1. The crew did not detect the FMC data entry errors.
2. The crew did not apply effective crosschecking procedures following FMC modification.
3. The PNF was not required to provide any altitude or distance support calls during the approach until after the FAF.

When the crew of the 737 inadvertently omitted waypoint/altitude constraint data from the flight management computer (FMC) LEGS page, they received no warning from the FMC. The only way to safeguard against this type of erroneous data manipulation is the application of standard operating procedures (SOPs) for FMC data entry and cross-checking.

Additional defences could be achieved by implementing SOPs for the monitoring of flight profile throughout critical stages of flight. The operator had published instructions requiring crew members to provide support calls from the Final Approach Fix (FAF). As a consequence, no calls were initiated by the pilot not flying (PNF) until well after the aircraft had descended below 6,500 ft.

The benefits of adhering to a proven set of SOPs are well documented. The procedures must be appropriate, well understood, and complied with.

The crew reported that when they made preparations for descent the Cairns Automatic Terminal Information Service (ATIS) nominated runway 15 for landing. The crew selected the appropriate approach and landing charts and programmed the flight management computer (FMC) for an arrival to runway 15.

Three minutes before the crew commenced descent, the ATIS was changed to indicate that arriving aircraft from the south could expect to carry out the runway 33 Locator/Distance Measuring Equipment (LOC/DME) approach to runway 33. The crew was not aware of the change to the landing runway until the air traffic controller assigned the crew a HENDO-Three standard instrument arrival (STAR) which included a NORMA transition. In order to comply with the STAR, the 737 was required to proceed via waypoint NORMA and then track 031 degrees to waypoint¹ HENDO, which, as the last waypoint of the STAR, also became the initial approach fix (IAF) for the runway 33 LOC/DME approach (Refer figure 1 & 2).

The crew selected the STAR and the runway 33 LOC/DME approach from the FMC navigation database and incorporated the required tracking and altitude requirements into the active FMC-LEGS page. During that interaction the crew did not select waypoint HENDO as the IAF when prompted by the FMC to do so and consequently critical 'Not below 6,500 ft' altitude constraints at the HENDO and 20 DME Cairns waypoints were omitted. The STAR and runway 33 LOC/DME approach became part of the FMC active flight plan and the FMC provided the crew with lateral and vertical navigation guidance.

While on descent, the crew was instructed by air traffic controller to reduce speed and fly a heading that would take the aircraft away from the published STAR track. The crew subsequently reported that this was done to ensure separation with preceding traffic. Once the required separation had been achieved, the crew was instructed to rejoin the localiser approach. They modified the route legs (RTE LEGS) page of the FMC by selecting track direct (TRK DCT) to HENDO. That action removed waypoints from the active route that were no longer required.

At 24 DME, the FMC calculated descent point, the aircraft left 6,500 ft. That occurred because the waypoints and their associated altitude constraints, which ensured that the aircraft would not descend below 6,500 ft until passing 20 DME had been omitted. The autopilot was engaged with vertical navigation (VNAV) mode active.

Analysis of information recovered from the flight data recorder showed that the 737 passed 20 DME Cairns at an altitude of 5,860 ft. The crew reported to air traffic control (ATC) that they were in cloud.

Figure 1: Cairns LOC DME Rwy 33

Reprinted with permission of Jeppesen Sanderson Inc.

Figure 2: Cairns - HENDO THREE ARRIVAL

Reprinted with permission of Jeppesen Sanderson Inc.

Cairns runway 33 LOC/DME approach

To ensure appropriate terrain clearance, flight crews conducting the Cairns runway 33 LOC/DME approach must maintain track within appropriate tolerances and not descend below the minimum altitude specified on the instrument approach chart until passing the next step-down point. Once the next step-down point has been passed, the aircraft is permitted to descend to the next lower minimum altitude. That allows crews to progressively descend, remaining safely clear of terrain as the aircraft approaches the airport for landing.

In order to accommodate aircraft arrivals from various directions, the Cairns runway 33 LOC/DME approach has a number of IAFs. Depending on the STAR issued by ATC, the crew should operate the aircraft to track via COBUN, HENDO, or ZORBA. The IAF forms an integral part of the runway 33 LOC/DME approach.

Flight Management Computer

The FMC system fitted to the 737 provided lateral and vertical flight path guidance as well as performance information to the crew. The FMC can also provide control and guidance information to the autopilot.

The 737 autopilot and flight director system has a number of descent modes. The crew reported that they conducted the Cairns runway 33 LOC/DME approach using VNAV² path.

Before the crew could utilise the FMC to provide vertical navigation guidance, the FMC needed to compute a descent path, which conformed to the requirements of the instrument approach. Waypoints and associated altitude constraints required by the FMC to compute an accurate approach profile that corresponded to the LOC/DME approach path gradient had been inadvertently omitted by the crew.

The FMC database contained the Cairns runway 33 LOC/DME approach, which the crew selected. They were then prompted to select from one of three transitions; COBUN, HENDO, or ZORBA. During the occurrence flight, the FMC operated as designed.

Communication of safety information

The Operator's Operations Manual included detailed instructions to ensure that crews selected the HENDO transition when activating the runway 33 LOC/DME approach into the FMS.

The operator experienced similar events on three occasions during December 2003. In response to those occurrences the operator produced an article for its Safety Shorts Operational newsletter, which warned crews to follow published procedures when conducting a Cairns runway 33 LOC/DME approach.

The operator's manuals contained the following instruction:

The 737 Flight Crew Training Manual (FCTM)
Set all mandatory altitude restrictions and at or above constraints in the MCP altitude window. The next altitude may be set when the restriction has been assured, and further clearance has been received.

The Flight Crew Operations Manual (FCOM) (Part B) Volume B1 Approaching intercept heading, select flaps 5 and select LNAV or other appropriate roll mode. Approaching the FAF, select gear down and flap 15. Set the [altitude] minima in the MCP altitude window.

Crew communication during instrument approaches

The FCOM issued to crew members referred to the crewmembers as either the pilot flying (PF) or pilot not flying (PNF) and contained the following instruction regarding support calls:

Operations Manual (Part B) 2.11.3.9
On a non-ILS instrument approach, including DME Arrival, the PF shall brief the descent profile to be flown from the Final Approach Fix (FAF). After passing the FAF, the PNF shall call the profile (distance/altitude) at each published or briefed distance/altitude. The PF shall acknowledge the call and initiate any profile correction. The PNF shall then call the next profile distance/altitude.

Note: In the case of the Cairns 33 LOC/DME approach the FAF is at 8 DME.

The PNF was required to provide support calls (of distance/altitude) from the FAF. No calls were required to be made between the IAF, and the FAF. In the case of the Cairns runway 33 LOC/DME approach, the aircraft may be as low as 1,600 ft (the minimum altitude at 8 DME) before any calls were required to be made. The aircraft was between the IAF and the FAF when it descended below 6,500 ft.

1. Predetermined and accurately known geographical position forming start or end of route segment.
2. Vertical navigation mode

On 27 May 2004, a Boeing 737-705 (737) aircraft, registered VH-VBX, was being operated on a scheduled passenger service from Melbourne, Vic, to Cairns, Qld, in instrument meteorological conditions.

Throughout this analysis it should be noted that the pilots of both aircraft and the ATS controller involved in the occurrence complied with the rules and procedures for operation in Class E airspace associated with the NAS phase 2b, implemented on 27 November 2003.

