On 23 November 1997, a McDonald Douglas MD 83 being operated by French Airline AOM Minerve on a scheduled passenger flight from Marseille to Paris Orly made an unintended premature descent almost to terrain impact at 4nm from the destination runway in day Instrument Meteorological Conditions (IMC) before a go around was commenced. A subsequent approach was uneventful and a normal landing ensued. There was no damage to the aircraft or injury to the occupants.
An Investigation was carried out by the French Bureau d'Enquêtes et d'Analyses (BEA). It was found that co pilot line training had been in progress over a two day period using an experienced instructor pilot as pilot in command whilst two newly type-qualified co pilots alternated between the right hand seat as operating co pilot and the observer seat. For the incident approach, the aircraft commander had been PF, fog prevailed at destination and an Instrument Landing System (ILS) approach to Runway 07 was accepted with radar vectoring to the ILS LLZ and an autoland planned.
The detailed sequence of events was established using a combination of Flight Data Recorder (FDR) data, ATC recordings and flight crew interviews. The aircraft commander as PF selected track 258° on NAV 1 instead of the correct 065° but the co-pilot did not notice the error. From a closing heading for the ILS LLZ of 020° level at 3000 feet and 160 KIAS, capture occurred and was announced by PF who, looking at the mis-set HSI, brought the heading indicator to the tail of the ILS LLZ track. He subsequently stated that he thought at that time that he had brought the heading indicator to the head of the arrow and not to the tail. He then noticed that the heading indicated 078° while he expected the LLZ track for Runway 07 of 065°. Thereafter, the PF did not announce the actions he took relative to the automatic systems and mis-management of FD Mode selections resulted in initial failure to descend on the ILS GS and a deviation left of the LLZ track. With the FD in VS and HDG modes, the aircraft descended through the ILS GS. At a height of 916 feet agl, a GPWS/TAWS Mode 5 ‘Glideslope’ Alert occurred and shortly afterwards, the aircraft went from clear skies into the fog bank which covered the area. The Autopilot was disconnected and, at 783 feet agl, a Mode 2 "Terrain" warning occurred. The GPWS "Glideslope" Alert started up again as soon as the overidng "Terrain" warning ended. The PF attempted to bring the aircraft back onto the approach track and then re-engaged the Autopilot at 415 feet agl and armed the "autoland" mode. At 279 feet agl, a further GPWS "Terrain" warning occurred lasting nine seconds. At approximately 200 feet agl, the PF disconnected the Autopilot again and began a go around. As the transition to a climb was made, a minimum radio-height of 67 feet was recorded at the ILS Outer Marker situated 4nm from the landing runway threshold. The co-pilot later stated that he saw the ground and read a radio-height of about 50 feet. The go around was continued into a re-positioning circuit under radar vectoring and a further approach resulted in a successful autoland on Runway 07.
On the subject of ‘Crew Behaviour’ the Investigation found that:
“The Captain acted alone. The co-pilot was aware of a problem in the vertical plane because he announced the glideslope track and looked outside. The first officer in the observer seat was also aware of an abnormal situation. However, neither of them intervened. Several factors contribute to an explanation of inadequate crew performance:
- The two First Officers in training had an MD83 type rating but had not yet finished their LOFT.
- They had very little experience, unlike the Captain, which generated a passive attitude on their part.
- The functions of the trainee first officer in the observer seat were not clearly defined, either by the airline or by the Captain. She neither had the role of backup co-pilot, nor the competence to perform the task. She had, however, performed this role for another airline on a different type of airplane.
During the approach, the two first officers on LOFT behaved like students facing an instructor. They were unaware of their responsibilities in the conduct of a commercial flight. Their attitude was passive, although the Captain was demonstrating a landing in poor visibility. The context of a learning situation inhibited Crew Resource Management. They never had the feeling of being in any danger.
In conclusion, the Captain had to perform the roles of captain, co-pilot and instructor during the incident flight. There was no check of the Captain’s actions or decisions, nor any mutual crosschecking. Neither the role of co-pilot responsible for safety nor the role of First Officer were filled on board the aircraft, which is unacceptable from the point of view of the safety of a passenger-carrying flight.”
On the subject of Fatigue, the Investigation found that:
“the Captain had built up a high level of fatigue due to the number of flights as instructor, the simulator sessions, work on the ground and exceeding the regulatory number of flying hours. This apparently contributed to the ILS heading selection error and diminished his performance when faced with a situation requiring a high workload.”
