On 31 October 2012 a Boeing 737-800 (LN-DYC) being operated by Norwegian Air Shuttle on a scheduled passenger flight from Trondheim to Oslo as NAX 741 lost separation in day Instrument Meteorological Conditions (IMC) against another Boeing 737-800 (LN-NOM) also being operated by Norwegian Air Shuttle on a scheduled passenger flight from Oslo to Trondheim as NAX 740 after the former had advised that a go around was being flown from the approach to the same runway from which the other aircraft had just departed. Both aircraft began the left turn given only to the aircraft on go around and continued climbing on similar tracks until the aircraft on go around altered their heading prior to a similar action being given by ATC. There were no abrupt avoidance manoeuvres and none of the total of 232 people on board the two aircraft were injured.
An Investigation was carried out by the Accident Investigation Board Norway (AIBN). DFDR data from both aircraft and recorded ATC radar and communications were available to support the Investigation but no action was taken to prevent the 30 minute Cockpit Voice Recorder (CVR) recording on either aircraft from being overwritten.
The Investigation noted that all the pilots involved were experienced in their assigned roles and generally, as well as being very familiar with Oslo Airport and its air traffic procedures and the fact that the Gardermoen CTZ was class ‘D’ airspace. It was established that on-the-job training (OJT) was in progress at the Oslo APP position, the final radar frequency worked at the Oslo ACC before inbound traffic was handed off to the TWR. It was also noted that there had been a changeover at the TWR position just four minutes prior to the investigated occurrence. The Automatic Terminal Information Service (ATIS) broadcast accessed by the crew of the arriving aircraft was noted to have included the remark “tailwind on final to 2000”.
It was found that at the time of the event, the airport was operating with a single runway (01L) in use for both take offs and landings due to periodic 15 minute closures of alternate (parallel) runways to facilitate snow clearance operations. Three landings had been planned on 01L during the period of single runway operations based on a target distance between successive landings of 8nm - one of them being the aircraft which was involved in the investigated conflict whilst on go-around.
At time 07:35:28. Correlation of FDR data shows that the minimum distance between the two aircraft was 0.17 NM horizontally and 500 ft vertically. Both aircraft were in clouds at this time, which is not depicted in the image. (reproduced from the Official Report)
It was apparent to both the flight crew and ATC from an early stage that the aircraft on approach was both too high and too fast relative to a normal Instrument Landing System (ILS) closure and profile. The Investigation noted that although the aircraft was established on the ILS LOC at a range of 9.5nm, this occurred at 220 KIAS and above the ILS GS at in excess of 4000 feet aal. The Automatic Terminal Information Service (ATIS)-alerted tailwind component was also present.
Although the aircraft commander of this aircraft told the Investigation that “it was difficult to get down to the glide path and at the same time reduce the speed because of the strong tailwind”, they had chosen to continue the approach, “hoping that the tailwind would decrease”. He stated that “they had not applied airbrakes during their descent or during the final approach stage because he considered the effect to be small and it would have led to vibrations in the aircraft, which would have increased further when they set flaps”.
The aircraft was transferred from Radar APP to TWR at 6nm out with an instruction to reduce speed to 180 KIAS or less - airspeed was still indicating in excess of 210 knots at that time with a ground speed of 250 knots. The crew inadvertently selected the wrong TWR frequency and called APP instead so that actual check in with TWR was delayed. By this time, the aircraft was 3.7nm from the runway threshold at an altitude of 2100 feet (equivalent to 1400 feet aal) and still had a ground speed of 250 knots.
Immediately after acknowledging the aircraft checking in, the TWR controller issued take off clearance to the departing aircraft - which had already been advised to commence any required run up and be read to roll immediately they were cleared to do so. Although the inbound aircraft crew heard this clearance, they stated that they did not appreciate its significance as they continued their approach attempt. They had selected Flap 15 shortly before visual contact with the runway when there was approximately 1nm to go but they were still at 700 feet aal (and therefore 400 feet above the ILS GS altitude) and with an airspeed still around 190 knots. At this point - well beyond the 1000 feet aal stabilised approach ‘gate’ which formed part of their operating procedures, the aircraft commander determined that a go around was necessary and advised TWR as it was initiated. TWR acknowledged and added “departing traffic below, essential traffic”.