The regulations, procedures and educational material associated with that implementation stated that there was a shared responsibility by pilots of IFR and VFR flights to see-and-avoid each other in Class E airspace.

Prior to the implementation of NAS phase 2b on 27 November 2003, both aircraft involved in this occurrence would have been operating in Class C airspace. As such, they would have been subject to an ATS airways clearance and would have been provided with separation in accordance with Class C airspace rules and procedures. In order for two aircraft to pass in close proximity at these flight levels in Class C airspace, those rules and procedures would need to have been compromised. As the pilots of both aircraft and the ATS controller complied with the rules and procedures for Class E airspace under NAS, those rules and procedures do not preclude an IFR high performance, high-capacity regular public transport aircraft from passing within such close proximity as to generate a TCAS RA on either known or unknown VFR traffic.

The controller's relatively low workload and other factors, such as the Lancair pilot submitting flight notification details, and broadcasting his departure from Maroochydore, assisted the ATS controller to detect a possible conflict. Although there was no requirement for the controller to pass traffic information to the pilot of the Lancair about the location of the 737 under NAS Class E airspace procedures, the controller provided traffic information to both aircraft with respect to each other.

Part 5 of MATS also stated that controllers shall issue a safety alert when, in the consideration of the controller, such an advice was warranted to avoid conflict. In the circumstances of this occurrence, the controller had provided traffic information to the crews of both aircraft, and the Lancair pilot had broadcast that he had the 737 in sight. Accordingly, the onus was then on the Lancair pilot to avoid the 737. In those circumstances, the provision of a safety alert, which may have included a suggested course of action, may also have complicated the situation, if that suggestion was contrary to what the pilots of each aircraft considered necessary.

Provision of a safety alert, in the circumstances of this occurrence, was not required. However, MATS did not provide any guidance to controllers on the circumstances under which the provision of a safety alert would be appropriate. Publication of those guidelines may assist controllers to determine when a safety alert should be issued.

In Class E airspace, the provisions of CAR 163A required the crews of both aircraft to 'see and avoid' each other. The 737 crew were unable to see the Lancair despite their attempts to do so. The Lancair pilot reported that he had the 737 in sight. When the 737 crew observed the position of the Lancair on the TCAS navigation display, they commenced action to avoid a confliction prior to receipt of both the TCAS TA and RA. In concert with that action, they continued in their attempts to visually acquire the Lancair, in accordance with Class E airspace see-and-avoid requirements.

In the circumstances of this occurrence:

• the ATS controller took more actions than those required by the published requirements for Class E airspace and MATS;
• the Lancair pilot took more actions than those required by the published procedures for Class E airspace under NAS;
• the early action taken by the 737 crew to avoid the conflict was not contrary to the published procedures for Class E airspace under NAS.

Despite those actions, the two aircraft came into such proximity that a TCAS RA was generated in the 737.

Both aircraft were operating in Class E airspace that was introduced as part of the National Airspace System (NAS) phase 2b on 27 November 2003. (An ATSB research report titled National Airspace System Stage 2b: Analysis of Available Data was released in July 2004.) As no prescribed separation standards are applicable in these circumstances, there was no infringement of separation standards. However, ATS audio tapes, radar and TCAS data, and information obtained from the air traffic controller and the pilots of both aircraft were consistent and indicate that the aircraft came into such close proximity that a threat to the safety of the aircraft may have existed. Therefore, the incident has been classified by the ATSB as an airprox event.

The incident at Canty IFR reporting point, on 3 December 2003 (ATSB report 200304963) was also classified as an airprox event.

Unlike this incident north of Brisbane on 7 April 2004 and the Canty incident, the occurrence near Launceston on 24 December 2003 (ATSB report 200305235) was classified as a serious incident due to the lack of radar coverage in the Launceston area and the absence of radio broadcasts from the pilot of the Tobago, which created an unalerted see-and-avoid environment for the crew of the Boeing 737. The air traffic controller also was unaware of the Tobago.

On 7 April 2004, a Boeing 737-7BX (737) aircraft registered VH-VBT, operating under the instrument flight rules (IFR), was en route from Townsville and descending for a landing at Brisbane. A Neico Lancair IV-P aircraft registered VH-LDJ, operating under the visual flight rules (VFR), was en route from Maroochydore to St George, on climb to flight level (FL) 165¹. Both aircraft were operating in radar Class E airspace at the time of the occurrence.

The Lancair pilot reported to the Air Traffic Services (ATS) controller that he had departed Maroochydore at 0718 Eastern Standard Time², although in accordance with the National Airspace System (NAS) procedures, there was no requirement for him to do so³. The controller issued the Lancair pilot with a discrete transponder code to assist with his situational awareness. Published NAS procedures stated that:

In Class E airspace, IFR and VFR flights are permitted. IFR flights are provided with an air traffic control service, are separated from other IFR flights, and receive traffic information on VFR flights as far as is practicable. VFR flights receive a Radar Information Service (RIS)⁴ on request.⁵

Those procedures also stated that, for VFR aircraft operating in Class E airspace, '…no flight notification was required…' and pilots of aircraft conducting operations under the VFR were required to monitor the appropriate radio frequency. The Lancair pilot submitted flight notification details to ATS prior to departure from Maroochydore, and reported that departure to ATS.

In accordance with those procedures, the controller was not providing traffic separation⁶ to either aircraft, and advised the 737 crew accordingly. Recorded ATS audio information showed that the controller provided traffic information about the Lancair to the 737 crew on three occasions between 0721:58 and 0725:08. At 0722:17, and again at 0725:38, the controller provided the Lancair pilot with traffic information about the 737, although there was no requirement for him to do so in Class E airspace. During the first transmission at 0722:17, the controller advised the Lancair pilot that the 737 was crossing traffic which would pass in about 15 NM and was on descent into Brisbane. During the second transmission at 0725:38, the controller advised the Lancair pilot that the 737 was 'in your 2 o'clock at 8 miles now'. ATS radar data showed that, at 0725:38, the 737 was 8.8 NM from the Lancair. At 0725:47, in response to that broadcast, the Lancair pilot advised the controller that he had the 737 in sight.

As the 737 was passing through about FL157 at 0726:01, the crew reported that they observed a traffic alert and collision avoidance system (TCAS)⁷ traffic symbol on the aircraft's navigation display, about the Lancair. They attempted to visually acquire the Lancair, but were unable to see that aircraft. Recorded data from the 737's flight data recorder (FDR) showed that, at that point, the crew commenced manoeuvring the aircraft by reducing the rate of descent. At about 0726:18, the crew received a TCAS traffic advisory (TA)⁸. The crew reported that they were still unable to visually acquire the Lancair and were uncertain of its relative position. Recorded FDR data indicated that at 0726:34, the crew disengaged the autopilot and commenced a right turn away from the Lancair. They subsequently levelled the 737 at FL153 and then climbed to FL154. At 0726:40 and at 15,420 ft the 737 received a TCAS resolution advisory (RA)⁹ aural warning instructing them to climb, in response to the proximity of the Lancair. They subsequently climbed the 737 to FL166 and continued the turn to about 15 degrees right of track. The duration of the RA was 10 seconds and commenced when the Lancair was about 650 ft vertically lower and about 0.7 NM to the left of the 737.¹⁰ Recorded ATS radar data showed that, at about 0726:45, the Lancair altered track 8 degrees to the right away from the 737, just before passing behind and below the 737. The Lancair pilot reported that he did not experience the effect of wake turbulence from the 737. The minimum distance between the two aircraft was about 600 ft vertically at about 0.3 NM laterally.¹¹

The Class E airspace in which the 737 and the Lancair were operating at the time of the occurrence was introduced as part of NAS phase 2b on 27 November 2003. Prior to NAS phase 2b, that volume of airspace was classified as Class C airspace. In Class C airspace, both aircraft would have been subject to an ATS airways clearance and would have been separated in accordance with prescribed standards.