The aircraft was found to have been serviceable.
It was noted that Minimum Safe Altitude Warning (MSAW) was not installed at Paris Orly at the time of the event and that there were no ATC procedures requiring any monitoring of aircraft established on ILS approaches.
It was determined by the Investigation that “the incident resulted from the decision to put the airplane into descent when, as a result of a display error, it was neither on the localizer track nor on the glide path, and with no context defined for this improvised manoeuvre. The importance that (the Operator) attached to accelerated training given to new co-pilots and to undertaking commercial flights contributed directly to the incident.” Other contributory factors were found to be:
- the aircraft commander’s fatigue;
- the imbalance in the flight crew, made up of a very experienced instructor and an under-trained co-pilot, which led to the abrupt disappearance of teamwork and procedures the moment the workload increased;
- aircraft warning system ergonomics and a fault in the automatic pilot system.
A large number of Safety Recommendations were made during and upon the conclusion of the Investigation. Not all were specifically targeted. A preliminary report issued during the course of the Investigation contained six Safety Recommendations:
- That Jeppesen make its Orly approach chart compatible with the official French chart.
- That Météo-France remind forecasters that they must amend TAF’s whenever a threshold amendment is forecast.
- That the DGAC ensure that pilot training clearly demonstrates that the majority of METAR’s contain trends, which are meteorological forecasts valid for two hours, and can be used for the preparation of flights of short duration,
- That the DGAC remind crews that in case of uncertainty or of
contradiction between different parts of the meteorological file, they can obtain direct information from the meteorological centre.
- That the DGAC ensure that airlines possess information which allows them to modify flight planning so as to avoid pilots exceeding the
statutory work time.
- That the DGAC' accelerate installation of the MSAW system and prioritise aerodromes with heavy traffic.
A total of 18 further Safety Recommendations were issued at the conclusion of the Investigation as follows:
- That, in liaison with the JAA and the FAA, the DGAC modify the certification regulations so as to ensure a better balance in the presentation of horizontal and vertical position data on new generation aircraft.
- That the criteria for performance of flights under supervision guarantee the effective presence of an additional pilot trained in supervision.
- That (the Operator)ensure that flights counted by pilots in training are really performed as members of the crew.
- That the regulations on flight crew work time take into account all aspects which cause fatigue.
- That the Manufacturer takes immediate steps to warn operators of MD83’s that, in certain circumstances, the active modes displayed on the FMA can differ from those which the airplane is in fact engaged in at that moment.
- That the FAA, in liaison with the DGAC, immediately require modification of the MD83 so that the active modes displayed on the FMA indicate what the airplane is in fact doing at that moment.
- That the DGAC ensure that the French and European certification regulations have clear specifications concerning ILS capture.
- That the certification requirements be modified so that certification takes into account the overall management of alarms in the cockpit;
- That the implementation of (the immediately previous) recommendation, along with Recommendation 44.3 in the Report into the Mont Sainte Odile accident of 20 January 1992, be made a priority by the DGAC and the JAA. (Editors Note: The Mont St Odile Report and its Safety Recommendations were not published online in English translation.)
- That the DNA study the possibility of presenting controllers with vertical airplane positions expressed in altitude when an airplane is below the transition level.
- That the DGAC significantly increase the number of in-flight inspections, particularly in case of a major increase in an airline’s activity.
- That the DGAC establish a scaled range of penalties for problems uncovered to allow systematic, rapid and appropriate measures to be taken
- That the DGAC forbid the extension of time limits set for the correction of problems identified.
- That the DGAC modify the procedure for supplying meteorological information when the latter is less reliable so as to ensure that crews are informed of the exact visibility at the moment they must decide whether to continue an approach.
- That Aircraft Operators ensure that documentation given to crews is valid and, where necessary, draw the crew’s attention to differences or errors identified.
- That the recording of images from the cockpit instrument panel on protected recorders be required, the images being synchronized with those of other mandatory recordings.
- That the DGAC install some means of reconstituting images as seen on a radar screen.
- That Aircraft Manufacturers ensure that the denomination of parameters on flight recorder decoding grids correspond to the parameters really recorded.
The Final BEA Report of the Investigation may be seen in full at SKYbrary bookshelf: BEA Report f-mc971123a (English)