On 5 July 2015, a Boeing 777-300ER (A6-EBU) being operated by Emirates on a scheduled international passenger flight from Melbourne to Singapore as EK405 which had taken off from runway 34, a Boeing 737-800 (VH-VYE) being operated by Qantas on a scheduled domestic passenger flight from Canberra to Melbourne as QF819 which was on a go around from an approach to the same runway, and another Boeing 737-800 (VH-VXS) being operated by Qantas on a scheduled domestic passenger flight from Sydney to Melbourne as QF449 also going around, all came into proximity as they followed ATC instructions at night.
An Investigation was carried out by the Australian Transport Safety Bureau (ATSB). It was established that the TWR position was being occupied by a trainee controller accompanied by an OJTI who had held a Melbourne Aerodrome Control Rating for 6½ years. A ‘Tower Coordinator’ was also present - he had held a Melbourne Aerodrome Control Rating since 1996 - as was another controller tasked with Ground Movement Control and Clearance Delivery.
The trainee controller instructed the 777 to line up on runway 34 following discussion with her OJTI. At this time, the two 737s were on approach, one for runway 34 and one for the shorter runway 27. Both were about 5 nm from the threshold of their respective runways with Land and Hold Short Operations in use. About 30 seconds later, as the 777 continued to taxi towards the runway, it was cleared for an immediate take-off and, after making a 90° turn to line up on the full length of the runway, commenced its take-off.
This manoeuvre took longer than the trainee controller and her OJTI had expected and as the 777 began its take-off roll, the 737 on approach to land on the intersecting runway 27 had about 2½ nm to go and the 737 on approach to land on runway 34 had less than 2 nm. The trainee tower controller requested the flight crew of the 737 approaching runway 27 to reduce to minimum speed and its crew responded that they already were at that speed. Her OJTI then told the trainee to tell the crew that in the event of a missed approach, they should expedite climb to 4,000 ft as there was traffic departing runway 34. In response the 737 crew, descending through approximately 900 feet, announced that they were going around. Meanwhile, the other 737 was passing 800 feet on approach to runway 34.
At this point the Tower Coordinator “expressed concern” that the departing 777 was still on runway 34 and that the required runway separation standard would not be maintained against the approaching 737. After he had “received what he deemed an unsatisfactory response from the tower OJTI” he then instructed the trainee controller to instruct the 737 on final to runway 34 to go around. As she did this, a runway 34 occupancy caution activated because the 737 was less than 15 seconds from the threshold. This caution almost immediately changed to a warning which is triggered when an aircraft is less than 10 seconds from the threshold. The 777 was still less than half way along the runway and the 737 going around from runway 27 was less than 1 nm from the 27 threshold and climbing through 1,100 feet. The trainee controller, intending to advise the flight crew of the 737 she had just instructed to go around from runway 34 that another 737 was currently going around from runway 27, transmitted instead traffic information that another 737 was going around from runway 34.
As the 737 going around from runway 34 overflew the runway 27/34 intersection climbing through 600 feet, the 777 was still on the runway south of the intersection and the 737 going around from 27 was almost over the runway 27 threshold and climbing through 1,200 feet. As the 737 going around from runway 34 passed 1,000 feet in the climb still over the runway (see the illustration below) the trainee controller was instructed to tell it to turn right onto north whilst it was still below the 2,000 feet Minimum Vectoring Altitude (MVA). At this time, the 777, now airborne, was north of the runway 27/34 intersection climbing through 800 feet and the 737 was just east of it and climbing ahead through 1,700 feet. Soon after this the trainee controller issued a wake turbulence caution to the flight crew of 737 going around from runway 34 in relation to the 777 which had just taken off from the same runway.
The relative positions of the three aircraft as the 737 ground tracks were about to cross. [Reproduced from the Official Report]
Subsequently, all aircraft were “tracking well clear of each other” on divergent flight paths. The 777 continued en-route and the two 737s were then re-sequenced to land at Melbourne without further event. It was noted that at no time had traffic information been provided to the 777 in respect of either of the 737s. Surveillance data was subsequently examined and showed that separation between the two 737s had reduced to about 0.9 nm and 900 feet as the one going around from runway 27 had crossed runway 34 in front of the one going around from that runway. It was noted that responsibility for maintaining separation lay with the TWR controller and the only available standard was visual separation which, because of “the limitations of human vision at night and the disposition and trajectory of traffic at the time” could not be assured. It was also found that although the 737 sent around from its approach to runway 34 did not fly through the flight path of the departing 777, it was only 1½ nm behind it instead of the applicable wake turbulence separation minimum of 5 nm.
A number of aspects of the event and its context were examined during the Investigation, including the following:
- The entrance to the full length of runway 34 had a yellow marked centreline intended for use by aircraft exiting from the 16 direction as well as taxiway lighting which when followed results in a 90° turn to line up for takeoff, as Illustrated below. Both the trainee and her OJTI stated that when the 777 was issued with clearance for an immediate take-off, they had expected it to line up via the 16 exit line which aircraft sometimes do. However, it was noted this assumption was at variance with the requirements of an immediate takeoff clearance which are only that the aircraft shall taxi onto the runway and commence takeoff in one continuous movement. The aircraft operator also subsequently confirmed that performance calculations for the departure involved had required use of the full runway length.
The runway 34 threshold area. [Reproduced from the Official Report]
- Use of LAHSO at Melbourne was noted to increase runway capacity from about 24 movements per hour to about 44 movements per hour and at the time of the investigated event was also in use at Adelaide and Darwin. LAHSO clearances may only be given to domestic operators who have been granted corresponding regulatory approval and at Melbourne require a cloud base not less than the 2000 feet MVA within 8 nm of the ARP (Aerodrome Reference Point) and a prevailing visibility of no less than 8km. During LAHSO, the TWR controller “is responsible for maintaining visual separation in the event of a missed approach or a dual missed approach […] until such time as another separation standard can be applied, either 3 nm or 1,000 feet”.