In Class E airspace, the pilots of aircraft operating under the IFR and VFR were required to:

…maintain vigilance so as to see, and avoid other aircraft.¹²

The NAS Implementation Group reference guide, distributed as part of the NAS phase 2b implementation, stated that:

The other important change is that the pilot of a VFR flight should not make broadcasts on ATC frequencies.¹³

It also stated that:

Pilots of VFR flights may monitor the ATC frequency to enhance situational awareness.

Please do not make broadcast transmissions or engage in chatter on an ATC frequency. The safety of others depends on you not doing this.

Pilots are not precluded from responding to any ATC or pilot transmission when they believe their safety is at risk from another aircraft.

Part 2, Section 2, paragraph 2.2.4.1 of the Manual of Air Traffic Services (MATS) stated that:

Before providing a radar service to an aircraft, radar identification shall be established.

Although the controller did not advise the Lancair pilot that the Lancair was radar identified, the controller issued a discrete transponder code to the Lancair pilot and radar-identified the aircraft before providing a Radar Information Service to the pilot.

Part 4, Section 1, paragraph 4.1.1.1 of MATS contained information regarding ATS controller responsibilities for providing aircraft separation. The manual stated that:

Separation shall be provided by ATC using approved separation standards and procedures.

In the circumstances of this occurrence, the ATS controller was not required to provide separation to either aircraft in respect of the other and there were no separation standards applicable in these circumstances in Class E airspace.

Part 4, Section 1, paragraph 4.1.1.3 of MATS stated that:

Nothing in this chapter precludes a controller from using discretion and initiative in any particular circumstance where these procedures appear to be in conflict with the requirement to promote the safe conduct of flight.

In the circumstances of this occurrence, the ATS controller provided traffic information to the crews of both aircraft in respect of the other, although under NAS procedures there was no requirement for him to provide traffic information to the pilot of the Lancair on the location of the 737.

Part 5, Section 1, paragraph 5.1.13 of MATS provided information regarding provision of safety alerts. Paragraph 5.1.13.1 stated that:

A safety alert shall be issued to an aircraft when a controller is aware the aircraft is in a situation which is considered to place it in unsafe proximity to terrain, obstructions, or other aircraft.

The controller reported that, once the Lancair pilot broadcast that he had the 737 in sight, there was no necessity to broadcast a safety alert to either the Lancair pilot or the 737 crew. The controller also reported that if that pilot had not broadcast that he had the 737 in sight, his next option was to issue a safety alert. MATS did not provide any guidance to controllers on what might be considered '…unsafe proximity…', or when to issue a safety alert.

The NAS Implementation Group reference guide contained information for VFR pilots regarding separation from other aircraft when operating in Class E airspace. Page 16 of that guide stated that:

When you are flying in Class E airspace you are responsible for separation from other aircraft. The onus is on you to look out and see and avoid other aircraft.

Part 12, The Rules of the Air, Division 1, of the Civil Aviation Regulations (CAR) 1988, contained information regarding right of way, prevention of collision, operating near other aircraft and responsibilities of flight crew to see and avoid aircraft. More specifically, CAR 161 contained information regarding right of way and stated that:

(1) An aircraft that is required by the rules in this Division to keep out of the way of another aircraft shall avoid passing over or under the other, or crossing ahead of it, unless passing well clear.¹⁴

(2) The pilot in command of an aircraft that has the right of way must maintain its heading and speed, but nothing in the rules in this Division shall relieve the pilot in command of an aircraft from the responsibility of taking such action as will best avert collision.

CAR 162 (1) contained information regarding prevention of collision and stated that:

When 2 aircraft are on converging headings at approximately the same height, the aircraft that has the other on its right shall give way…

Although not specifically referring to converging aircraft, CAR 162 also stated that '…each shall alter its heading to the right…', and when referring to the aircraft other than the aircraft having right of way, '…shall keep out of the way of the other aircraft by altering its heading to the right…'.

CAR 163 (1) contained information regarding operating near other aircraft and stated that:

The pilot in command of an aircraft must not fly the aircraft so close to another aircraft as to create a collision hazard.

The 737 was on the Lancair's right and, in accordance with CAR 161 and CAR 162 (1), had right of way. The Lancair pilot reported that he had the 737 in sight. While the crew of the 737 had observed a traffic symbol on the TCAS display, they reported that they did not see the Lancair, despite attempts to do so.

Information obtained from the crews of both aircraft, the ATS controller, recorded flight data from the 737, ATS audio recordings and radar data, was consistent and indicated that the crews of both aircraft and the ATS controller complied with the published procedures for Class E airspace under NAS.

Based on the factual data, and the definition contained in Regulation 2.2 of the Transport Safety Investigation Regulations 2003, the incident was classified as an airprox event.¹⁵

¹ 16,500 ft with an altimeter pressure sub-scale setting (QNH) of 1013.2 hPa.
² The 24-hour clock is used in this report to describe the local time of day, Eastern Standard Time (EST), as particular events occurred. Eastern Standard Time was Coordinated Universal Time (UTC) + 10 hours.
³ Aeronautical Information Publication (AIP), ENR 1.1, Sections 3.4 and 18.3.2.
⁴ Radar Information Service (RIS) is defined in Part 10, Section 1 of the Manual of Air Traffic Services as:

An add-on ATC service within radar coverage, which provides information to flights, not otherwise receiving a separation service, in order to improve situational awareness, and assist pilots in avoiding collisions with other aircraft.

⁵ AIP, ENR 1.4, Section 2.1.4 and pages 12-13 and 40 of the NAS Implementation Group Reference Guide - How to Operate in the National Airspace System, effective from 27 Nov 2003.
⁶ Separation is defined in Part 10, Section 1 of the Manual of Air Traffic Services as:

A controlled condition using defined standards to prevent collisions between aircraft.

⁷ The Boeing 737-NG Operations Manual, Volume 2, states that:
 

TCAS alerts the crew to possible conflicting traffic. TCAS interrogates operating transponders in other airplanes, tracks the other airplanes by analysing the transponder replies, and predicts the flight paths and positions. TCAS provides advisory, flight path guidance, and traffic displays of the other airplanes to the flight crew. Neither advisory, guidance, nor traffic display is provided for other airplanes which do not have operating transponders. TCAS operation is independent of ground-based air traffic control.

