The flight was planned as a familiarisation flight for the passenger. The pilot did not detect any fault with the engine during the pre-flight inspection, and he reported that the compression on all cylinders appeared normal. The aircraft was close to its maximum weight at take-off.

During the take-off roll engine RPM was 2800. Maximum RPM permitted was 3300 however, the manufacturer reported that the engine/propeller combination fitted to the aircraft was only expected to produce between 2900 and 2950 RPM. Although the take-off performance appeared below normal, the pilot believed it was caused by the aircraft being close to its maximum weight. Witnesses reported that the aircraft used almost all of the 900 m strip for its take-off roll. The manufacturer reported that the take-off roll in a Jabiru 1600 at maximum weight should be between 250 and 300 m.

After becoming airborne the aircraft would not climb, and engine RPM had fallen to 2700. The pilot turned to follow the road in case the engine lost more power. He completed trouble checks but was unable to identify the reason for the loss of performance. Shortly after, the engine regained some power, and the pilot decided to attempt a return to the airstrip.

During the subsequent turn the engine stopped completely. The pilot then attempted a forced landing in a tree covered area. Whilst attempting to avoid a medium sized tree, at a height of approximately 20 ft above the ground, one wing collided with the tree and the aircraft stalled and crashed. The fuselage turned through 180 degrees and slid backwards into another tree before coming to a stop.

Although the aircraft was extensively damaged the cockpit area was relatively intact. A fuel filter had fractured allowing fuel into the cockpit however, there was no fire. The pilot turned the fuel off and secured the electrical system before he and the passenger exited the wreckage.

An inspection of the engine disclosed that the exhaust valve, seat and spring on the No. 3 cylinder had been subjected to excessive heating. The valve guide was loose in the cylinder head and the valve spring had lost its tension. There was an accumulation of grease and metal shavings in the rocker cover which had been there for sufficient time to congeal indicating that the valve, seat and spring had been in this condition for some time. The faults meant that the valve might or might not seat correctly during engine operation. The variable nature of the valve operation could lead to an intermittent power loss at those times when the valve was not seated correctly. It is possible that the valve was seated correctly when the pilot checked the compression during the pre-flight inspection; however, the post-accident investigation was unable to reproduce normal cylinder compression. It is possible that the low RPM and excessive take-off distance, reported by the pilot, was a symptom of the valve problem.

The engine was fitted with a reconditioned cylinder head and the manufacturer reported that there had been other cases of valve guides coming loose. To manage this type of problem the manufacturer had included a compression check and, if this proved irregular, a tappet check in the daily inspection requirements. Although the precise reason for the valve guide failure could not be determined the guide is held in place by a using a temperature-dependent shrink-fit. As there was an indication of excessive heating, it is possible this caused the fit to come loose.

Although the valve position could lead to a loss of power, experience indicates it would not normally be sufficient to stop the engine. It could not be determined why the engine stopped. The manufacturer reported that any reduction in power, during three cylinder operation, might stop the engine. The pilot reported that he did not reduce power.

At the low speeds, normally used during a forced landing, any abrupt manoeuvres made to avoid trees would probably cause the aircraft to stall.

SAFETY ACTION

The manufacturer has discontinued production of the 1600 cc engine and replaced it with a 2200 cc engine which is manufactured using a different process.

This will alleviate the problem with the valve guides. The manufacturer is also working to relocate the fuel filter to prevent fuel from entering the cockpit in the event of a failure.

The owner reported that the tail rotor of the helicopter came into contact with a small tree as the pilot was pushing some cattle into a yard. The helicopter then spun out of control to the ground, landing heavily on the left skid before rolling over. The pilot was able exit the helicopter without injury. The owner advised the helicopter was damaged beyond economical repair.

Post flight examination showed that one of the tail rotor blades had separated in flight when it came into contact with the tree.