The go around was commenced as per the procedure, straight ahead on the runway extended centreline and 15 seconds later, the aircraft overflew the threshold of the runway with a groundspeed of 190 knots. At this point the departing aircraft had just become airborne with a groundspeed of 154 knots approximately a mile ahead. Shortly afterwards, as the departing aircraft passed 700 feet agl, the TWR controller instructed the aircraft on go around “when able, left turn heading 270.” This instruction was then acknowledged by both aircraft - although ATC heard only the incorrect acceptance from the departing aircraft and did not respond to either aircraft - and the aircraft on go around did not hear the incorrect acceptance transmitted by the departing aircraft.
Soon after the departing aircraft initiated the left turn, it received a TCAS TA with the other aircraft symbol showing 400 feet above on the TCAS display and the crew acted to reduce their rate of climb. Since the TWR controller no longer had either aircraft in sight due to the low cloud base but could see from the radar monitor that both had or could be expected to shortly begin left turns, they advised the departing aircraft of “essential traffic above on missed approach turning left shortly” which was acknowledged. Twenty seconds later, minimum separation of 0.19nm was reached with the aircraft on go around passed an altitude of 1800 feet (equivalent to approximately 1100 feet agl/aal) some 500 feet above the and ahead of the departing aircraft - which was still below 1000 feet agl/aal. With the conflict resolved, the TWR controller transferred both aircraft to the next appropriate frequency.
Subsequently the aircraft on go around also reported receiving a TCAS TA when in cloud and after starting its left turns. When their TA occurred, the flight crew of this aircraft did not see a height tag against the traffic symbol for the departing aircraft but observed that the traffic symbol for it was almost co-located with their position on the TCAS display. They stated that, upon realising for the first time that the other traffic ahead of them had just taken off, they increased thrust to increase their rate of climb. Noticing that it was also turning left, they reported that they had also responded by reversing their direction of turn and going to the right ahead of the complimentary instruction from the TWR controller being received soon afterwards instructing a right turn onto a heading of 360°.
Correlation of FDR data showed that the minimum horizontal distance between the two aircraft occurred with the departing aircraft climbing through 1300 feet QNH (approximately 600 feet aal) and the go around aircraft climbing through 1800 feet QNH when they were 0.19 nm apart.
The Investigation noted (listed below in no intended order of significance) that:
- There were no instructions at Oslo ACC to prescribe where OJT instructors should be seated when they were working in position with trainees, although the ability to override trainee communications was available at all positions.
- Short Term Conflict Alert (STCA) had “no practical application in this incident” because it was configured to be active only if at least one of the aircraft in a conflict was above 5500 feet.
- The Norwegian Air Shuttle Operations Manual stated unequivocally that any approach should be stabilised before the aircraft reaches 1000 feet aal.
- ATC controller training is delivered in accordance with the EUROCONTROL specification for ‘Common Core Content’.
- Although the use of a tower simulator for controller training was not mandatory in Norway in line with EU Regulation 805/2011 at the time of the investigated event, such a simulator had been in use at Oslo for many years.
- the first leg of the outbound flight procedure from runway 01L and the missed approach procedure for runway 01L both initially follow magnetic heading 014°, i.e. straight ahead along the extended runway centreline.
- the prevailing separation standards after the inbound aircraft had commenced its missed approach were 3 nm horizontally or 1000 feet vertically.
- Go around procedures did not address separation between an aircraft on missed approach and a simultaneous departure off the same runway, only the case of normal parallel runway operations with one runway for take off and one for landing.
- There was no standardisation amongst those controllers qualified as TWR Supervisor in respect of what they considered was a sensible number of landings to be accepted on a runway also being used for departures during 15 minute snow clearance periods.
- The failure to secure CVR data meant that the Investigation lacked relevant information which could have helped to understand the sequence of events in detail.