⁸ A traffic advisory (TA) is generated when the other aircraft is approximately 40 seconds from the point of closest approach, dependent upon aircraft altitude.
⁹ If the other aircraft continues to close, a resolution advisory (RA) is generated when the other aircraft is approximately 25 seconds from the point of closest approach, dependent upon aircraft altitude. The RA provides aural warning and guidance as well as manoeuvre guidance to maintain or increase separation from the traffic.
¹⁰ The recorded ATS radar data and the 737 flight recorder data was consistent. However, the accuracy of that information is dependent upon the tolerances associated with both aircraft altimeters and the ATS radar equipment.
¹¹ The horizontal distance was determined using the radar positions that were recorded every 5 seconds. By interpolating between those points to derive a position every second, the minimum horizontal separation was 0.3 NM, dependent upon the tolerances previously mentioned.
¹² Civil Aviation Regulations 1988, 163A - Responsibility of flight crew to see and avoid aircraft.
¹³ Page 8 of the NAS Implementation Group Reference Guide - How to Operate in the National Airspace System, effective from 27 Nov 2003.
¹⁴ The investigation was unable to locate a CAR definition of '…well clear…'.
¹⁵ An airprox event is defined in Regulation 2.2 of the Transport Safety Investigation Regulations 2003, as:

…an occurrence in which 2 or more aircraft come into such close proximity that a threat to the safety of the aircraft exists or may exist, in airspace where the aircraft are not subject to an air traffic separation standard or where separation is a pilot responsibility.

Airservices Australia

As a result of this occurrence, on 26 November 2004 Airservices advised the ATSB that a review of the AIP and MATS will be conducted. This review will ensure that, when a RVR is quoted in the aerodrome information provided by the controller or detailed in the ATIS, the RVR clearly relates to the runway touchdown zone quoted in the information. Airservices will also incorporate the events of the occurrence into a refresher training module to be used by staff at towers with low visibility operations procedures.

As a result of Airservices? advice of this proposed safety action, the ATSB will continue to monitor its progress until evidence is received of the implementation of the proposed actions.

Bureau of Meteorology

The BoM advised that the occurrence was reviewed at a Fog Workshop on 13 July 2004, which resulted in the establishment of the Forecasting at Sydney Airport? project. The aim of the project is to review the existing forecasting methodologies and guidance material used to predict fog at Sydney aerodrome. The BoM intends to use the results from the review and subsequent real-time testing to implement a structured set of objective guidelines that are intended to improve the accuracy of fog forecasting at Sydney aerodrome.

The project is a joint undertaking by staff from the Bureau?s Sydney Airport Meteorological Unit and the Bureau of Meteorology Research Centre, in collaboration with researchers at Monash and Macquarie Universities. The project team members are currently constructing an expanded climatological database for Sydney aerodrome and reviewing the predictors for fog. The planned completion date for the project is December 2005.

Operator

The aircraft operator advised that staff from its safety department gave a presentation to controllers at the Sydney Terminal Control Unit (TCU) on 15 April 2005. The presentation provided an explanation of the operational aspects of the occurrence to ATC from an operator and flight crew perspective. The operator also advised that the same presentation was given at the Air Traffic Service/Airline Forum held on 20 April 2005. The operator plans to give the presentation to Brisbane and Melbourne TCU controllers later in 2005.

The operator also advised that an article describing the occurrence will be published during 2005 in a newsletter distributed to flight crew.

As a result of this advice of proposed safety actions by the operator, the ATSB will continue to monitor its progress until evidence is received of the implementation of the proposed action.

Related Documents: | Media Release |

The aircraft landed at Sydney in weather conditions that were below the specified landing minima. Although a landing in conditions below the specified meteorological minima was permitted under Civil Aviation Regulation 257(5) in an emergency situation, a landing at Sydney only became a necessity after 0618, the latest divert time to Canberra. This analysis examines the development of the occurrence and discusses the factors that influenced the crews decision to continue the approach to a landing at Sydney.

The aircraft departed Perth with sufficient fuel for the flight, based on the current Sydney aerodrome forecast. There was no requirement to divert to Melbourne at the flight planned Designated Point All Engines Operating (DPA), because the forecast weather conditions were not below the alternate minima for a landing at Sydney. However, the aircraft passed the DPA at about the same time as low-level cloud was clearing at Sydney. The dissipation of the cloud resulted in the subsequent unforecast movement of fog over the aerodrome. At 0540 the crew was aware that a fogbank had formed to the north-west of the aerodrome, but the other weather information available to them at that time did not indicate that fog would be a problem for their arrival at Sydney.

After commencing the descent, the crew used weather information provided by controllers and from the onboard datalink when making decisions in response to the deteriorating visibility at Sydney aerodrome. However, during the latter stage of the flight the crew were not provided with all of the weather information that was available to ATC. This reduced the crews situational awareness of the effects of the rapid progression of fog across the runway complex. While information was available via the onboard datalink, the crew was busy setting up for the approach into Sydney and would have had an expectation that ATC would advise of any significant deterioration in the weather conditions.

Although the crew considered that the Canberra 0530 special automated weather observation indicated marginal weather conditions, the reported conditions were just above the applicable alternate minima for an ILS approach to runway 35. Had the crew obtained or been provided with the Canberra 0600 trend type forecast, which predicted that the weather conditions would not significantly vary for the following three hours, they may have been more confident about the weather trend at Canberra.

The declaration of minimum fuel? and reiteration of the requirement to use 16R in the crews PAN broadcast at 0559 indicated their concern about having enough fuel to land at Sydney. However, the crews advice to ATC five minutes later, that their latest landing time was 0618,?? and then we?d need to go straight to Canberra? suggests that they had enough fuel to land at Sydney but were attempting to expedite their arrival to enable an attempted landing while retaining Canberra as an alternate for as long as possible.

This was supported by the pilot in commands statement that the PAN broadcast was made because there was insufficient fuel to proceed to an operator approved suitable aerodrome, ATC were vectoring the aircraft for a holding pattern and the weather conditions at Sydney were deteriorating at an unknown rate. The pilot in command also stated that, without the existence of the PAN broadcast, the crew could not use Canberra because, in accordance with the operator's aerodrome categorisation, this aerodrome could only be used in an emergency.

The crews decision to require an approach to runway 16R was based on runway visual range (RVR) information that did not reflect the actual conditions at the time. Had the crew been aware that the RVR for runway 34L was greater than the landing minima for that runway, they may not have advised ATC at 0556 that they required runway 16R. Without the delay caused by commencing that approach, the crew would probably have landed on runway 34L before the weather conditions deteriorated below the specified landing minima for that runway.

The report at 0614 from the crew of the Boeing 747, that they were visual during the approach to runway 34L, was subsequently supported by the advice from ATC that the RVR for that runway was 2,000 m, which was greater than the landing minima. This information was received by the crew of QPC one minute before the nominated last divert time and, given the pilot in commands statement that Canberra would be a preferred option only in the event of weather conditions deteriorating to below the landing minima at Sydney, probably influenced them to continue tracking to runway 34L rather than diverting the aircraft to Canberra.

Although the crew could see the runway 34R threshold during the ILS approach to 34L, an orbit to enable a change of runway to 34R would have delayed the landing with the risk of a further deterioration in weather conditions at both thresholds. Given that the crew was committed to a landing at Sydney, the decision by the pilot in command to continue the runway 34L approach reduced the risk of conducting an approach at a later time when the visibility was likely to be worse. Use of the autoland system by a trained crew constituted the safest way to conduct a landing in conditions below the specified meteorological minima.

Although the operator calculated that the amount of fuel remaining on the aircraft after the landing was sufficient to proceed to Canberra after an approach at Sydney, the crews fuel calculations were conservative and were performed in a dynamic high workload situation.

The occurrence highlights the significant safety issues that unforecast weather conditions present to aircraft when these conditions occur during the latter phase of flight. During this phase, the remaining fuel on board may limit the options available to flight crews to minimise the effect of the unforecast conditions on the operation of the flight.

The occurrence also highlights the need for information sharing between flight crews and air traffic controllers in a timely manner so that the situational awareness of crews and controllers is maintained at a high level and crew decision making is optimised in dynamic weather situations.