Weather at the time of departure from Emerald was reported to be 6 to 7 OKTAS of low cloud with patches of rain in the area. Light rain was falling at the time the aircraft taxied. The wind was light with a slight crosswind from the right. Visibility was in excess of 10 KM. The runway was damp and there were several patches of water on the runway.

The pilot reported that the take-off proceeded normally until approximately 50 knots when he disengaged the nose-wheel steering. At this point, the aircraft started to diverge to the right of the centreline. In an attempt to return the aircraft to the centreline the pilot used rudder, but this was ineffective. The pilot then rejected the take-off by moving the power levers to idle, but the aircraft continued to diverge to the right and departed the sealed surface of the runway.

The pilot used reverse thrust in an attempt to slow the aircraft, but this appeared ineffective, and the aircraft continued to turn to the right and slide on the wet grass surface. The aircraft by this time was heading for a small levee bank beside a drainage ditch. The pilot assessed that the aircraft would probably climb over the levee bank, so he elected to introduce power on the left engine in an attempt to turn the aircraft before the levee bank. However, this was only partially successful, and the aircraft came to rest adjacent to the levee bank facing approximately 180 degrees from the take-off heading. The left propeller contacted the bank as the aircraft stopped.

The aircraft had travelled 121 M after leaving the runway. The distance at which the aircraft left the centreline of the runway could not be accurately determined due to the runway surface. The nose-wheel track was approximately 1 metre to the left of the right-hand wheel track and continued at this distance until approximately 5 metres from the end of the tracks, at which point it crossed over the top of the right wheel track.

The aircraft suffered damage to the left propeller and engine. Examination of the engine when removed from the aircraft confirmed that it had suffered considerable internal damage consistent with a propeller strike at high power.

The operator reported that there had been several minor problems with the nose-wheel steering system in the 5 weeks prior to the accident. These problems, however, caused the system to fail in the off mode.

The nose-wheel steering system was examined and a series of tests conducted. It was found that with the nose-wheel steering armed, any movement of the speed levers caused an uncommanded movement of the nose-wheel to maximum right deflection. The fault was traced to the steering amplifier and this was subsequently removed for further testing. Once the amplifier was removed from the aircraft and replaced with a serviceable unit, the fault did not re-occur. Extensive testing of the amplifier has been unable to reproduce the fault.

The patches of water on the runway were not considered to be a factor in this accident.

A significant proportion of this report is based on information provided by the operator.

Whilst approaching Bindook at FL180, on a scheduled air transport flight from Dubbo to Sydney, the crew of a SAAB 340B observed the right engine air intake caution light illuminate. The abnormal procedures checklist for this condition was carried out and the caution light extinguished.

A short time later a strong burning smell became evident on the flight deck. However, there were no associated warning indications. Discussions with the flight attendant (FA) revealed that the smell was isolated to the cockpit, indicating that the burning smell may have been associated with the right air conditioning system, which supplied conditioned air to the cockpit. The 'Air Conditioning Smoke' emergency checklist was not utilised. The right engine bleed air valve was closed in an attempt to further diagnose the problem. Several minutes later the bleed air valve was re-opened. The burning smell in the cockpit intensified. Reports from the FA continued to indicate no noticeable smell in the cabin.

The co-pilot then looked out the right, side window and reported to the pilot in command (PIC) that there was a fire in the right engine. There were no fire warning indications in the cockpit. The engine fire checklist was carried out, the right engine fire bottle was discharged, and the engine shut down. An inflight emergency was declared to Air Traffic Control, who initiated a distress phase and expedited the entry of the aircraft to Sydney.

All three crewmembers then prepared for an emergency landing at Sydney. The PIC advised the FA that the landing would be normal, and would be followed by a normal disembarkation on an adjacent taxiway. With local emergency services in attendance, the aircraft subsequently landed safely on runway 07 and was stopped on taxiway G3. All occupants disembarked normally through the main cabin door and were transported to the terminal.