In its analysis of the facts gathered, the Investigation’s observations included the following:
- By going around at an earlier stage, the crew (of the approaching aircraft) would have had more time to implement necessary corrective measures (and) the tower controller would also have had more space to accommodate separation from other aircraft.
- Since information on the radar monitor in the tower showed that the aircraft on approach was maintaining an abnormally high speed, the tower controller would have been wise to wait before giving take off clearance to the departing aircraft and to also instruct the approaching aircraft to go around.
- If the aircraft on approach had been able to complete a landing, the separation between its landing position and the departing aircraft would have been less than the full runway length.
- The Tower controller’s assessment not to cancel the take off clearance given to the departing aircraft once the potential conflict had arisen was reasonable since the aircraft was about to rotate and had consequently continued too far to initiate a rejected take off.
- Although aircraft should not normally deviate from ATC clearances, the decision of the crew of the aircraft on go around to turn away from the aircraft which had just taken off ahead of them and was located below them was a sound assessment.
- The decision of the crew of the departing aircraft to reduce their rate of climb when they became aware of the conflicting aircraft immediately above them was a sound one.
A similar event at Bergen on 4 October 2011 which had also been investigated by the AIBN was noted. In this case, a Norwegian Air Shuttle Boeing 737-800 had discontinued its approach to runway 17 just before landing when its crew considered it unlikely that a Boeing 737-300 on the runway and in receipt of a take off clearance would be clear of the runway before their likely time of landing. They were subsequently unaware that the aircraft they were concerned about had actually taken off and was ahead of them until they saw it and made an immediate turn off the same track. Although minimum separation had been only 0.5 nm horizontally and 400 vertically, the Investigation concluded that there had not been any real risk of collision despite the potential for it. This Investigation, with the Final Report published only in Norwegian, had led to the issue of three Safety Recommendations to the State ANSP Avinor, two of which were still open in October 2013, one of which was as follows and was considered to be equally relevant to the Oslo event:
- that Avinor assess whether other air traffic control units that bear responsibility for control tower and/or approach control have satisfactory local guidelines for handling traffic management during a missed approach. [SL No. 2012/06T]
The formally-stated Conclusion of the Investigation was as follows:
“The two aircraft came into conflict with each other during go-around and simultaneous departure, respectively. AIBN’s assessment is that there was a real danger of collision during this incident. But once the situation was a fact, situational awareness and sound assessments on the part of the aircraft crews and the tower controller prevented a further escalation of the conflict. AIBN considers established and practiced procedures to be conducive to recognising situations and averting them before they become critical.”
Three Significant Findings which were considered to be of decisive significance for the sequence of events or are especially important viewed from a flight safety perspective were also highlighted as follows:
- The crew of the aircraft on approach had unrealistic expectations of becoming stabilised not later than (the prescribed) 1 000 feet aal. The decision to implement a missed approach was taken at a late stage.
- The Final Approach and Tower controllers’ expectations that the speed of the approaching aircraft would be reduced sufficiently were not fulfilled and as a result he aircraft came into conflict with the simultaneous departure of the other aircraft.
- Since runway 01L was being used both for landings and departures, the Tower controller needed to secure sufficient separation during any go-arounds as well as for landings and take offs. ATC did not have documented local guidelines that could help the tower controller assess whether there might be enough time to take a departure ahead of a landing under the prevailing conditions. Nor were there any documented guidelines for how the tower controller should handle traffic management during missed approach and simultaneous departure from the same runway.
The Investigation was completed on 7 November 2013, the Final Report published in Norwegian on 11 November 2013 and the Final Report in English was subsequently published on 25 February 2014.
The only information documented in respect of significant Safety Action taken during the course of the Investigation was that to address callsign confusion risks, the aircraft operator involved had since made changes to a number callsigns for flights likely to be operating in the same area at the same time, including changing the callsign for the departing NAX 741 to NAX 1YE to avoid possible confusion with the arriving NAX 740. No new Safety Recommendations were made because the content of one which might otherwise have been was already contained in the previously quoted and still open one made to ANSP Avinor following the similar conflict at Bergen on 4 October 2011.