Information sharing between flight crews and air traffic controllers was also a factor in two other occurrences, involving high-capacity aircraft, which were investigated by the ATSB (ATSB investigation reports BO/200100213 and BO/200304400).

Sequence of events

On 6 April 2004, at about 0625 Eastern Standard Time, an Airbus A330-301 aircraft landed on runway 34L at Sydney aerodrome in weather conditions that were below the applicable landing minima¹. The aircraft, registered VH-QPC, was being operated on an instrument flight rules (IFR) scheduled passenger flight from Perth to Sydney. During the latter stage of the flight unforecast fog developed at Sydney aerodrome, which resulted in the deterioration of visibility to below the landing minima.

The flight had been planned using a valid aerodrome forecast for Sydney, which predicted rain showers and visibility greater than 10 km, with 1 to 2 oktas² of cloud at 1,200 ft and 5 to 7 oktas at 5,000 ft. Those weather conditions were above the Sydney alternate minima³ of 1,479 ft cloud ceiling and 6 km visibility. As there were no operational requirements due to the forecast weather conditions, the flight departed Perth without fuel being specifically carried for a diversion to an alternate aerodrome after an approach at Sydney.

The flight planned ?Designated Point All Engines Operating? (DPA)⁴, based on a diversion to Melbourne, was the IFR waypoint TOBOB, located 314 NM west of Sydney. Prior to TOBOB the crew obtained the 0400 and 0430 Sydney trend type forecasts (TTF). The TTFs were attached to the Sydney aerodrome weather reports and detailed the weather conditions expected to affect the aerodrome for the 3 hours following the time of the weather report. TTFs were routinely issued every 30 minutes for Sydney aerodrome and the 0400, 0430 and 0500 TTFs all indicated that, at the time of the aircraft?s estimated time of arrival (ETA), the visibility and cloud ceiling at Sydney would be above the alternate minima. The crew did not obtain the 0500 TTF. The aircraft passed the TOBOB position at 0518.

The crew commenced descent for Sydney at 0540 and, at about the same time, air traffic control (ATC) broadcast to aircraft that the Sydney automatic terminal information service (ATIS) had changed to include the remark that there was a fog bank to the north-west of the aerodrome. The ATIS was also reporting visibility greater than 10 km, with 1 to 2 oktas of cloud at 1,000 ft and 3,000 ft. The ATIS consisted of a continuous and repetitive broadcast of pre-recorded information about the actual weather conditions at the aerodrome.

The pilot in command reported that following receipt of the ATIS information, the crew obtained weather information about three aerodromes near Sydney; Williamtown, Richmond and Canberra. These aerodromes were available for use during non-normal operations (see Availability and selection of alternate aerodromes for more information). The pilot in command stated that, at the time, he considered the weather conditions at Williamtown might deteriorate to below the landing minima and Richmond was unavailable as the conditions were below the landing minima. He assessed Canberra as presenting potential difficulties due to the cloud ceiling being just below the alternate minima for runway 17 and just above the alternate minima for runway 35. In addition, the crew estimated that the fuel on board was insufficient to allow them to conduct a missed approach at Sydney and then divert the aircraft to Canberra. The pilot in command also stated that Canberra was a relatively unfamiliar aerodrome and he did not have any Notice to Airmen (NOTAM) operational information about Canberra.

The crew obtained the 0530 Sydney TTF soon after commencing descent. It indicated that, at the time of the aircraft?s ETA, there could be periods of up to 60 minutes when the visibility at Sydney would be 7,000 m in rain showers, with 5 to 7 oktas of cloud at 1,000 ft. These conditions were below the alternate minima, but above the special alternate weather minima of 400 ft cloud ceiling and 2 km visibility that applied to instrument landing system (ILS) approaches at Sydney. The aircraft was fitted with navigation equipment that allowed the use of the special alternate minima.

At 0555, as the aircraft descended through flight level 180 about 60 NM south-west of Sydney, ATC advised the crew that the Sydney ATIS had changed to include a hazard alert. This was in response to unforecast fog that had reduced the runway visual range (RVR)⁵ to 800 m. The ATIS also advised flight crew to expect ILS approaches to runways 34 Left (34L) and 34 Right (34R). One minute later, in response to that broadcast, the crew advised ATC that they required runway 16 Right (16R) for landing. This was due to the runway 16R ILS category one landing minima of 220 ft cloud ceiling and 800 m visibility being less restrictive than the runway 34L ILS category one landing minima of 270 ft and 1,500 m.

At 0558, ATC advised the crew that there would be a delay due to their requirement for an approach to runway 16R and instructed them to turn onto a heading of 230 degrees (taking the aircraft away from the aerodrome). One minute later, the crew broadcast a PAN⁶, declaring an urgency condition due to ?minimum fuel, require runway 16R for arrival?. The aircraft was then radar vectored for an ILS approach to runway 16R. The pilot in command stated that the PAN call was made because there was insufficient fuel to proceed to an alternate aerodrome approved by the operator for normal operations, ATC were vectoring the aircraft for a holding pattern and the weather conditions at Sydney were deteriorating at an unknown rate.

At about the same time, the crew of a Boeing 737 on final approach for runway 34L advised ATC that they could see the threshold and the first 1,000 m of the runway. The crew of QPC, being on a different radio frequency, did not hear that information, nor was it subsequently provided to them by ATC.

At 0604, in response to a request by ATC for the crew?s latest acceptable landing time at Sydney, they replied ?? time one eight [0618] and then we?d need to go straight to Canberra?. The pilot in command stated that a diversion to Canberra would have been the selected option only in the event that weather conditions at Sydney had deteriorated to below the landing minima. He also considered that a diversion would only ensure a ?minimum fuel state, at best?, upon arrival at Canberra.

At 0611, when the aircraft was about 17 NM north-north-west of the airport and tracking to intercept the runway 16R ILS localiser, the crew requested information about the visibility at the airport. After receiving a report from an approved observer in a motor vehicle positioned near the runway threshold, ATC advised the crew that the visibility on the runway 16R threshold was 400 m. The crew then requested the visibility on runway 34L. They were advised that the crew of a Boeing 747 approaching runway 34L ?reported visual at 1000 ft and he?s been visual all the way to 34L?. The crew of QPC then requested a vector ?for short final 34L? and were provided with radar vectoring for an approach to runway 34L.

The Boeing 747 landed on runway 34L at about 0614 and its crew reported to ATC that ??we were visual the whole way to touchdown, but I?d say this vis [visibility] at midpoint is a thousand metres or thereabouts and it?s rolling through to the south and thickening. It won?t be long before it will be a bit worse at the threshold at 34L?. The crew of QPC did not hear that information as they using a different radio frequency, but ATC advised them at 0615 that ?? the fog is now moving around at the airfield. I can?t guarantee you the same visibility as [the] company 747 of yours [that] landed on 34L?. Soon after, at 0617, one minute before the nominated divert time, ATC advised the crew of QPC that the RVR for runway 34L from the threshold was 2,000 m, which was 500 m more than the landing minima.

At 0622, the aircraft was established on the ILS approach for runway 34L. At about the same time, the crew of a preceding aircraft on approach to runway 34L conducted a missed approach because they were unable to see the runway lights. The controller passed that information to the crew of QPC. During the approach, the pilot in command of QPC saw the threshold of the adjacent parallel runway 34R and he advised ATC that he would need a right orbit onto the ILS of the adjacent runway to enable a landing. However, soon after, he decided to continue the approach to runway 34L as the aircraft was configured for an autoland. The pilot in command reported that he was concerned that if a missed approach was conducted and the aircraft was then manoeuvred for an approach to the adjacent parallel runway 34R, the heavy fog to the north may have moved across that runway during the next approach.

