On 13 August 2018, a Boeing 737-800 (EI-FJW) being operated by Norwegian on a scheduled international passenger flight from Stewart USA to Edinburgh as a Line Check sector and about to land at destination lost prescribed separation against an Airbus A320 (OE-IVC) being operated by Easyjet Europe on a scheduled domestic passenger flight from Edinburgh to Luton as a Line Training sector as the latter was about to get airborne from the same runway in day VMC. The TWR control position was occupied by a trainee controller under the supervision of a qualified instructor who took over control at a late stage.
A Field Investigation was carried out by the UK AAIB. There was a delay in the mandatory reporting of the event direct to both the AAIB as a Serious Incident and to the UK CAA as an MOR which meant that relevant data from both CVRs were overwritten. However all relevant radar and other ATC recorded data and airport CCTV data were available.
It was noted that the 737 flight was a line check flight with a trainee Captain occupying the left seat, a Training Captain acting as PF occupying the right seat and a second Training Captain acting as the Check Captain for the trainee occupying the supernumerary crew seat. The 42 year-old Captain in command and acting as PF had a total of 6,800 flying hours which included 6,650 hours on type. All three pilots were familiar with the 737 and with operations into and out of Edinburgh and the trainee Captain undergoing his Final Line Check had recently joined the operator with previous command experience. The A320 crew consisted of a 37 year-old Training Captain occupying the left seat and supervising a trainee First Officer acting as PF who had recently joined the operator “from flight school” and had since completed “around 70 hours flying in the A320 with the operator”. To qualify as such, the OJTI (on-the-job training instructor) responsible for the TWR position had recently completed the required 5-day “practical instructional techniques” course and so was still “relatively inexperienced in that role”.
It was established that APP radar had vectored the inbound 737 to the ILS and once established asked them to maintain “at least” 160 knots to 4nm from touchdown rather than issuing the more usual instruction to fly “at” 160 knots to that position. At 8.8nm from touchdown, the flight was transferred to TWR. With the 737 6nm from touchdown and making a groundspeed of 181 knots, the trainee TWR controller cleared the departing A320 to line up on runway 06 but forgot to extinguish the illuminated stop bar at the runway holding point so the crew could not proceed. Another exchange on the TWR frequency which had immediately followed acknowledgment of the line up clearance prevented the crew alerting the controller to this oversight but once it had finished, the A320 Captain transmitted “stop bar” and it was then extinguished. This resulted in a half-minute delay to the A320 beginning to taxi onto the runway. After a further half minute, with the A320 still at 90° to the runway centreline, it was given take off clearance and the 737, at this point at 3nm from touchdown, was advised to expect a late landing clearance. With a cloud base of 300 feet, the 737 was not yet visible to either the TWR or the departing A320.
Once lined up, the A320 stopped for around 20 seconds before beginning its takeoff roll with the 737 just over ½ nm from touchdown. Almost immediately, the 737 crew, who now had the A320 in sight having emerged from cloud, called TWR to remind the controller than they were only ½ nm from touchdown without a landing clearance. The OJTI reacted by taking control of the position from the trainee. Prior to the A320 beginning its takeoff roll, the RIMCAS stage 1 Alert (issued 30 seconds prior to a predicted collision and visual only) had already been activated at the TWR position and the stage 2 Alert (issued 15 seconds prior to a predicted collision and both aural and visual) had been activated as the 737 had been told to continue approach when at ½ nm from touchdown. Minimum separation between the two aircraft - 875 metres - occurred when the 737 was 6 seconds from touchdown and passing 40 feet agl overhead the runway and the A320 was accelerating through 148 KCAS on its takeoff roll. One second after the A320 got airborne, the 737 was cleared to land and it touched down one second after this.
It was noted that since the two aircraft had not been on the runway at the same time, there was no actual runway incursion. However, “separation had been significantly eroded” and had breached the UK Manaul of Air Traffic Services (MATS) Part 1 general requirement that “a landing aircraft shall not be permitted to cross the beginning of the runway on its final approach until a preceding aircraft, departing from the same runway, is airborne” unless specific local procedures have been approved. No such specific procedures had been approved for Edinburgh.