An engineering examination revealed that the right engine lower intake anti-ice duct had short-circuited and was excessively burnt around the inner-upper lip at the particle separator mounting flange. The damage was consistent with the duct material smouldering for some time. Based on the recollection of the co-pilot, there had been no visible flames but rather a very bright red/orange glow inside the intake. The lower duct, and the particle separator, were replaced with serviceable items and the aircraft returned to service.

The engine intake anti-ice system consists of electrically heated intake ducts. The heater elements are made of a copper alloy material sprayed onto a glass fibre mat and covered with an outer glass mat. The mat is bonded to the metal inlet, making them an integral unit. Should the duct intake surface suffer damage which allows moisture to penetrate to the heater elements, short-circuits can occur. Although required as part of the engine fire emergency checklist, the activation of the fire extinguisher has no effect on the engine intake duct. The intake duct anti-ice remote circuit breaker is designed to trip as a result of a short-circuit. In this case the circuit-breaker failed to trip.

The aircraft manufacturer has been aware of short-circuiting intake duct heaters in aircraft equipped with this particular type of intake, and had developed a specific checklist titled 'Intake Sparks' to cover this particular situation. At the time of this occurrence the checklist was being distributed as an amendment to the Aircraft Operations Manual, but had not been available to the flight crew.

Local safety action

As a result of this occurrence the company has introduced improvements to flight operations training and procedures, designed to assist flight crews in the recognition and handling of intake duct problems. In addition, both flight and cabin crew emergency disembarkation procedures are being reviewed.

Other safety actions include increasing the frequency of engineering inspections and resurfacing of duct heater elements, as well as streamlining procedures to ensure new information from the aircraft manufacturer is distributed to all crews without delay.

The recently rated agricultural pilot was conducting circuit and landing practice in a Cessna Ag wagon. Whilst landing on runway 04 the right brake lost effectiveness, resulting in the pilot losing directional control. The aircraft ground looped to the right, damaging the right landing gear and right wing, before coming to rest adjacent to the runway. The weather was CAVOK with light winds.

FACTUAL INFORMATION

A Reims F406 aircraft was conducting a coast watch surveillance flight from Darwin to Gove. The aircraft was approximately 100 NM north-west of Gove operating below 2,000 ft in non-controlled airspace. The crew heard a broadcast on the marine very high frequency - frequency modulation (VHF-FM) radio. This broadcast was on the nominated emergency frequency of 156.8 MHz (which was Channel 16 on marine transceivers) and was from a foreign warship notifying an intention to commence live weapon firing in the area from the surface to 45,000 ft.

The F406 crew attempted to contact the warship crew on the marine frequency to advise them of the location of their aircraft, which was within the firing area. Reception with the foreign warship was unreadable, but eventually the F406 crew were able to talk to the radio operator of a Royal Australian Navy (RAN) warship that was also in the area. The F406 crew sought advice of a safe distance to avoid the ships. The radio operator in the RAN warship advised the crew of the ship's position and to remain outside a circle of 20 NM radius centred on that position. The intended track of the F406 was within 10 NM of the ships.

The F406 crew altered the aircraft's course to avoid the firing area. At the same time the crew reported their position to Adelaide Flight Service (FS) and also advised that they were aware of the live firing being conducted in the area. The FS operator checked the aeronautical information service for a notice to airmen (NOTAM) regarding the firing activity. There was no NOTAM for the activity. The operator requested the crew of the F406 to relay to the foreign warship to contact FS on high frequency (HF) radio. The foreign warship contacted FS and it was established that the firing was an unnotified live firing on the high seas. Shortly after, the operator on the warship advised FS on HF and broadcast on the marine VHF frequency, that firing was completed. The firing had been conducted for approximately 40 minutes.

The F406 crew were monitoring six radios which were selected to both aviation and marine, emergency and normal operating frequencies (HF, VHF and ultra high frequency (UHF)). The only notification of the firing activity that they heard was on the marine VHF emergency channel. No broadcasts were heard on the aviation emergency or the normal operational very high frequency - amplitude modulation (VHF-AM) or HF frequencies.