The aircraft was certified for autoland operations using the auto-flight system to control the aircraft during the approach and landing phase, and the crew had undergone the required training to conduct an autoland. The instrument landing systems for the various runways at Sydney were only certified to facility performance category one criteria, which meant that pilots were not permitted to continue approaches, including autolands, to runway 34L when the weather conditions were less than the specified category one landing minima.

The crew conducted a runway 34L ILS approach using the aircraft?s autoland capability and landed at 0625 in weather conditions that were less than the specified landing minima.

Civil Aviation Regulation (CAR) 257(5) specified that ?? if an emergency arises that, in the interests of safety, makes it necessary for an aircraft to land at an aerodrome where the meteorological minima is less than that determined for that aircraft operation at that aerodrome? then CAR 257(4) did not apply. CAR 257(4) stipulated that if an element of the meteorological minima for the landing of an aircraft at an aerodrome was less than that determined for the aircraft operation at the aerodrome, the aircraft was not to land at that aerodrome.

Availability and selection of alternate aerodromes

The aircraft operator provided information about aerodromes approved for A330 operations in the Route Manual Supplement (RMS), which was carried onboard the aircraft for flight crew use. If an aerodrome was approved for normal operations and the weather conditions were forecast to be better than the alternate minima, the aerodrome was classified as ?suitable aerodrome? and could be used as an alternate aerodrome for the destination. Brisbane, Coolangatta, Melbourne and Adelaide were the closest aerodromes that could be used as ?suitable aerodromes? for operations to Sydney.

The RMS also listed other aerodromes that were categorised as either emergency or adequate aerodromes that could be used during non-normal operations. Emergency aerodromes had runways of a sufficient length that could be used by A330 aircraft in an emergency situation. Nowra and Canberra were the closest emergency aerodromes to Sydney.

Adequate aerodromes met the requirements detailed in Civil Aviation Order 82.0, Appendix 2. If the weather conditions were forecast to be better than the specified ?adequate criteria?, an adequate aerodrome was considered to be a ?usable adequate aerodrome? and could be used by A330 aircraft following a critical system failure. The closest adequate aerodromes to Sydney were Williamtown and Richmond.

During the latter phase of the occurrence flight, the closest ?suitable aerodrome? was Melbourne and, once the aircraft flew past the DPA at 0518, the fuel on board was insufficient for a diversion to that aerodrome. When the crew became aware of the presence of fog near Sydney at about 0540, they obtained information about two adequate aerodromes, Williamtown and Richmond, and an emergency aerodrome, Canberra (see Sequence of events for the crew?s assessment of the weather conditions at these aerodromes).

The pilot in command stated that he considered that ATC had to be advised of an emergency condition in order to use an emergency aerodrome listed in the RMS. The operator also advised that the decision to divert to Canberra would have required the broadcast of PAN.

Aircraft fuel load

The aircraft departed Perth with 28,900 kg of fuel, which was 2,500 kg more than the minimum amount required by the operator?s approved fuel policy for the flight. The planned flight time from Perth to Sydney was 210 minutes and the fuel load gave the aircraft an endurance of approximately 315 minutes. As noted earlier, the aerodrome forecast did not require additional fuel, for either holding or a diversion, to be carried on this flight to Sydney.

The operator?s Flight Administration Manual (FAM) contained the fuel policy that specified the minimum mandatory fuel requirements approved by the operator for the operation of its aircraft. The FAM stated ?? it is recognised that there are occasions when a flight may pass DPA with the required fuel on board and a subsequent deterioration in forecast weather may then result in the minimum mandatory requirements ? not being met?.

After the occurrence, the operator calculated that the total amount of fuel required to divert the aircraft from a missed approach at Sydney to a landing at Canberra with reserves intact was 6,700 kg. Information from the aircraft?s flight data recorder showed that the aircraft landed at Sydney with about 6,900 kg of fuel remaining.

Meteorological information

The Bureau of Meteorology advised that fog at Sydney aerodrome was difficult to forecast as it was an uncommon event with between four and eight occurrences per year. Although the fog forecasting models had been upgraded in 1999, those models did not predict the formation of fog on the date of the occurrence. The classic fog scenario occurs when radiative cooling in the north-west area of the Sydney basin produces fog, which then moves to the aerodrome in a north-westerly surface airflow. For this airflow to form, several hours of clear sky is required to produce enough cooling so that the surface airflow uncouples from the gradient level airflow, which is 2,000 ft to 3,000 ft above ground level and may be moving in a different direction.

As the cloud was expected to persist during the morning of the occurrence, the duty operational meteorologist considered that the airflow uncoupling would not occur and the surface airflow would remain the same as the gradient airflow, which was from an east to north-easterly direction. At 0230, the high resolution computer forecasting models indicated precipitation and an east to north-east surface airflow. Although fog had formed in some western areas of the Sydney basin by 0330, the meteorologist considered that the rain showers near Sydney aerodrome would assist in mixing the slightly stronger east to north-easterly gradient airflow down to the surface airflow and hence reduce the risk of fog drifting in from the north-west.

The low-level cloud at the aerodrome cleared at about 0500, which coincided with the surface airflow tending to be from the north-west. About 40 minutes later fog was observed to the north-west of the aerodrome and soon after it started to move over the runways. The deterioration in visibility associated with the movement of the fog was rapid. A reduction in visibility from 2,000 m to 400 m was reported during an 11 minute period from 0552 to 0603. The fog did not clear until 0830.

The Canberra 0530 special aerodrome weather report (SPECI) obtained by the crew indicated that the wind was from the south at 3 to 4 kts, visibility was greater than 10 km, with 7 oktas of cloud at 2,100 ft above the aerodrome. The Canberra 0600 TTF was not obtained by the crew, but it did not vary significantly from the 0530 SPECI. According to the applicable Jeppesen Airport Chart, Canberra was available as an alternate aerodrome provided the cloud base above the aerodrome was not below 2,084 ft (2,144 ft for runway 17) and the visibility was not less than 6 km (7 km for runway 17). That minima was applicable with an actual QNH, which was available at Canberra, and was being broadcast on 116.7 MHz.

Provision of weather information to the flight crew

Airservices Australia (Airservices) was the air traffic service (ATS) provider. The procedures for the provision of weather information to flight crew by air traffic controllers were detailed by Airservices in the Manual of Air Traffic Services (MATS). The Aeronautical Information Publication (AIP), also published by Airservices, was an Australian operational document used by pilots. The information in the AIP that related to the provision of weather information was in accordance with the MATS.

The MATS required air traffic controllers to monitor and assess information from various reports. These included weather forecasts, amended weather forecasts and observations and reports that indicated the weather conditions at the destination aerodrome had deteriorated below the alternate minima. If that information was assessed to be of ?an unexpected and critical nature?, controllers were required to broadcast a hazard alert to flight crews.

At 0554, soon after the fog was observed moving over the aerodrome, the controllers amended the ATIS to include a hazard alert due to unforecast fog. The ATIS also advised pilots to expect an ILS approach to either runway 34L or 34R and that the RVR had reduced to 800 m in fog. This information was also broadcast at that time by the controller handling QPC.

