If you wish to contribute or participate in the discussions about articles you are invited to join SKYbrary as a registered user

SH33 / MD83, Paris CDG France, 2000

From SKYbrary Wiki
Summary
On the 25th of May, 2000 a UK-operated Shorts SD330 waiting for take-off at Paris CDG in normal visibility at night on a taxiway angled in the take-off direction due to its primary function as an exit for opposite direction landings was given a conditional line up clearance by a controller who had erroneously assumed without checking that it was at the runway threshold. After an aircraft which had just landed had passed, the SD330 began to line up unaware that an MD83 had just been cleared in French to take off from the full length and a collision occurred.
Event Details
When May 2000
Actual or Potential
Event Type
AGC, GND, HF, RI
Day/Night Night
Flight Conditions On Ground - Normal Visibility
Flight Details
Aircraft MCDONNELL DOUGLAS MD-83
Operator Air Liberté
Domicile France
Type of Flight Public Transport (Passenger)
Origin Paris/Charles de Gaulle Airport
Intended Destination Madrid-Barajas Airport
Flight Phase Take Off
TOF
Flight Details
Aircraft SHORT 330
Operator Streamline Aviation
Domicile United Kingdom
Type of Flight Public Transport (Cargo)
Origin Paris/Charles de Gaulle Airport
Intended Destination London Luton Airport
Flight Phase Taxi
TXI
Location - Airport
Airport Paris/Charles de Gaulle Airport
General
Tag(s) Flight Crew Training
Aircraft-aircraft collision
Inadequate ATC Procedures
Airport Layout
Inadequate Airport Procedures
Copilot less than 500 hours on Type
AGC
Tag(s) Phraseology
Language Clarity
Multiple Language use on Frequency
HF
Tag(s) ATC clearance error
ATC Unit Co-ordination
Ineffective Monitoring
Plan Continuation Bias
Procedural non compliance
GND
Tag(s) Aircraft / Aircraft conflict
ATC clearance error
Both objects moving
RI
Tag(s) ATC error
Incursion pre Take off
Ground Collision
Phraseology
Outcome
Damage or injury Yes
Aircraft damage Major
Injuries Few occupants
Fatalities Few occupants (1)
Causal Factor Group(s)
Group(s) Aircraft Operation
Air Traffic Management
Safety Recommendation(s)
Group(s) Air Traffic Management
Airport Management
Investigation Type
Type Independent

Description

On the 25rd of May, 2000 a Shorts SD330 (G-SSWN) being operated by Streamline Aviation on a scheduled cargo flight from Paris Charles de Gaulle to London Luton as 'Streamline 200' and about to depart was involved in normal night visibility in a high speed collision on runway 27 with an MCDONNELL DOUGLAS MD-83 (F-GHED) being operated by Air Liberté which was departing Paris on a passenger charter flight to Madrid as 'Liberté 8807'. Both aircraft sustained significant damage but there was no post-impact fire. The casualties were the two pilots of the SD330, one of whom was killed and the other seriously injured as a direct result of the impact of MD83’s left wing with the SD330 flight deck.

Investigation

An Investigation was carried out by the French BEA. The MD83 was equipped with a FDR and a CVR but the SD330 was fitted only with a CVR - there was no requirement to fit an FDR. Data from the available recorders were successfully downloaded for use by the Investigation. ATC recorded data were also available. A Preliminary report containing the initial factual information available at that time was published on 30 June 2000.

It was found that the SD330 flight crew consisted of a 41 year-old Line Training Captain with 2,440 hours total flying experience including 1,005 on type who was overseeing a 43 year-old First Officer newly qualified on type for which the accident flight was his 6th line training sector with the same Captain and his third duty on the night return flight from London Luton to Paris CDG. This First Officer had 4,370 hours total flying experience which included 14 hours on type. The MD83 flight crew consisted of a 55 year-old Captain who had 11,418 hours total flying experience and had been type rated on the MD83 for three years and a 47 year-old First Officer who had 11,104 hours total flying experience and had been type rated on the MD83 for 9 months.

At the time of the accident, the ATC function for the airport was grouped at a single control position as follows:

  • a Delivery Position occupied by the 31-year old TWR Manager.
  • a 36 year-old TWR controller who was an ENAC instructor on his sixth day of re-qualification experience who had originally qualified in the position 9 years earlier before transferring to ENAC about 18 months prior to the accident.
  • a 29 year-old GND controller who had first qualified in that position about 3½ years prior to the accident.
  • an Approach/Departure Position "occupied by the APP controller(grouped with the IFR room)".

Although all three runways were in use, all traffic was being controlled by this team from 'north' positions. In addition, because of advance awareness of an abnormally busy late evening period due to spectators returning to Spain after a major football match in Paris, the Head of ATC was also present in the TWR and seated between the TWR and GND control positions passing strips between them.