Royal Australian Navy procedures

Royal Australian Navy personnel had briefed the crew of the foreign warship on the RAN procedures for weapons firings.

Warships planning to conduct firing operations were required to notify the appropriate local area command 4 days prior to the event. This was to enable the command to arrange for the issue of a NOTAM notifying the location, times and vertical limits of the operation. The investigation was unable to establish the reason for the short notice of the firing operation in this instance.

Operations could be conducted without this notification subject to a number of criteria being satisfied. In this situation the warship intending to conduct the firing had to either visually or electronically scan the area to be used. This scanning was to ensure that no vessels or aircraft were within the firing area. The firing was to be conducted no closer than 10 NM to controlled airspace or a published air route, and outside of 20 NM from land and 10 NM from offshore man-made structures.

The warship was also required to broadcast intentions prior to commencement and on completion of each operation on the marine VHF-FM and HF (2828 KHz) emergency frequencies. The actual frequencies on which the foreign warship did broadcast were unable to be confirmed by the investigation. There was no requirement to broadcast on aviation emergency or normal operational frequencies.

Civil aircraft radio requirements

Civil aircraft are only required to fit and use VHF-AM and/or HF radios. Civil aircraft would normally not be capable of receiving or transmitting on the VHF-FM band. Because of the nature of the coast watch task undertaken by the F406, that aircraft was fitted with and used a number of non-aviation radios.

Air routes

Part of air route B472 between Gove and Ambon, Indonesia and air route J199, the published route between Kupang, Indonesia and Port Moresby, Papua New Guinea, traversed the area in which the firing was conducted.

Adelaide FS advised Brisbane air traffic control (ATC) of the firing once they were aware of the details. Subsequently, Brisbane ATC routed aircraft clear of the area until the activity was completed.

Airspace

Non-controlled airspace extended from sea level to flight level (FL) 245 in the firing area. FS was responsible for communications in non-controlled airspace. Above FL245 and up to FL450 the airspace was controlled airspace. Brisbane ATC was responsible for the management of this airspace.

ANALYSIS

The crew of the foreign warship would appear to have not conducted an adequate scan of the firing area prior to the firing operation and was unaware of the proximity of the coast watch aircraft.

The warship's notification of firing was not broadcast on the aviation, emergency or normal operating frequencies. Consequently, civil aircraft, which may have been in the area, would not have received warning of the firing operation unless they had special equipment to monitor the marine VHF-FM emergency frequency.

Air traffic services (ATS) used VHF-AM and HF radios to communicate with crews of civil aircraft. The foreign warship crew could have used either of these radios to establish communications with ATS to notify the firing activity. FS would have then been able to notify aircraft in non-controlled airspace by broadcasting a hazard alert to known aircraft in the area. Also, crews of aircraft unknown to FS in the area, would probably have also been alerted if they were monitoring the appropriate area frequency. In controlled airspace ATC would have had prior notice to enable them to route aircraft clear of the firing area.

SIGNIFICANT FACTORS

1. A NOTAM was not issued for the firing activity.

2. The foreign warship conducted a firing activity at short notice.

3. The foreign warship crew was unaware of the approaching coast watch aircraft.

4. Royal Australian Navy procedures did not require broadcasts to be made on either the aviation emergency or normal operational frequencies

5. Brisbane air traffic control was unaware of the firing activity that impinged on controlled airspace.

6. The firing activity was conducted in close proximity to two air routes in controlled airspace.

SAFETY ACTION

Following investigation of aspects of the notification of weapons training areas the Bureau of Air Safety Investigation issued the following recommendations to the Australian Defence Force and to Airservices Australia on 29 June 1998:

"IR980086

The Bureau of Air Safety Investigation recommends that the Australian Defence Force, with the assistance of Airservices Australia, review airspace activation procedures to ensure that appropriate and fail-safe notification actions have been completed prior to the commencement of military activities.