A special aerodrome weather report (SPECI) was issued at 0603, and this report indicated that the visibility had reduced to 400 m in fog, which was below the landing minima of any runway at Sydney. The TTF appended to the SPECI indicated that these conditions would continue until 0800. The 0603 SPECI information was not passed to the crew of QPC. However, they became aware of the deterioration in the weather conditions at 0611 when ATC provided information about the RVR on runway 16R.

The RVR was assessed by approved observers in motor vehicles positioned near the runway thresholds. To determine the RVR, the observers counted the number of visible runway lights or identifiable features adjacent to the runway. This information was then broadcast to the aerodrome controller in the control tower. During the occurrence, there were observers in vehicles located near the thresholds of runways 16R, 34L and 34R.

Landing information, including RVR, was provided to flight crews either by controllers, or by the crews listening to the ATIS broadcast. The MATS section 5.1.11 contained instructions regarding the format of weather information provided in the ATIS. The MATS specified that, when the visibility was between 1,500 m and 800 m, the ATIS may report RVR information. It also stated that when visibility was 800 m or less, the RVR ?shall be reported? in the ATIS broadcast. The MATS and the AIP did not specify that the RVR information in an ATIS broadcast was to be for the touchdown zone of the runway in use.

When the ATIS was amended at 0554 to include the 800 m RVR information, the actual visibility at the touchdown zone of runway 34L was greater than that figure.

The ICAO Annex 3 ? Meteorological Service for International Air Navigation (Annex 3), contained the standards and recommended practices regarding the provision of metrological services in the aviation sector. Paragraph 4.7.4 of Annex 3 recommended that RVR observations should be made on precision approach runways for category one instrument approach and landing operations. Runway 16R and 34L at Sydney were category one precision approach instrument runways. ICAO Annex 14 ? Aerodromes defined a category one runway as ?? an instrument runway served by ILS ? and visual aids intended for operations with a decision height not lower than 60 m (200 ft) and either a visibility not less than 800 m or a runway visual range not less than 550 m?. Paragraph 4.7.2 of Annex 3 required that:

Runway visual range observations shall be representative of the touchdown zone and, depending on the category of operation for which the runway is intended and the length of the runway, of the mid-point and stop-end of the runway.

Following the amendment of the Sydney ATIS at 0554, the weather conditions continued to deteriorate with the visibility reducing to 400 m by about 0600. However, the next amendment of the ATIS was not until 0633. Requirements for the revision of ATIS information were covered in the MATS section 5.1.12. Paragraph 5.1.12.1 specified that when the current value of the visibility information was less than 1,500 m, controllers were to amend the ATIS as required if the value of the visibility was expected to vary and remain that way for at least 15 minutes.

Paragraph 5.1.12.2 of the MATS stated that in situations where controller workload precluded the amendment of the ATIS in accordance with the MATS requirements:

Tower controllers shall ensure that aircraft under their control are advised of sudden and perhaps unexpected changes to the aerodrome information, pending an amended ATIS.

Notwithstanding these requirements, there was no provision in the MATS for controllers, who were located remote from the control tower and who were passing ATIS information to flight crews, to be made aware of changes to the aerodrome information pending the updating and broadcasting of an amended ATIS.

¹ The landing minima are the Civil Aviation Safety Authority (CASA) specified meteorological conditions of cloud ceiling and visibility. In order for an aircraft to land at an aerodrome, the actual weather conditions need to be at or above the landing minima.
² Okta is a meteorological unit, equal to the area of one eighth of the sky.
³ Alternate minima are the cloud base and visibility values specified by CASA for a particular aerodrome such that, if the conditions are forecast to be worse than the alternate minima, the pilot in command must provide for a suitable alternate aerodrome.
⁴ The term DPA was used by the operator to indicate the point along the flight plan track that equated to the last position furtherest removed from the departure aerodrome from which an aircraft could divert to a nominated off-track alternate aerodrome. The minimum fuel required to be onboard at the DPA comprised the flight fuel to the alternate aerodrome, a variable fuel reserve, approach fuel, a fixed fuel reserve and holding fuel (if required).
⁵ The International Civil Aviation Organization (ICAO) Annex 3 - Meteorological Service for International Air Navigation defined runway visual range as '? the range over which the pilot of an aircraft on the centre line of a runway can see the runway surface markings or the lights delineating the runway or identifying its centre line'.
⁶ PAN is an expression, spoken three times in succession, used in the case of an urgency condition. ICAO Annex 10 - Aeronautical Telecommunications, Volume II, defined an urgency condition as '? a condition concerning the safety of an aircraft or other vehicle, or of some person on board or within sight, but which does not require immediate assistance'.

Analysis

The investigation determined that the bull gear failed as a result of a previously known high cycle fatigue cracking mechanism.

The Civil Aviation Safety Authority Airworthiness Directive (AD) requiring compliance with the manufacturer's Service Bulletin (SB) A72-2087 had been completed on the engine. However, the bull gear failed at less than half of the manufacturer's projected component life.

The diaphragm housing had been extensively damaged following the release of the section of the bull gear rim. That damage had prevented the investigation determining the housing's pre-failure condition and whether its condition had contributed to the failure.

At the time of the failure, the spectrometric oil and filter analysis program (SOAP) analysis had been carried out and assessed in accordance with the manufacturer's procedures. While the filter weight increases noted in the engine on 3 February 2004 and 1 April 2004 were within the manufacturer's 'normal sample' range, the above average filter weight, coupled with the traces of carbon steel may have been an indicator of the impending bull gear failure.

The crew handled the engine failure appropriately in accordance with the operator's procedures. The report of smoke in the cabin of the aircraft during the failure was consistent with the ingestion of engine oil into the compressor assembly immediately following the uncontained failure.

Factual Information

At 1100 Central Standard Time on 16 April 2004, a British Aerospace Plc, J32, Jetstream aircraft registered VH-OAE, with 2 pilots and 19 passengers on board, was on descent, during a scheduled passenger flight from Melbourne, Victoria to Mount Gambier, South Australia. As the aircraft passed through flight level (FL) 140, approximately 37 NM from Mt Gambier, the crew reported hearing a bang from the right engine. Simultaneously, the aircraft yawed to the right and they heard something impact the right side of the fuselage. Some smoke was evident in the cockpit.

A check of the aircraft's engine instruments confirmed a problem with the right engine and the crew shut down the engine and feathered the right propeller in accordance with the operator's quick reference handbook drills. The crew then advised air traffic control of the situation and continued for a landing at Mount Gambier Airport.

An inspection of the aircraft by the operator revealed that there had been an uncontained failure of the propeller reduction gearbox on the right TPE 331-12UHR-702H turboprop engine, serial number P66338C. There was also evidence of impact damage on the right side of the fuselage, below the co-pilot's side window area. That impact had not breached the aircraft's pressure hull.

An examination of the engine, supervised by the Australian Transport Safety Bureau (ATSB), found that a section of the spur gear teeth from the outer rim of the reduction gearbox bull gear, had detached during engine operation (See Figure 1). Spur gear teeth are radial, uniformly spaced around the gear's outer periphery and parallel to the shaft axis¹. The detached section of gear had penetrated the diaphragm housing (intermediate gearbox housing) and the gearbox accessory case, before exiting the engine through the compressor air intake.

