Description

On 29 June 2010, an Airbus A319 (HB-JZQ) being operated by Easyjet Switzerland on a scheduled passenger flight from Palma to Basel Mulhouse as DS1058 and another Airbus A319 (F-GRHA) being operated by Air France on a scheduled passenger flight from Basel-Mulhouse to Paris CDG as AF7343 lost separation whilst following clearances given by a trainee controller which brought them into conflict. One of the cabin crew of the Easyjet aircraft, which was crewed by a Captain under training and a Training Captain, sustained minor injuries during the abrupt manoeuvring which followed the annunciated RAs.

Investigation

An Investigation was carried out by the French Bureau d'Enquêtes et d'Analyses (BEA). Flight Data Recorder (FDR)/QAR data was available for both aircraft but Cockpit Voice Recorder (CVR) data was available only for the Air France aircraft after no attempt was made to prevent overwriting of the Easyjet aircraft CVR.

It was established that with procedural ATC service being provided due to radar unserviceability, the Air France A319, with the aircraft commander as PF had been cleared by the trainee APP controller to climb on the LUMEL 5T SID to FL 110, a generally westerly track. Approximately a minute later, the same trainee controller also cleared the inbound (Easyjet Switzerland) A319, being flown by a Training Captain as PF in the right hand seat supervising a trainee Captain occupying the left seat and flying a generally northerly track to the west of the airport, to descend to the same level. The OJTI did not notice the error until he noticed, on a radar display not available for use because of the promulgated unserviceability of the system, that the Air France aircraft was passing FL 100 and immediately alerted the trainee controller accordingly.

Shortly after the outbound (Air France) aircraft had had a request for a slight left turn for weather avoidance (there were convective cells in the area) approved and was approaching FL110 at 3000 fpm (a rate which was contrary to both Airbus recommendations and Air France requirements), coordinated TAs occurred simultaneously in both aircraft. Three seconds later, about the same time as the OJTI had noticed and pointed out the conflict, an Short Term Conflict Alert (STCA) visual activation occurred at the APP controller’s workstation. The trainee controller responded to the alerting by his OJTI by instructing the Air France aircraft, as it was passing FL106 for its cleared level FL110, to descend to FL100. Then, 10 seconds after the TAs, co-ordinated RAs occurred - a preventive ‘MONITOR VERTICAL SPEED’ for the Easyjet aircraft which was level at FL110 and a (nominally) corrective ‘MAINTAIN VERTICAL SPEED, CROSSING MAINTAIN’ for the Air France aircraft.

In response to their initial preventive RA, the Training Captain in command of the Easyjet aircraft and occupying the right hand seat as PF, immediately disconnected the AP and A/T, pitched up and increased thrust. After a 50 feet gain in altitude, the thrust levers were moved to Idle. Nine seconds after the initial RA, ‘DESCEND CROSSING DESCEND’ was annunciated for just two seconds and was followed immediately by a ‘CLIMB’ RA. The PF applied a sharp nose up input, taking the pitch attitude from -5.6° to +13.8° whilst at the same time turning the aircraft 26°to the left. As the aircraft reached a minimum altitude of 10870 feet, an ‘INCREASE CLIMB’ RA was annunciated and the thrust levers were moved from Idle to TO/GA with the vertical load factor reaching 2.04 G. An unspecified RA was advised to ATC.

In response to their initial RA, the crew of the Air France aircraft, who had just begun to respond to the ATC ‘descend’ instruction received as they passed FL106 by reducing pitch attitude, recommenced the climb. The RA was announced by the First Officer incorrectly to ATC as a “TCAS Climb”. Eleven seconds later, at the same time as the other aircraft received its ‘Climb’, the Air France aircraft received a DESCEND RA. The necessary pitch down was made, ATC were advised correctly of a TCAS Descent with the aircraft reaching a maximum of 11,040 feet before commencing the descent.

The tracks and altitude changes in above two illustrations are annotated with numbers which identify equivalent points on both. In effect, the initially applied TCAS logic was to resolve the imminent loss of separation by keeping the Air France aircraft climbing - if necessary beyond its cleared level of FL110 and ensuring that the Easyjet aircraft, already level at FL110, stayed there. In line with the TCAS RA process, this resolution assumed no changes of track. But what happened was that, with a rate of climb that was abnormally high when so close to a cleared level, the Air France aircraft took more time to reverse its climb than would otherwise have been the case and at the same time, the Easyjet aircraft was put into a descent towards it. This resulted in both aircraft getting further RAs to reverse their respective climb and descent to which both responded and the Easyjet aircraft was additionally turned approximately 30°to the left of its previous track. Shortly afterwards, both aircraft received RA’s to adjust their vertical speed and the “Clear of Conflict” messages came 36 seconds after the first RAs.