IR980087

The Bureau of Air Safety Investigation recommends that Airservices Australia assist the Australian Defence Force to review airspace activation procedures to ensure that appropriate and fail-safe notification actions have been completed prior to the commencement of military activities."

The pilot reported his aircraft had been hangared since April 1994. He had become seriously ill in November 1994 but, in the past 12 months, he had made a good recovery. He decided that he would go with a friend, who was a qualified pilot, to pick up his aircraft from Roma where it was hangared. The aircraft was flown from Roma to his friend's property.

On the afternoon of the accident, the pilot was using a 220 m runway for take-off, but this distance was well within the performance capability of the aircraft. The friction nut for the throttle control in the aircraft was not functioning correctly, and it was possible for the throttle control to slide back from the full throttle position by about 2.5cm, if the throttle was not held in position manually.

The pilot stated that after becoming airborne in about 70 m he went to apply full flap after building up some speed to clear some low trees at the end of the runway. He said that as he let go of the throttle to apply extra flap, the throttle control slid back with a corresponding reduction in engine power. By the time he realised what had happened and reapplied full throttle, it was evident that the aircraft would not clear trees at the end of the strip. He said he was then left with no alternative but to pull the nose up and 'belly' the aircraft through the trees.

The Boeing 737 departed Darwin for Adelaide at 1250 CST. During cruise at FL330, with Autopilot A selected in Command mode, the crew reported that the aircraft exhibited Dutch roll tendencies and that an Air Data Computer malfunction had occurred. Autopilot A was disengaged, and the aircraft was flown manually for approximately 2 minutes before Autopilot B was selected in Command Mode. Sometime later Autopilot A was re-engaged with no problems observed. (Each Autopilot receives inputs from a different Air Data Computer.)

Engineering staff removed the digital flight data recorder from the aircraft in Adelaide and sent a copy tape to the Bureau for analysis. The data showed that at 1357 CST five cycles of left then right roll of increasing amplitude occurred. The total duration of the five cycles was 40 seconds. Approximately 1 1/2 minutes later a similar disturbance occurred followed quickly by a third disturbance of lesser magnitude. The maximum roll experienced during these disturbances was 3 degrees. Only very small heading changes were associated with this roll. Autopilot A provided aileron control inputs to counteract the roll.

Occasional large fluctuations in recorded Air Data Computer parameters (Pressure Altitude, Computed Airspeed, Mach Number, Total Air Temperature and Static Air Temperature) were observed in the recorded data commencing at 1401 CST. These indications are consistent with a malfunctioning Air Data Computer.

Neither Rudder Position nor Yaw Damper status are directly recorded so the cause of the Dutch roll could not be positively determined however the yaw damper does have airspeed as an input. While rudder pedal position is recorded there is no mechanical feedback from the yaw damper to move the rudder pedals.

At 1600 CST shortly before top-of-descent Autopilot A was re-engaged in Command mode. No anomalies were observed in the recorded data at this time.

The Air Data Computer suffering the malfunction was removed from the aircraft in Adelaide and quarantined for fault-finding. During subsequent examination, a fault was found with this unit. The aircraft's in-service performance has been closely monitored by the operator since this occurrence and no anomalous behaviour has been reported to date (2 July 1997).

There was no safety action as a result of this occurrence.

FACTUAL INFORMATION

The crew of an instrument flight rules (IFR) Boeing 737 (B737) had been cleared to descend to flight level (FL) 120 while tracking inbound to Darwin. The aircraft was tracking via the 163 radial of the very high frequency omni-directional radio range (VOR) navigation aid. The crew had transferred from Brisbane En Route Control to Darwin Approach Radar Control at approximately 60 NM from Darwin. An IFR Airbus Industrie A320 had departed Darwin for Alice Springs and was tracking via the 163 VOR radial on climb to FL370. The approach radar controller was busy with aircraft operating predominantly within 20 NM of Darwin aerodrome and did not recognise the potential for conflict between the aircraft.