Why it happened
The Investigation sought to examine how the conflict had been able to develop when the applicable minimum separation requirements of the Edinburgh MATS Part 2 for a landing aircraft following a departure from the same runway specified that an aircraft on approach must be at least 6nm from the runway to continue an approach, a requirement which had been met prior to the loss of separation developing. Local practice in respect of this minimum separation appeared to be ensuring that the take off clearance has been issued by the time the approaching aircraft has reached 3nm from touchdown.
Interviews with the flight crew of both aircraft involved indicated that the A320 crew had no reason to suppose that their takeoff clearance required an immediate departure and that with a supervised First Officer trainee as PF, the time taken to line up and then to beginning rolling was entirely commensurate with the very low experience of the trainee. These interviews also showed that the experienced 737 crew had complied in a reasonable manner with the APP radar controller’s request for them to continue at or above 160 KIAS until 4nm whilst still performing a stabilised approach. There was no evidence that an observed Line Check which was being performed on the sector had affected the way the aircraft was operated and its crew were unaware of the departing traffic until seeing it on emerging from the overcast base at 300 feet aal. The trainee controller’s response of “continue and expect a late landing clearance” in response to their “½ nm to go” call was considered to have “confirmed in the minds of the crew that the controller was fully aware of the position of both aircraft”.
It was noted that with a relatively low cloud base such as prevailed prior to and at the time of the investigated event – even though still in excess of requiring LVPs to be in force - a TWR controller gains less information by looking out of the window than in clear conditions. Consequently, both the OJTI and the trainee controller needed to rely more on information presented on the various screens. However, it was found that “the design of the screens is such that there is not a good view of them from any position other than the controller’s position” (see the illustration below). For this reason, an OJTI usually uses a higher chair for the explicit purpose of being able to see all the screen information available to the trainee and if such a chair is not available, then an OJTI “may need to be more active in changing positions to see the information they need to effectively monitor the trainee controller”. It was found that only one of these higher chairs was provided in the TWR control room and that at the time of the investigated event, both TWR and GND positions were occupied by trainees overseen by OJTIs with the only available high chair being used by the GND position OJTI.
The TWR controller workstation at Edinburgh. [Reproduced from the Official Report]
It was established that in the first instance, the trainee controller had not recognised the implications of his failure to turn off the stop bar lights when clearing the A320 to line up and although he had issued an “expect late landing clearance” to the 737, he had not also qualified his A320 line up clearance with “be ready immediate” or given his A320 take off clearance as cleared “immediate” takeoff. Had he done both, or even one of these, it is likely that the Training Captain in command of the A320 would have responded in a way that would have reduced the delay getting airborne. It was concluded that at the time of the loss of separation, “the trainee controller was still relying on techniques more suited for use in better weather conditions (and was) also using a rule-based strategy which checked the gap was sufficient as the preceding aircraft touched down”. The trainee did not appear to have made any use of the approach radar display showing approach traffic and its groundspeed and was not effectively monitoring aircraft speeds or the size of the gap between them. It was also concluded that he may well have had little or no on-the-job experience of the RIMCAS conflict alerting system. He certainly did not become aware of the rapidly reducing gap between the two aircraft until he saw the 737 emerging from the cloud and the RIMCAS stage 2 Alert began to sound and had little time to react to the situation given his limited capacity to recognise and deal with the deteriorating separation. In particular, he had “little or no experience of instructing a go-around” and this “inexperience, together with the short time period available to act after becoming fully aware of what was happening resulted in an inability to recover the situation”.
It was considered to be of considerable significance that the OJTI had also missed the developing situation and he also only became aware of its seriousness when the 737 became visible and the RIMCAS Stage 2 alert began to sound. He reported having been “startled by the suddenness of the situation” and that this “caused a further delay in his reaction”. However, although he could see out of the window that both aircraft were close, he was at this late stage less concerned about any further gap closure than with the risk which would follow from instructing the 737 to go around when it was too late to stop the A320 taking off since this would result in a potentially unsafe proximity as both aircraft disappeared into cloud where they could not be visually separated. Having taken over the position, he therefore made the decision to allow the 737 to land. It was observed that an earlier intervention could have allowed a safe go around by the 737 by keeping the A320 on the ground.