- The applicable runway separation standards in this event mean that an aircraft landing behind a departure must not cross the runway threshold until the preceding aircraft is airborne and has either commenced a turn or is both beyond the point on the runway at which a landing aircraft could be expected to complete its landing roll and there is no risk of conflict if it were to go around.
- Controllers must apply wake turbulence standards to an aircraft departing or going around behind another aircraft. This may be based on either time or distance. The 777 was a ‘Heavy and the 737 was a ‘Medium’ which meant that the applicable minimum separation was 2 minutes or 5 nm. It was noted that as Melbourne was a radar tower environment, only the distance standard was applicable.
- Traffic Information should be provided when, in the controller’s judgement, one aircraft may observe another aircraft and could be uncertain of their intention. Its provision and the information provided rely on the controller’s “assessment of the underlying need”. No traffic information was given to the departing 777. Had the (visiting) 777 crew been made aware of the 737 on final for runway 34 and the other 737 on final to runway 27, they would have had the option of declining the line-up and take-off clearance given.
- Compromised separation recovery actions are important emergency response actions and need to be implemented promptly and accurately when separation standards have been, or will shortly be, compromised. They need to be regularly practiced. Controllers are required to issue safety alerts as a priority when they become aware that aircraft are considered to be in unsafe proximity unless a pilot advises that action is being taken to resolve the situation or that they have the other aircraft in sight. No safety alerts were issued by the trainee tower controller during this proximity event.
- There were no TCAS RA activations during this event because they are inhibited below 1,000 feet agl which was where the closest proximity occurred.
- When LAHSO is in use, runway 27 is the designated departure runway and runway 34 is normally used only for landings. However, since 27 is the shorter of the two runways, larger/heavier departing aircraft sometimes need to use runway 34 as in this case. The computerised sequencing tool used at Melbourne, ‘MAESTRO’ displays each arriving aircraft’s landing runway and its position in the sequence so that all controllers have to do is use speed control, vectoring or holding to achieve an “orderly traffic flow”. However, runway 34 departures are not taken into account of by MAESTRO and the controller must fit such departures into the arrival sequence based on their best judgement, ensuring that the landing aircraft will not touch down until the departing aircraft is airborne.
The Investigation formally documented six Contributing Factors as follows:
- The decision to clear the flight crew of A6-EBU for an immediate take-off, combined with the aircraft’s slower than anticipated rate of movement, resulted in it coming into proximity with VH-VXS and VH-VYE.
- The proximity between A6-EBU, VH-VXS and VH-VYE resulted in the flight crew of VH-VXS on final for runway 27 electing to go around and the flight crew of VH-VYE on final for runway 34 being instructed to go around on the direction of the Melbourne Coordinator.
- Although initiated due to a safety concern, the decision by the Melbourne Coordinator to instruct the crew of VH-VYE to go around resulted in an airborne loss of separation compared to a potential loss of separation on the ground.
- The simultaneous go-arounds conducted by VH-VYE and VH-VXS, sequenced to land on intersecting runways under land and hold short operations at Melbourne Airport, resulted in the controller issuing a radar vector to the flight crew of VH-VYE while the aircraft was below minimum vector altitude to assure surveillance and wake turbulence separation.
- The radar vector issued at night to the flight crew of VH-VYE when no other options were available, though intended to ensure wake turbulence and surveillance separation behind A6-EBU and separation assurance with VH-VXS, did not assure terrain and obstacle clearance.
- The hazard associated with the inability to separate aircraft that are below the appropriate lowest safe altitude at night was identified but not adequately mitigated. This resulted in a situation where, in the event of a simultaneous go-around at night during land and hold short operations at Melbourne Airport, there was no safe option available for air traffic controllers to establish a separation standard when aircraft were below minimum vector altitude. [Safety Issue]
Two Other Factors that increased risk were also identified:
- The lack of night-time compromised separation training scenarios for the Melbourne Air Traffic Control Tower controllers increased the risk of the controllers responding inappropriately when aircraft were in proximity at night.
- The automated sequencing system used by Airservices Australia at Melbourne Airport (MAESTRO) did not ensure that two aircraft would not arrive at the intersection of the runways at the same time during land and hold short operations, increasing the risk of unsafe proximity at the intersection.
Safety Action taken by ANSP Airservices Australia as a result of the investigated occurrence whilst the Investigation was in progress was recorded as having included the following:
- Night LAHSO were suspended at Melbourne and Adelaide airports on 10 November 2015. An assessment of the obstacles in the Melbourne Airport area was then conducted which confirmed that aircraft complying with controller-issued headings of up to 50° left of the extended centreline of runway 27 and right of the extended centreline of runway 34 following a missed approach would safely clear all relevant obstacles. On 22 April 2016, the Civil Aviation Safety Authority issued an exemption to Airservices Australia to radar vector aircraft below the minimum vector altitude in relation to aircraft carrying out simultaneous go-arounds on runway 27 and runway 34 at Melbourne Airport at night during LAHSO. On the same day, night LAHSO were reinstated at Melbourne.
- Training in night-time compromised separation recovery during LAHSO was completed by all Melbourne Tower controllers in March 2016.
- On 10 March 2016, an arrival stagger was introduced for LAHSO arrival pairs at Melbourne to ensure that, in the event of a missed approach, the aircraft would not be in unsafe proximity at the runway intersection.
The Final Report was released on 6 August 2018. No Safety Recommendations were made.