When the B737 was approximately 45 NM from Darwin, the approach radar controller was alerted to the conflict with the A320 by another controller. The approach radar controller instructed the crews of both aircraft, in separate transmissions, to turn right 30 degrees. In the first transmission, the B737 crew did not understand the instruction and the approach radar controller had to repeat the instruction. Subsequently, the aircraft passed with less than 1,000 ft vertical separation and 2.4 NM lateral separation. The required minimum separation standard was either 3 NM laterally or 1,000 ft vertically. There was a breakdown of separation. B737 crew The B737 was fitted with a traffic alert and collision avoidance system (TCAS) and consequently, the crew was aware of the approaching A320. They were prepared to comply with any avoiding instructions expected from the approach radar controller. However, the initial instruction was given very quickly and was difficult to understand, and the crew had to ask the controller to repeat the transmission. Radar display The approach radar controller was responsible for managing the airspace encompassed by a circle with a radius of 50 NM, centred on Darwin. The approach controller used a main radar display with the range set to 50 NM and a smaller secondary display window with the range set to 100 NM.

The main display was primarily used to separate and manage aircraft close to Darwin while the smaller window was used for the management of aircraft that were further than 50 NM from Darwin. To avoid clutter on the display, the radar display labels on the secondary window were suppressed for all aircraft within 30 NM of Darwin. Just prior to the occurrence, the A320 would have been displayed on the main display with a full label and on the secondary window without a label. The B737 would have been displayed on the secondary window with a full label. Approach radar controller The approach controller was busy with aircraft conducting practice instrument approaches at Darwin and other arriving and departing aircraft. The controller was using flight progress strips and radar to manage aircraft in the area of responsibility. The Brisbane sector 11 controller advised the approach radar controller of the radar identification of the B737 and the assigned level of FL120 when the aircraft was 68 NM from Darwin. The approach controller reported the departure of the A320 to the sector 11 controller during the same coordination exchange. Neither controller remarked on a need to coordinate a separation procedure for the aircraft. Coordination The approach coordinator advised the sector 11 controller when the A320 taxied for departure.

The approach coordination controller and the sector 11 controller did not recognise the potential for conflict between the B737 and the A320. There were no specific procedures to ensure that Brisbane and Darwin controllers coordinated separation procedures for aircraft that were likely to pass near the boundary of the respective areas of responsibility. Generally, the approach coordinator would have alerted the approach radar controller to the potential for conflict before a taxiing aircraft departed. The investigation did not ascertain why the approach coordination controller did not advise the approach radar controller of the situation.

ANALYSIS

It is probable that due to the busy nature of the traffic environment at the time, both the approach coordinator controller and the approach radar controller became so fixated on the coordination and separation of aircraft close to Darwin that they were unable to comprehend the developing conflict. This aspect is further supported by the fact that the approach radar controller did not appreciate the potential conflict when the B737 was handed off from Brisbane and the departure time for the A320 was advised to Brisbane, in the same coordination exchange. The display of the radar tracks and labels for the aircraft on separate windows would have constrained the approach radar controller's ability to develop an appreciation of the potential conflict. However, flight progress strips were available and if they were scanned regularly by the approach radar controller, they should have provided a prompt for that controller. The aircraft passed at a position approximately 35 NM south-south-east of Darwin, which was well within Darwin's area of responsibility. Consequently, while the coordination of a procedure for the separation of the aircraft between Brisbane and Darwin Air Traffic Control would have assisted in the occurrence, it was not considered to be a significant factor.

SIGNIFICANT FACTORS

1. The approach coordination controller did not warn the approach radar controller of the potential conflict between the aircraft.