The Investigation calculated that “the minimum angular separation in the sequence…corresponded to a separation of 0.29 nm horizontally and 1760 feet vertically”. It was also found that “the two aircraft were at the same flight level for the second time at a horizontal distance of 2.2 nm, i.e. (for) a flight time of about 16 seconds with a closing speed of approximately 500 knots”. By the conclusion of their respective descent/climb, the Air France aircraft had reached FL105 and the Easyjet aircraft almost FL126. The effect of the turn made by the Easyjet aircraft during the RA sequence was analysed and it was noted that it had resulted in the aircraft passing in front of the Air France aircraft instead of behind it which it would have done if it had maintained its track and it “did not increase the separation”.

The TCAS RA response procedure was reviewed. It was found that the Airbus procedure, adopted in full by Easyjet, required no action until an RA occurred when, for all RAs, both preventive and corrective, the AP and both FDs should be disconnected. In the case of Air France, their procedure at the time required that the FD be switch off as soon as any TA occurred. The Easyjet procedure was not properly followed but the Air France one was.

The Investigation noted that the Training Captain in command of the Easyjet aircraft had a total of five years command experience with the Company, two of those as a Training Captain. The trainee Captain he was with had over 3000 hours on type but no previous multi crew command experience. The Air France crew had substantial experience in general and on type in proportion to their rank. The trainee controller was “in training for their first controller qualification” and the OJTI had been unit qualified for 9 years and had acquired an instructor endorsement.

The radar unserviceability which had been the cause of the procedural service which had prevailed when the incident had occurred was found to be a long running issue which had been subject to NOTAM and was especially well known to crews based at Base-Mulhouse such as the Easyjet Switzerland crew involved. However, despite a recent history of prolonged area radar outage at the Unit, it was found that neither the OJTI nor the trainee controller had controlled operationally using procedural service prior to the beginning of the shift cycle in which the incident occurred. It was also noted that “nothing restricts instruction being undertaken in a procedural control situation”. It was considered that the selective activation of the monitors and the viewing on them of “an uncertain radar image” which was occurring at the time of the incident and which was not the subject of any procedure, “may have helped create confusion about the positions of the aircraft”.

Workload on the APP frequency at the time was low. In the hour prior to the conclusion of the incident, just ten inbound and four outbound aircraft had been worked with a maximum of four on being on the frequency at the same time just twice during the hour. When the Air France aircraft was cleared to climb to FL110, there were two other aircraft on frequency.

The Investigation “sought to identify the different types of RA leading to erroneous crew inputs, in particular ‘monitor vertical speed’”. However, it was found that neither operator used flight data algorithms which would enable the detection of contradictory corrections for this particular RA.

Interviews with the controllers and pilots involved led to the following findings:

• The trainee controller said that he “had cleared the crew of AF7343 to climb to FL110 thinking he had said FL100” and “did not pay attention to the crew read back as he was thinking about the strategy to adopt for a second incoming aeroplane that was descending to FL120”.

• The Training Captain in command of the Easyjet aircraft said, in relation to his response to the RA sequence, that:
  • he did not intentionally make a nose-up input from level flight when the first (preventive) RA, ‘monitor vertical speed’, which required no alteration to the flight path, occurred.
  • He felt that the occurrence of an RA of whatever type may provoke in him a need for a reaction due to the implicit risk of collision and anxiety generated by the aural and visual alarms.
  • due to the speed of the sequence and his state of stress, he disabled the AP but did not, as required, disable the FDs or ask the PNF to disable them. For this reason, being aware that the FDs were still engaged and the possible consequence on the management of thrust on the A/T, he decided to disable it and operate the thrust levers manually during the various resolution manoeuvres.
  • Being aware of the position of the conflict on his right, he “voluntarily made a left turn in order to increase the separation”. He reported having seen AF7343 during the turn.
  • He feared a collision during the incident.
  • He “did not think about safeguarding the CVR” after landing at Basle.
• The trainee Captain on the Easyjet aircraft said that:
  • The series of powerful vertical accelerations, including the negative acceleration, made him lose sight of the instruments.
  • He considered that he had been influenced by the situation and was surprised by various factors such as the reversal of the RA or the left turn made by the PF.
  • In retrospect, he wondered whether he ought to have intervened, in particular to moderate the extent of the manoeuvres.

