amounts of fuel for the power setting and contributed to the fire witnessed coming from the left engine exhaust as reported during the 5 January 2002 occurrence on this aircraft (see ATSB report 200200006).

Local safety action

As a result of this occurrence, the third party maintenance organisation advised, 'As a preventative measure we have briefed our maintenance personnel to trim or adjust to mid band position when trimming or adjusting to a given band.'