1.16.4.2 Warning lights

Examination of the bulbs of the warning lights did not show whether they were lit at the moment of impact.

1.16.4.3 Flight Instruments and equipment

Most of the equipment and flight instruments could be examined in the laboratory. Although blackened by the fire, they were, in general, in good external condition with few deformations. The values noted essentially confirm the values found on the site, though this does not necessarily prove the relevance of the display. Manipulation of the instruments had, however, modified certain values that were therefore discarded.

• EPR left engine (needle) 0.985 (flagged)
• EPR left engine (index) 2.176 (flagged)
• EPR right engine (needle) 1.0 (flagged)
• EPR right engine (index) 2.178 (flagged)
• Fuel flow meter left engine 5,400 kg/h
• Right engine fuel flow meter < 200 kg/h
• Left tank fuel quantity 4,400 kg
• Center tank fuel quantity 820 kg
• Right tank fuel quantity 4,400 kg

Note: the quantities read correspond to a weight of 9,620 kg. By adding taxiing allowance (200 kg) and the fuel consumed on take-off (225 kg), on the basis of nine tons per minute, a value of 10,045 kg is reached, consistent with the weight of on-board fuel.

• EGT left beyond the upper limit
• EGT right 0
• N2 left Indicator zero on principal and secondary scales
• N2 right Indicator 22 on principal scale, 2.5 on secondary scale
• N1 Indicator left zero on principal and secondary scales
• N1 right Indicator zero on principal and secondary scales.
• Electrical altimeter setting at 1016 mb
• ADF RMI left heading of 005°; the two indices set at 270°

1.16.5 Airplane performance on take-off

1.16.5.1 Performance calculations

A calculation of performance was carried out in the United States (cf. appendix 5). Its goal was the calculation of certain parameters not recorded by the FDR, for example pitch, roll, and engine thrust, by correlating the following various data:

• aerodynamic, kinetic and propulsion characteristics of the B737-200;
• the CVR data (noise from the stick shaker, from the GPWS alarm and from the sharp thumping noise);
• FDR parameters;
• observations made on the wreckage (position of the flaps, of the landing gear and impact marks on the ground);
• the weight, balance and thrust at take-off and the airplane's speeds (V1, VR and V2);
• the meteorological conditions on the day.

The results of these calculations showed:

• a loss of airplane speed accompanied by a continuous nose-up pitch attitude, leading to a stall situation;
• a significant reduction of thrust on the right engine a few seconds after the loss of power on the left engine. This greatly reduced the performance of the airplane during initial climb;
• that with one engine out, and with a constant V2 (150 kt), the maximum vertical speed in the climb of the airplane is around 150 feet per minute with the landing gear down. This vertical speed rises to around 450 feet per minute when the landing gear is in retracted position.

1.16.5.2 Flight simulator session

At the request of the Commission of Inquiry, and in the presence of some of its members, several tests on a flight simulator were carried out by a qualified crew using the conditions on the day of the accident. These tests were run with a left engine failure at rotation speed. During the first test, the crew was asked to follow standard procedures. During the second and third tests, the crew was asked to keep the landing gear down, to maintain the speed at V2 in one case and to maintain a constant pitch of 18° in the other.

From this session the following results emerged:

• a change in the magnetic heading by about fifteen degrees towards the left after correction by the crew;
• vertical climb speed between 1,000 and 2,000 ft/min when the pitch is maintained at 18° from the moment of loss of power in the left engine and with the landing gear down;
• vertical speed fluctuating between - 300 and + 300 ft/min, when the V2 speed is held constant;
• stick shaker activated between 135 and 140 kt;
• difficulty for the pilot to take the controls during the initial climb with one engine out, especially after a simulator pause comparable to a transfer of control in a critical situation.

1.17 Organizations and Management

1.17.1 Crew on Boeing 737-200 by Air Algérie

1.17.1.1 Typical type rating training course

The course has in three main phases that cover:

• theoretical aspects;
• simulator sessions (Full Flight Simulator - FFS);
• in-flight sessions.

The first phase is completed by groups of about ten people and lasts eleven days. It is made up of sixty-five hours of lectures based in particular on the utilization of training documents. These theoretical courses include:

• study of airplane systems;
• study of both normal and emergency procedures;
• aspects of CRM;
• aspects of safety and rescue.

Visualization of the layout of the controls is supported by visits to the simulator. The study of each system is completed by a debriefing by an instructor responsible for technical follow-up of the Boeing 737-200. This debriefing allows for significant points to be clarified and checks the level of knowledge acquired. The final two days are dedicated to aspects of CRM and to aspects of safety and rescue.

The second phase is open to trainees who have passed the relevant exam in the theoretical section. It is made up of fourteen sessions of four hours each in the FFS simulator, carried out as a crew under the direction of a pilot-instructor, preceded by a briefing in the auditorium of about two hours and followed by a debriefing. These sessions concern basic flying and flying on autopilot. Students study the flight envelope limits, various failures and emergency situations.