It was noted that work was under way for the construction of a fourth runway to the north of and parallel to the runway where the collision occurred - runway 09/27. The working area was outside the runway protected area. Weather conditions were good and were assessed not to have been a factor in the collision. Runway 09/27 was active in a westerly direction and traffic was being controlled from an 82 metre-high TWR which was located 1,800 metres from the runway threshold and 1,450 metres from the intersection of the runway and taxiway 16. Both the threshold of the runway and its intersection with taxiway 16, which were 1,730 metres apart, were at an altitude of 117 metres and between these two points, the runway surface was slightly concave to a low point of 114 metres. As shown in the diagram below, on the night of the accident, taxiways 16.1, 17.1, 17 and HP1 were closed and clear of taxiway B1, and work was in progress to construct taxiway HP2.

The taxiways serving runway 27 departures showing those out of service (X) [Reproduced from the Official Report]

It was established that the MD83 had been cleared to taxi to the holding point for 27 almost half an hour before the SD330 but although it had taxied as cleared, its take-off had then been delayed whilst the crew taxied clear and transferred back to GND to resolve a technical problem. Whilst with GND, that controller noticed that the aircraft was on a taxiway which led to the southerly departure runway 26 and proposed that runway for departure and altered the MD83 strip. Ten minutes later, the MD83 was ready to depart but requested 27, which was agreed and the aircraft transferred back to TWR. The corresponding strip, which had been altered by the GND controller to read '26' with the associated SID was only partially corrected so that the TWR controller initially thought that the aircraft was now departing from 26 and cleared it to do so. The crew then advised that they were at the 27 Holding Point and were cleared to line up and wait on runway 27 behind a landing 737. Three seconds after this clearance, GND transferred the SD330, which had requested and received approval for departure from taxiway 16 to the TWR frequency. The TWR controller "did not notice" the circled '16' annotation on the strip. Just over two minutes after the transfer to listen out on TWR, with the 737 having meanwhile landed and passed in front of the SD330 waiting on taxiway 16 before clearing the runway, TWR cleared the MD83 (in French) for take-off. Five seconds after this, the SD330 was instructed by TWR (in English) to "line up runway 27 and wait, number 2". On receipt of this clearance, its crew then began to taxi onto the runway "whilst looking for the Number 1" just as the MD 83 approached the intersection at high speed. Seconds before impact, the SD330 Captain noticed the MD 83 anti-collision beacon and braked. At about the same time, with the aircraft about to rotate, the MD 83 crew reported noticing the SD330 "on the edge of the runway". With no time to take avoiding action, the left wing of the MD 83 collided with the right propeller and cut through the SD330 flight deck. The MD 83 rejected its take-off and informed TWR that they had just hit another aircraft. The ground tracks and timings for both aircraft are shown in the diagram below.

The trajectories of both aircraft (red = MD83 , buff = Shorts) [Reproduced from the Official Report]

The damage to both aircraft was significant - see the illustrations below. The flight deck of the SD330 was partially destroyed, its First Officer was killed outright as the MD83 wing penetrated the flight deck from the right and the Captain was seriously injured. No fire resulted from the impact and none of the 157 occupants of the MD83 were injured.

The SD330 showing flight deck damage caused by the impact of the MD83 outer wing [Reproduced from the Official Report]
The MD83 showing impact damage to the outer leading edge of the left wing [Reproduced from the Official Report]

It was noted that the sight line from the SD330 flight deck whilst at the holding point would have precluded any view of the beginning of the runway and that this had only changed when the aircraft had been manoeuvred slightly to the right in preparation for line up. It was also noted that although the MD83 had understood the line-up clearance given in English to the SD330, it was not possible for them to realise that this aircraft was further down the runway rather than in line behind them.

The way in which the TWR controller had formed and continued with his incorrect mental map was considered during the Investigation. It was noted that:

  • There had been no verbal coordination between the GND and TWR controllers when the GND controller agreed that the SD330 could taxi to and await take-off clearance - as it often did - from taxiway 16 although the strip was so annotated.
  • In accordance with normal procedures, GND made no further mention of aircraft position when transferring either aircraft to TWR and in both cases this was done, again as permitted by procedures, by instructing the aircraft to change frequency and await a call from the TWR controller.
  • Given that he believed that the work to construct new taxiway HP and the associated taxiway closures meant that taxiways 16 and 17 "would not be in use", the TWR controller was functioning on the basis of "an erroneous perception of the situation […] according to which all aircraft having to take off from runway 27 were directed towards the runway threshold”. It was considered that the fact that there had been no TWR team briefing - which in any case procedures did not require - had "facilitated this erroneous perception". Nothing subsequently disturbed this false mental picture.
  • A direct visual check was difficult to perform because of the two areas of construction works and consequent light pollution and verification using ground radar was difficult because of the particular nature of the display. In any case "bearing in mind his mental picture of the situation, there was no reason for his attention to be drawn to taxiway 16".
  • During radio communications between TWR and the SD330, there was no mention of its position, nor did procedures require this.
  • The previously quoted words used by the TWR controller to instruct the SD330 to line up were predicated on his mistaken belief that the aircraft was at the threshold holding point behind the MD83.