Analysis of all the available evidence led to a number of observations in respect of the TCAS RA sequence and the reversal which occurred in particular:

• The excessive rate of climb as the Air France aircraft neared FL110 may have had an influence on the occurrence of the TA on the one hand, and on the interval between the TA and RA on the other but it was not the cause of the reversal of the RA.

• When the first RAs occurred, both the decrease in vertical speed of AF7343 in accordance with the instruction from the controller (maintain FL100) and the increase in vertical speed of DS1058 due to the inappropriate nose-up input of the PF “were contrary to the initial resolution logic of the TCAS” and “although they were short, they may have played a part in reaching the thresholds at which a reversal became necessary”.

• The left turn made by the Easyjet PF during the RA sequence was at variance with the principle of vertical separation on constant tracks on which TCAS is founded.

• Given that the Easyjet Training Captain had “received the training required to carry out his duties as a Captain and an Instructor” and “was experienced and competent in terms of the checks he had performed” it was considered that “the incident shows that these criteria do not guarantee a consistently high level of performance, especially when the pilot is subject to stress”.

• It is likely that simulator-based TCAS training does not realistically reflect operational situations such as those encountered during this event.

The formally-recorded Conclusion of the Investigation was as follows:

• The loss of separation that characterised this serious incident was due to an error in speech by the trainee controller, who cleared AF7343 to climb to FL110, a flight level to which he planned to clear the crew of DS1058 to descend and to which he actually cleared them shortly thereafter, and the non-detection of that error by the instructor controller.
• The following may have contributed to the error in speech and to the fact it was not detected:
  • the safety study and rapid implementation of CPUs that were unserviceable at the time of the serious incident;
  • the unusual situation for each of the two controllers, combining conventional control procedures, the use of a radar declared inoperative and weather avoidance requests;
  • the contact role undertaken by the instructor controller between the trainee controller and the coordinator controller was not conducive to the supervision of the trainee controller.
• It is possible that the vertical speed of AF7343 played a role in the sequence in which the TA and then the RA occurred.
• The worsening of the loss of separation, evidenced by the reversal of the TCAS RAs, was due to the conjunction of:
  • the tendency to decrease the rate of climb of AF7343 further to an instruction from the trainee controller asking the crew to maintain FL100, given prior to the triggering of the TCAS RA «maintain vertical speed, crossing maintain», inviting the crew on the contrary to maintain a constant rate of climb;
  • brief nose-up input by the PF of flight DS1058 when the AP was disengaged after the TCAS RA «monitor vertical speed» inviting the crew not to climb was issued.
• The minor injury to a cabin crew member of flight DS1058 was due to the abrupt manoeuvres by the PF, at the time subject to increasing levels of stress in response to the successive TCAS RAs.
• The malfunctions of the radar display system were due to the mismatch between the equipment as it was configured and the traffic liable to be taken into account. Inadequate coordination between the services and the time constraints did not facilitate the detection of this anomaly during the safety studies carried out for changes made prior in the incident.

Four Safety Recommendations were made as a result of the Investigation as follows:

• that ICAO study the inclusion of resolution automation when autopilot is engaged as an ACAS standard. Such a study should specifically consider the direct and indirect effects of automation on the ability of crew members to remain aware of the situation and fly manually when necessary. [FRAN 2013 058]

• that ICAO study the impact on safety of classifying all resolution advisories (RA) of the preventive type as traffic advisories (TA) only. [FRAN 2013 059]

• that the DSNA [The French State ANSP] define a clear instruction on the use of radar imagery when it is uncertain and when procedural control is in use. [FRAN 2013 060]

• that EASA study setting a standard for aeroplanes’ smooth vertical flight paths when approaching a level selected by the crew. [FRAN 2013 061]

The Final Report of the Investigation was published in English on 12 May 2014.

Further Reading

• Loss of Separation
• Airborne Collision Avoidance System (ACAS)
• ACAS Resolution and Traffic Advisories
• Safety occurrences during on-the-job training
• ATC Operations in Weather Avoidance Scenarios
• Loss of Separation During Weather Avoidance
• TCAS II 7.1 - Reversal RA
• TCAS II version 7.1 - Overview for pilots
• TCAS Occurrence (OGHFA SE)
• ACAS: Guidance for Controllers
• ACAS X
• Incorrect use of TCAS Traffic Display
• Crew Resource Management
• Stress