At the conclusion of these FFS sessions, a final exam lasting four hours is taken that covers the different types of problems and failures studied during the fourteen sessions.

In the context of this type rating training course, engine failure is tackled in the sixth session. Failures are generally simulated around V1 speed when the plane is rolling for takeoff. Three types can be simulated: engine flame out, damage to the engine (engine severe damage fire) and a surge.

Note: the procedure to follow after an engine failure on take-off after V1 is included in appendix 6.

The third phase (called base training) involves three flights of one hour and thirty minutes each followed by a test flight of one hour and thirty minutes, which leads to type rating certification on this model. Engine failure after V1 is included in the second and third flights as well as for the test flight.

Air Algérie' s program is based on a total of thirty-one days training (including exams).

Following successful type rating, a minimum of six legs is required on line flight training before the test flight.

1.17.1.2 Recurrent training and checks

Regulatory requirements concerning recurrent training and checks are: an annual recurrent course, two annual line checks carried out during a commercial flight and related to flying skills and application of Operations Manual requirements, an annual base check (on simulator) with emphasis on type rating exercises.

Air Algérie has included, in a single module, a recurrent training course and a base check. This takes place once per year. The program for this training course was approved by the DACM. The objective of this training course is to update pilots' knowledge of the Boeing 737-200, as well as their know-how in crew task-sharing. This sometimes provides the opportunity to take into account incidents that occurred during operations, and to revise both normal and emergency procedures.

The training course lasts three days. The first is dedicated to lectures in the auditorium, and the following two to the simulator: Two sessions of four hours are preceded by a two-hour briefing. The first is carried out on the FFS simulator, in the presence of a pilot-instructor from the airline, and focuses on the regulation type rating exercises. The second session consists of a simulated LOFT-type flight at the end of which a recurrent training certificate is issued.

1.17.1.3 Crew training

The two pilots' training and recurrent training files were examined by the commission. Their last exercise in engine failure during takeoff shows:

• for the Captain, 13 January 2003. This was an exercise involving an engine fire after V1 (case of a fire being brought under control after firing two extinguishers);
• for the co-pilot, 4 November 2002. The instructor had noted that it was necessary to call out the check-list and organize the fire fighting actions.

1.17.2 Feedback system

Since 1999, Air Algérie has had a Flight Safety Office that publishes safety bulletins for crews. This office provides Aviation Safety Report forms in order to get feedback, these being available on board planes. It is also possible to inform them confidentially of concerns. The office was recently equipped with a flight analysis station dedicated to Boeing 737-600 and -800; this analysis is not yet systematic or generalized to the whole Air Algérie fleet.

The first flight safety bulletin was published in August 1999. Two or three bulletins are published every year. Each issue includes studies of accident cases or incidents based on the example of other airlines. Of the twelve issues, only five occurrences that occurred at Air Algérie were analyzed:

• the first (a British investigation report) concerned an accident that occurred on 21 December 1994 at Willenhall (Coventry) to a Boeing 737-2D6C Cargo registered 7T-VEE;
• the second related to an accident to a Boeing 727 registered 7T-VEH concerning a lateral runway excursion that occurred on 31 January 1999 at Constantine airport;
• the third concerned runway excursion by a Boeing 737-200 analyzed by a national commission of inquiry;
• the two others concern technical incidents (the beginnings of a fire in the landing gear on an A310 and an engine oil leak leading to a return to the airport).

All of the case studies described since August 2000 concern foreign operations. No event reported by an Air Algérie crew was analyzed in a written fashion.

It appears that at the time of the accident, the system for providing feedback did not develop completely or systematically, in writing, events occurring within the airline. Such systematic investigation would allow crews to benefit from the experiences of other crews reporting unusual situations, with the goal of preventing them from happening again.

2 - ANALYSIS

2.1 Accident Scenario

2.1.1 Flight preparation and takeoff roll

Flight preparation was carried out by the co-pilot alone, as the Captain arrived late. It should be noted that the Chief Flight Attendant was present in the cockpit, a fact that contributed to the checklist that the co-pilot had started being cut off after the callouts for the selected takeoff speeds.

This lack of rigor in the preparation of the flight was also noticeable during taxiing, with the failure to observe the spirit of a "sterile cockpit". More generally, the CVR shows that at no time was there any dialogue or briefing related to a possible anomaly during takeoff, as required in the Air Algérie procedures. The crew was not apparently sufficiently psychologically prepared to face any possible problems that might occur.

Taxiing was immediately followed by engine power-up with an EPR of 2.18, which corresponds to the use of maximum thrust with the air conditioning switched on.