The Investigation found that "the organisation of ATC services at Paris Charles de Gaulle is defined by an Operations Manual that is not updated regularly and by a large number of unorganised internal memos". It also found that runway incursions at Paris CDG were relatively frequent - 20 had been recorded in the first 6 months of 2000 many of which evidenced recurring types of events:

  • aircraft cross the runway or go past a holding point,
  • aircraft line up in front of an aircraft on take-off or on final instead of lining up

behind,

  • there is a confusion in call-signs which leads to one aircraft moving other than that which the controller wished to call.

It was considered that these events "demonstrate that it is sometimes difficult for crews to identify where they are and so arrive at a predetermined point". Recurring features were found:

  • Holding points are not always identified.
  • The terminology used is sometimes ambiguous when several call signs resemble each other and no position information is given.
  • In particular, one aircraft lining up in front of another aircraft which is taking off is not exceptional, collisions being avoided because the crews notice each other in time - being able to see is a crucial factor in saving the situation which was absent in the case of the investigated accident.

In addition, a general lack of awareness amongst the Paris CDG controllers of such incidents and the lessons which could be learned from them was identified as was the fact that "no organised (incident) feedback exchange system exists between different French aerodromes, even less with foreign aerodromes". It was considered "regrettable that the absence of such a system [means that an] objective comparison of the advantages and disadvantages of the use of a single language, for example, is not possible". It was concluded that "this rather underdeveloped feedback system, associated with a lack of established rules for teamwork, results in a non-optimal use of the [safety-related] information available".

The formal determination of Probable Causes was as follows:

  • The TWR controller’s erroneous perception of the position of the aircraft, this being reinforced by the context and the working methods, which led him to clear the SD330 to line up.
  • The inadequacy of systematic verification procedures in ATC which made it impossible for the error to be corrected.
  • The SD330 crew not dispelling any doubts they had as to the position of the “number one” aircraft before entering the runway.

It was also formally determined that Contributory Factors included the following:

  • Light pollution in the area of runway 27, which made a direct view difficult for the TWR controller.
  • Difficulty for the TWR controller in accessing radar information: the 'ASTRE' image was difficult to read and the 'AVISO' image was not displayed at his control position.
  • The use of two languages for radio communications, which meant that the SD330 crew were not aware that the MD 83 was going to take off.
  • The angle between access taxiway 16 and the runway which made it impossible for the SD330 crew to perform a visual check before entering the runway.
  • The lack of co-ordination between the TWR and GND controllers when managing the SD330, exacerbated by the presence of a third party whose role was not defined.
  • An occurrence feedback system which was recent and still underdeveloped.

A total of 14 Safety Recommendations were made as follows:

  • that Aéroports de Paris and the DGAC together study all of the procedures and associated means for the simultaneous use of two different parts of a runway so as to guarantee, in all circumstances, the same level of safety as when the runway is used by only one aircraft.
  • that Aéroports de Paris and the DGAC should ensure that terminology used in practice by the ground controller include the systematic identification of the holding point specific to the required taxiway during the instruction to taxi towards the runway.
  • that Aéroports de Paris and the DGAC should ensure that terminology used in practice by the aerodrome air traffic control include the systematic identification of the taxiway from which the aircraft must line up.
  • that Aéroports de Paris and the DGAC should ensure that terminology used in practice by the aerodrome air traffic control systematically include, where a clearance is issued to line up behind a departing aircraft, the formal and unambiguous identification of said aircraft.
  • that Aéroports de Paris and the DGAC should ensure that the procedure for sequential line-up be defined, as well as conditions for its application.
  • that Aéroports de Paris and the DGAC should ensure that the control positions at Paris Charles de Gaulle aerodrome be equipped with the latest ground radar equipment.
  • that Aéroports de Paris and the DGAC should ensure that the use of high speed exits for line-ups be subject to the existence of arrangements which guarantee a level of safety equivalent to visual checks performed by the crew.
  • that the DGAC, in the light of the analysis of this accident and previously acquired experience, study the expediency and methods of implementation for the systematic use of the English language for air traffic control at Paris Charles de Gaulle aerodrome, as well as the extension of this measure to other aerodromes with significant international traffic.
  • that the DGAC establish a precise definition and a verification procedure for Control Tower Operations Manuals.
  • that the DGAC study the implementation in ATC practices in respect of systematic checking procedures.
  • that the DGAC accelerate and systematise the implementation of an ATC resource management training course, specifically taking into account questions related to co-ordination.
  • that the DGAC should ensure that the functions of a controller assistant (role, prerogatives and possible manning of the position) are defined.
  • that the DGAC should ensure that it is only permissible for the Tower Manager to fulfil a control position where this remains compatible with his other responsibilities.
  • that the DGAC should ensure that procedures for releasing controllers to duty are defined so as to exclude this being dependant on self-assessment.

The Final Report in an Official English translation and the definitive version of the Report in French were published in June 2001.

Related Articles