Overall, the evidence was considered to show that more effective support for the trainee by the OJTI could have avoided the situation developing in the way it did. Until a late stage in the developing loss of separation, the OJTI, who had gained five months experience since qualifying as a instructor had taken the view that as their (ab initio) trainee was over halfway through their training to qualify as a TWR controller and had generally been performing satisfactorily, he had “judged the trainee to be […] at the point where less OJTI input was needed”. This was within the general guidance for the trainee’s level of experience at the time which was that “in medium traffic/medium complexity conditions ‘minimal’ support is expected”. ‘Minimal’ was defined as meaning “minor support where necessary such as making reference to complex problems or asking the trainee to develop a course of action in time”. Only in “high traffic or high complexity conditions” would “some” support be expected.
Controller Training at Edinburgh
The background to the poor controller performance in the investigated event was noted as being a changeover of ANSP at Edinburgh which had taken place 4½ months prior to its occurrence and had led to the need for a sustained period of controller training due to a significant exit of previously employed controllers. It was found that in the 15 months leading up to the change of ANSP, a total of eight controllers with previous experience elsewhere were recruited and trained at Edinburgh by the previous ANSP and a further five ab-initio controllers had been recruited by the new ANSP but had not begun their training until after the changeover. This degree of training was unusual for the Unit - 13 new controllers trained within 18 months compared to the historic norm of “around one new controller a year”. This amount of training also required the new ANSP to upgrade a significant number of their newly re-qualified controllers as OJTIs, including some who had “little or no previous experience” in training ab-initio controllers.
It was also found that despite the use of pre-session training briefings to be conducted by OJTIs during the one week OJTI qualification course, at Edinburgh “there was no expectation that this would occur, and no time was set aside for it”. It was considered that despite the small size of the Unit, such a pre-session brief would have served to clarify expectations for a session and allowed OJTIs to proactively assess the trainee’s experience level relative to the prevailing operational conditions and review the appropriate ways to make use of the information available.
It was also found that there was no quick reference summary of the significant content on the reporting forms used by OJTIs at Edinburgh to record trainee progress relative to the challenges they had dealt with and also that the individual completed reporting forms, although often quite detailed, “rarely included details of the weather conditions experienced”. This was considered to “make it hard for OJTIs to quickly get an impression of a trainee controller’s recent experience and performance” to help decide what the emphasis for each session should be.
The formally documented Conclusion of the Investigation was as follows:
A succession of short delays to the departure of the A320 and the higher than normal approach speed of the 737 led to the rapid closure of the gap between the two aircraft. The loss of spacing went unnoticed by both the trainee controller and the OJTI until the 737 came out of cloud, which was just before the crew prompted them by calling at ½ nm. At this point the OJTI made the decision that it was safer to land the 737 than risk having two aircraft that he could not separate visually close to each other in cloud above the airport.
The crew of the A320 were completely unaware of the developing situation as they could not see the 737 nor had the trainee controller instructed them either to be ‘ready immediate’ or cleared them for an ‘immediate takeoff’. The crew of the 737 were confident in the TWR controller and were not initially concerned that they had not received a landing clearance. They were used to operations at London Gatwick where traffic levels are significantly greater than at Edinburgh. They became concerned enough to prompt the controller at ½ nm but the reply only served to reinforce their confidence that the controller was on top of the situation. As a result, they decided not to perform a go-around and continued to land once clearance was given. This led to a loss of separation between the aircraft at a critical phase of flight.
Safety Action taken by the Edinburgh ANSP and known to the Investigation was noted as having included, but not been limited to, the following:
- Introduced refresher training for all OJTIs.
- Increased the availability of higher OJTI chairs so that OJTIs can properly oversee the trainee position.
- Reminded OJTIs of the requirement in the Unit Training Plan which mandates the requirement for a pre-training briefing between the OJTI and a trainee controller prior to every training session or at least once every training day.
- Has added a one-sheet overview of trainee controller experience to their training file on which an OJTI must record significant operating conditions encountered and procedures used such as LVPs, high surface winds, go-arounds, significant slot delays, weather avoidance and snow.
The Final Report of the Investigation was published on 30 May 2019. No Safety Recommendations were made.