2.1.2 Failure of left engine during takeoff

Engine power up, airplane acceleration and the standard callouts were all carried out normally until rotation. The problems started suddenly, at the exact moment landing gear retraction was announced. All evidence gathered by the investigation, especially the airplane parts found on the runway near the place where the airplane lifted off, as well as the twelve degree change of heading to the left, show that the crew was then faced with a contained left engine burst.

Examinations showed that the problem originated in the turbine high-pressure stage located just after the combustion chamber. The rupture of a relatively large part caused the immediate destruction of the high-pressure turbine and the subsequent damage to the low-pressure turbine. The burst caused a sharp drop in the rotation speed and in engine thrust, without causing it to shut down.

Apart from the airplane' s tendency to yaw to the left, the damage to the left engine would normally result, in the cockpit, in a sudden reduction in the performance readings for that engine and a visual oil pressure warning, though without a fire alarm.

2.1.3 Management of the engine burst during initial climb

2.1.3.1 Crew coordination

The crew, which had been in a routine flight situation, was suddenly confronted with an emergency situation that required high levels of alertness, coordination and concentration, a situation for which, as previously noted, they were not specifically prepared. In this context, the Captain did not read back the order to retract the landing gear and did not retract it.

There was no announcement of the failure, no mention of any possible visual warnings such as the one associated with the oil pressure. Only a few exclamations from the co-pilot were heard. Even the callouts related to flying the airplane (speed, climb rate, trim, etc) were not made. Thus, after the failure, there was no longer any formalization of crew action, nor was there any coordination whatsoever on the management of the problem encountered.

A further factor then entered the picture: after eight seconds, the Captain took over the controls of the airplane, without any request from the co-pilot or any use of the standard terminology for handing over the controls.

It is impossible to know what led to this decision:

• did the Captain notice that the co-pilot was having difficulties controlling the airplane, or did he at least believe so?
• did he prefer to take back control, reckoning that it was his responsibility as Captain?

It is not possible to know whether his taking over the controls (words recorded on the CVR) was not preceded by an input on the controls.

In any event, the absence of any coordination between the pilots at the moment when tasks were transferred meant the Captain had to manage an emergency situation, which he had not had time to analyze completely. For the same reasons, he did not rely on the co-pilot, whom he simply asked on several occasions to let go (of the controls). He likely encountered some problems in taking over control, given that he repeated the terms "let go" and "take your hand off", this continuing until the end of the recording. The co-pilot appears to have carried out the Captain' s orders by reading back in the affirmative and indicating her willingness to act (proposal to retract the gear, radio message to the tower with the hand mike) though without being sure of the role she was supposed to play. This may, for example, have resulted in her placing her hands on the control column at the time of the stall warning alarms, which would explain the repeated requests from the Captain. She was thus not fully carrying out her role as PNF and did not monitor or at least call out the speeds as they were decreasing. The co-pilot' s offer to retract the gear was probably not even noticed by the Captain, due to his sudden excess workload.

2.1.3.2 Maintaining climb rate

Just after the rotation, the vertical speed calculated from the FDR data was between 1,400 and 1,800 feet per minute. This can be explained by the airplane' s initial speed, 160 knots, and the selection of a pitch attitude estimated at about 18°. Such a pitch angle corresponds to a climb with the engines operating nominally. The airplane' s speed fell progressively.

According to the airline' s procedures, the initial climb on one engine should be performed with the gear retracted, maintaining safety speed at V2 until the safety altitude. However, the landing gear was never retracted. With the conditions on the day of the accident (weight, aerodrome altitude and high temperature levels), the calculations and simulations performed show that it was difficult to maintain a positive rate of climb. Since the altitude was continuing to increase at the same rate, it was logical that there should be a progressive drop in speed.

Note: the airline' s procedure corresponds to a rate of climb of around five hundred feet per minute and a pitch attitude of about twelve degrees.

The Captain took over the flight controls in a critical situation while he was out of the loop with regard to flying. This probably led him to focus on a pitch attitude that was incompatible with one failed engine. His decision to take over the controls made it impossible for him to develop and supervise a strategy to adopt for the conduct of the flight.

The left engine failure coincided with the co-pilot' s request to retract the landing gear and disrupted that action. The co-pilot mentioned retracting the gear again but, at that moment, the task-sharing had been reversed. It was up to the Captain to ask for gear retraction. In the end, the gear remained extended until the impact.

The airplane' s aerodynamic performance subsequently deteriorated rapidly, especially as a result of the non-retraction of the gear, which added to the effect of maintaining a high pitch attitude. In addition, it should be emphasized that the high altitude of the aerodrome and the high temperature on the day, as well as a takeoff weight close to the maximum also contributed to limiting the airplane' s performance. During the initial climb, the high pitch attitude and the yaw induced by the failure of the left engine (which had the effect of increasing the drag at a critical moment), added to the factors previously mentioned.