Figure 1: Cutaway diagram of TPE 331 reduction gearbox

The hole in the accessory case had allowed engine oil to escape and flow over the engine cowling, with metallic debris and oil entering the engine's compressor intake. Once inside the compressor, the oil was able to enter the aircraft's compressor bleed air system that supplied the aircraft's air conditioning and pressurisation air. There was also significant associated damage to the diaphragm housing, the high speed pinion and compressor/turbine main shaft, with metallisation² observed on the turbine and exhaust sections.

An ATSB Technical Analysis report (Appendix A), on the mode of failure of the bull gear, part number 3108295-1, found that the gear had failed as a result of a mechanism known to the manufacturer. The report indicated that the progressive propagation of high cycle fatigue cracking within the gear web and rim transition region, had caused a section of the gear rim to separate from the gear.

The engine manufacturer had introduced several changes to the bull gear design to address 'reliability and reparability issues' that had occurred in the TPE 331 engine type. Among those were changes in gear relief, gear tooth roots had been ground and shot peened to improve fatigue life, the gear rim inside diameter was shot peened to increase fatigue resistance and a coating was added to the gear web to dampen gear vibrations. The engine manufacturer reported that despite those actions some of the re-worked and coated gears had a higher failure rate than non-reworked gears.

The engine manufacturer also investigated TPE 331 engine diaphragm housings, in which gears had failed, to ascertain if distortion of the housing could cause bull gear to pinion gear misalignment. Several problems were identified with those housings that may have contributed to the gear failures. These included bull gear to pinion gear centreline growth and misalignment, growth between diaphragm to gearbox alignment pins and out-of-round bearing bores.

In October 2001, the engine manufacturer issued Service Bulletin (SB) A72-2087³ in response to 16 in-service bull gear rim separations and 13 high speed pinion torque shaft failures. Four of those failures resulted in gearbox debris being ejected from the engine. One failure resulted in the penetration of the right side of an aircraft's pressure hull by a gear fragment. The bulletin indicated that high tooth loading on the bull gear to high speed pinion mesh, bull gear tooth profile, and distortion of the intermediate gearbox housings, had resulted in abnormal wear and subsequent failure of the assemblies.

Service Bulletin A72-2087 required replacement of the bull gear and high speed pinion with new, zero-time components, at intervals not to exceed 3,600 hours in service. It also required the inspection and the rework/overhaul of some gearbox components such as the diaphragm housing, plus a more stringent periodic inspection of specified gearbox components to ensure an optimum operating environment for the bull gear. At the time of failure, the bull gear assembly in this engine had accrued 1,199.55 hours and 1,523 cycles since installation. The engine had accrued a total of 10,755.7 hours and 12,295 cycles since new.

In Australia, the Civil Aviation Safety Authority (CASA) issued Airworthiness Directive (AD) AD/TPE 331/57 to require compliance with SB A72-2087. That AD became effective on 31 October 2001. Amendment 1 to that AD was issued in January 2002. The AD actions had been incorporated into the occurrence engine at the manufacturer's German maintenance facility on 20 December 2002.

Information received from the engine manufacturer following this occurrence, indicated that there had been three bull gear failures in post SB A72-2087 engines. One of those failures was the subject of a UK Air Accidents Investigation Branch (AAIB) investigation, published in AAIB Bulletin number 7/2005. Information from the AAIB on that failure indicated that the bull gear had failed in a similar manner to the gear in this occurrence.

The engine manufacturer reported that a spectrometric oil and filter analysis program (SOAP)⁴ was used to monitor an engine's in-service condition and to reduce the possibility of a premature mechanical failure. That program monitored the type and quantity of the deposits in the engine oil and oil filters over a specified period. A trending feature within that program could highlight an engine with a rapidly increasing filter 'weight' and indicate that further maintenance action was required. A high filter weight quantity of Carbon Steel in a sample could indicate a problem with the bull gear assembly. In November 2000, the engine manufacturer issued Alert Service Bulletin TPE 331-A79-0034⁵ that changed the SOAP interval periodicity to a fixed 100+/- 20 engine hours to minimise variability. On 25 January 2001, CASA issued AD/TPE 331/55 that required Australian compliance with that Alert SB.

The operator had complied with the engine manufacturer's and CASA's SOAP requirements, forwarding samples to the manufacturer's approved venue for testing. The operator reported that they had become concerned about a SOAP report for the occurrence engine that had been received on 3 February 2004. That report, although still within the manufacturer's 'normal sample' guidelines, had a significantly higher filter weight result than had been previously noted for the engine. When queried, the manufacturer confirmed the results of the sample and indicated that a higher reading may be seen following an engine oil change. The engine oil had been changed 101 engine hours prior to that sample being taken. In the subsequent SOAP sample taken 60 engine hours later, on 24 February 2004, the filter weight had returned to a similar level to that of the pre-3 February 2004 samples. The final sample taken prior to the occurrence, on 1 April 2004, was higher than usual and all of the samples had traces of carbon steel.

1. ASM International, Materials Information Society Handbook, Volume II.

2. Metal pulverised by the compressor becomes molten or burned in the combustion chamber and flows rearward, attaching to the turbine and exhaust assemblies (US Department of the Air Force.(1987). Safety Investigative Techniques (AF Pamphlet 127-1, Volume II). Washington DC: Author).

3. Honeywell Alert Service Bulletin - TPE 331-A72-2087, ENGINE - REDUCTION GEAR AND SHAFT SECTION - Replace Gearshaft (Sun and Bull Gear) Assembly, Part No. 3107037-9/10, 3107122-1, 3107162-1, or 3108222-1, or 3108294-1 with Part No. 3108294-1, issued October 2001 and revised 16 November 2001.

4.Service Information Letter - P331-97 - THE HONEYWELL SPECTROMETRIC OIL AND FILTER ANALYSIS PROGRAM FOR ALL TPE 331 ENGINES EXCEPT -14GR/HR ENGINES; Revision 10, Apr 5/02.

5. Honeywell Alert Service Bulletin - TPE 331-A79-0034 - OIL DISTRIBUTION - DECREASED TIME INTERVAL BETWEEN SPECTROMETRIC OIL (AND FILTER) ANALYSIS PROGRAM (SOAP) SAMPLING; Revision 4, Apr 5/02.

At 1100 Central Standard Time on 16 April 2004, a British Aerospace Plc, J32, Jetstream aircraft registered VH-OAE, with 2 pilots and 19 passengers on board, was on descent, during a scheduled passenger flight from Melbourne, Victoria to Mount Gambier, South Australia. As the aircraft passed through flight level (FL) 140, approximately 37 NM from Mt Gambier, the crew reported hearing a bang from the right engine. Simultaneously, the aircraft yawed to the right and they heard something impact the right side of the fuselage. Some smoke was evident in the cockpit.

The Boeing 747-400 aircraft, registered 9M-MPE, was conducting an international scheduled passenger flight from Kuala Lumpur, Malaysia to Melbourne. During cruise at flight level 390, the cabin crew reported an electrical smell near doors 1, 2 and 4. The flight crew immediately completed the "Smoke, Fumes, Fire, Electrical" checklist actions and the smell dissipated. Approximately one hour later, a similar smell became evident in the cockpit and the flight crew elected to divert the aircraft to Adelaide as a precaution.

After landing at Adelaide, an engineering inspection of the aircraft was conducted by the operator's engineers. This inspection revealed that the air conditioning right overhead recirculation fan had seized and the left overhead recirculation fan was causing the circuit breaker to trip. Both overhead recirculation fans were isolated electrically in accordance with the requirements of the aircraft's Minimum Equipment List and the aircraft was returned to service.