2. The approach radar controller's attention was diverted from scanning the flight progress strips for potential conflicts.

SAFETY ACTION

Local action

Darwin and Brisbane Air Traffic Control have developed a Manual of Air Traffic Services Supplementary Procedure which requires controllers to coordinate a separation procedure for aircraft that may conflict in the area between 40 NM and 80 NM from Darwin.

Additionally, Darwin Air Traffic Control has established an arrival position to reduce approach radar controller workload.

FACTUAL INFORMATION

The Australian-registered Boeing 767 (B767) had departed Brisbane on a flight to Christchurch, New Zealand and had passed the reporting point STUDA (30.18S 155.17E) at 1006 EST. The crew were maintaining the aircraft at flight level (FL) 370 on air route L503.

The foreign-registered B767 had departed Nadi, Fiji on a flight to Sydney and had passed Lord Howe Island at 1009 EST. The crew were maintaining the aircraft at FL390 on air route B450, which crossed the track of the Australian B767 approximately 90 NM south-west of Lord Howe Island.

Both aircraft were under the jurisdiction of Brisbane Air Traffic Control Sector 8. After receiving the foreign registered B767's position report at Lord Howe Island, the controller was required to forward the time estimate for the aircraft's next position report to Sydney approach control. This communication was to be carried out via a link that requires the receiving controller to depress a switch, indicated by a flashing light on the control console, in order for conversation to commence. On this occasion the Sydney controller was busy and it took almost seven minutes for the link to be opened.

There were five aircraft en-route from Australian airports to New Zealand that were under the control of the Brisbane controller. The flight times and disposition of these flights were such that several traffic separation conflictions would arise in Auckland's sector of responsibility unless corrective action was taken. At 1015, the Brisbane controller coordinated various matters with the Auckland controller and proposed some solutions which he could initiate to assist the Auckland controller in preventing the traffic conflicts. In particular, he suggested that the Australian B767 should climb to the non-standard FL390 and another aircraft climb to the non-standard FL350. The Auckland controller stated that he would consider these, and other, options before further coordination with the Brisbane controller.

At 1017, the crew of the Australian B767 requested a climb to FL390 due to turbulence. Although this was a non-standard level for eastbound flights, it did offer an immediate solution to four of the five prospective traffic conflictions. The Brisbane controller decided to approve the climb after discussing the issue briefly with the Auckland controller during a second coordination exchange, but without carrying out appropriate checks for lateral separation between the aircraft where air routes B450 and L503 intersect.

Just prior to approving the Australian B767's climb, the Sydney approach controller responded to the Brisbane controller's earlier attempt to coordinate details of the foreign B767. By this time the Brisbane controller could not recall the reason for initiating the call and the two controllers terminated the conversation without completing the intended task. The climb was approved at 1018.

At 1021, the Brisbane controller realised that he had not carried out all the required checks to ensure separation before approving the change in flight level. Although he did not make a specific scan of his flight progress strips, he was uneasy about the decision and elected to cancel the climb approval. He issued instructions for the crew of the Australian B767 to descend back to FL370.

The crew were passing FL385 on climb when they saw the other B767 some 25NM ahead and passing from left to right. They did not receive a warning from the traffic alert and collision avoidance system (TCAS).

The analysis of the radar recording was unable to establish the separation distance as both aircraft were outside radar coverage at the crossing point. Extrapolation of the last observed radar data for the Australian aircraft and the first observed radar data for the foreign aircraft, indicates that they passed with at least 25NM horizontal separation. However, as no procedural standard had been achieved, there was a breakdown of separation standards.

The Brisbane controller was working Sector 8 Radar and Sector 8 Ocean at the time of the occurrence. Together these two sectors form the Oceanic group. Workload was light to moderate during the period surrounding this occurrence. A team leader was rostered on duty but was conducting tasks outside the immediate work area at the time of the occurrence. Another controller was also rostered for duty on the Ocean group but was taking a rest break.

ANALYSIS

The Brisbane Sector 8 controller approved a change in flight level for the Australian B767 without carrying out the required checks for possible conflicts. This action may have been a result of the controller focussing his attention on resolving potential conflicts for the Auckland controller.

More specifically, the controller's omission occurred at a highly automised, skill-based level of performance. Following the Australian B767 crew's request to go to the non-standard FL390, a controller should normally undertake the following actions:

1. Check the flight strips and ensure there was no potential conflictions.

2. Communicate with relevant controllers to obtain approval for a non-standard flight level.

3. Approve or disapprove the change in flight level.

In this particular case, the controller had already discussed the possibility of a higher level for the B767 with the Auckland controller shortly before the flight crew requested the level change. When the Brisbane controller received the request, there would have been a strong, natural tendency to immediately provide this information to Auckland and therefore continue the previous conversation. Indeed, by continuing the conversation in a subsequent coordination call to Auckland just 2 minutes after the first contact, the controller had commenced his response to the B767 flight crew's request for a change in flight level. However, he had started the task at the second step in the sequence and omitted checking the other flight strips for potential conflictions.

As the request for the level change was due to turbulence, the controller may have felt some urgency to respond to the request. In addition, the Brisbane controller's awareness of the foreign B767 may have been reduced in the period prior to the occurrence. Firstly, the Sydney controller did not initially respond to the Brisbane controller's attempt to pass coordination for that aircraft. Secondly, the controller's attention was focussed on the other aircraft under his control during his discussions with the Auckland controller.

Skill-based errors, such as the one involved in this occurrence occur when a person performs a familiar task in an environment which is slightly different to that which is normal. In this case the difference was that the Brisbane and Auckland controllers were discussing a level change shortly before it was requested by the Australian B767 crew.

Skill based tasks are usually performed semi-automatically, with a low awareness of the actions taken by the individual. Reducing the incidence of such errors can be difficult due to this low level of awareness at which skill-based behaviour is performed. Airservices Australia will soon be introducing The Advanced Australian Air Traffic System (TAAATS). This new system will involve significant changes to the way many air traffic control tasks are performed. For procedural enroute controllers, a notable change will be the introduction of a display which gives a plan position indication for aircraft in their area of responsibility. This change will probably make the detection of potential traffic conflicts easier.

The likelihood of detecting and correcting such skill based errors before their consequences are significant could be improved through the introduction of an automatic conflict alert. Such an alert is currently being considered for radar tracks in TAAATS but not for other tracks. An alternative defence is the presence of other controllers to supervise or support a controller. Although there was provision for other controllers to be on duty, this defence was not present at the time of this occurrence.

SIGNIFICANT FACTORS

1. The Brisbane Sector 8 controller approved a change in flight level without carrying out the required separation calculations.

2. The Brisbane Sector 8 controller was attempting to solve future separation conflicts and may have been distracted from the immediate task.

SAFETY ACTION

As a result of this and several other occurrences, the Bureau of Air Safety Investigation (BASI) examined issues associated with Airservices Australia's operation of teams in air traffic services. This examination resulted in BASI issuing the following Safety Advisory Notice to Airservices Australia on 27 January 1998:

Safety Advisory Notice SAN 970137

Airservices Australia should take note of the safety deficiencies detailed by this document and take appropriate action.

The safety deficiencies referred to in the document were:

1. an undesirably low level of operational support provided by experienced controllers, including team leaders and other full performance controllers, to controllers working in operational positions.

2. an inappropriately low level of emphasis on team development activities, such as the provision of team leader training and support, and the regular provision of team days with structured learning content for team members.

3. performance evaluation systems for team leaders that do not ensure relevant team leader performance areas are measured, and also do not ensure that any degradation in a team leader's proficiency on operational positions will be detected.

4. an inappropriately low level of training and development for many controllers on human factors issues, particularly those associated with inter-controller coordination and communication.