2.1.4 Loss of control of the airplane

The high pitch attitude and the loss of speed put the airplane in a stall situation. About fourteen seconds after the noise and the "gear up" callout, the stick shaker (the device that warns the crew when 7% from the stall speed) began to operate intermittently, then continuously, until the end of the recording. The "don' t sink" aural warning told the crew that the airplane was dropping. The pitch attitude was apparently maintained until impact with the ground, as examination of the site showed. On the wreckage, the horizontal stabilizer trim was found in the position for a normal takeoff, which tends to support this hypothesis.

Examinations on the site show that the rear of the airplane hit the ground first in a stall situation with a high pitch attitude. Maintaining an excessive climb angle until the accident can be explained by the Captain focusing exclusively on this parameter ("refusing" the ground in a rocky environment making an emergency landing impossible).

The shutdown of the flight recorders before impact occurred while the airplane was in a stall situation. This shutdown is related to a loss of electrical power. The Commission was unable to identify the precise cause of this loss of electrical power, which occurred in an unusual situation a very short time before the impact.

2.2 Crew Task-sharing

The above scenario shows a lack of coordination, an absence of mutual control and deviation from procedures. The problems encountered were not subject to any formalized actions by the crew. The Captain' s takeover of the controls without agreement with the co-pilot, putting the latter out of the loop, occurred before the failure was identified. This sudden change destabilized task-sharing and thus crew teamwork. Management of the airplane' s track and flight strategy were severely affected.

2.3 Flight Analysis and Feedback

The existing regulations do not oblige operators to analyze flights systematically. No such analysis then existed for the Boeing 737. This may help to explain the lack of follow-up in the management of conversion documents. If it had not been possible to validate them, the flight parameters for the event might not have been usable, even though they were recorded on the FDR.

This absence of flight analysis can be assimilated with the non-systematic and unwritten nature of internal feedback. Formally registering the analysis of an event in written form allows all those who might be concerned to be informed in a coherent and standardized manner; it also allows a systematic record to be kept which will not be lost in the course of time.

3 - conclusions

3.1 Findings

• The airplane was certificated, equipped and maintained in accordance with the existing regulations.
• The airplane was airworthy. There were no deferred maintenance issues.
• The crew was trained and qualified in accordance with the requirements of Air Algérie and with the existing regulations.
• The co-pilot was pilot flying until the Captain took control.
• The weight and balance of the airplane were within the manufacturer' s weight and balance envelope.
• The Chief Flight Attendant spoke directly to the Captain during flight preparation and taxiing.
• The pre-start-up and pre-takeoff checklists were not performed. The emergency procedures were not mentioned.
• The power-up and acceleration were normal.
• Rotation was performed at about 150 knots, just above the planned speed.
• Immediately after the co-pilot' s request to retract the landing gear, the HP turbine on the left engine suffered a failure that led to a sudden loss of thrust and led the airplane to yaw. Landing gear retraction was not carried out, which increased drag significantly.
• Debris from the left engine was found on the runway near the area where the airplane lifted off.
• There was no communication between the Captain and the co-pilot concerning an emergency situation (identification of the nature of the problem).
• The Captain announced he was taking over the controls about eight seconds after the failure. In between times, the airplane' s excess yaw had been brought under control.
• Four seconds later, the co-pilot called the control tower to say that they had a problem.
• The rate of climb was maintained, the airplane' s speed dropped progressively until it reached stall speed.
• The maximum height reached by the airplane was about 400 feet above the aerodrome, about ten seconds before the end of the recording, the stick shaker being audible at that moment.
• Approximately ten seconds before the end of the recording, the "Don' t Sink" aural warning was heard.
• The FDR and CVR shut down before impact while the airplane was in a stall situation.
• The airplane' s initial impact with the ground occurred at a distance of about 1,645 m from the lift-off area, 12° to the left of the runway centerline.
• An intense fire broke out immediately. The airplane slid, losing various parts, hit and destroyed the airport boundary fence and then crossed a road before coming to a halt in flames.
• The control tower sounded the alarm. The airport fire service immediately headed towards the site of the accident.

3.2 Probable Causes

The accident was caused by the loss of an engine during a critical phase of flight, the non-retraction of the landing gear after the engine failure, and the Captain, the PNF, taking over control of the airplane before having clearly identified the problem.

The following factors probably contributed to the accident:

• the perfunctory flight preparation, which meant that the crew were not equipped to face the situation that occurred at a critical moment of the flight;
• the coincidence between the moment the failure occurred and the request to retract the landing gear;
• the speed of the event that left the crew little time to recover the situation;
• maintaining an inappropriate rate of climb, taking into account the failure of one engine;
• the absence of any teamwork after the engine failure, which led to a failure to detect and correct parameters related to the conduct of the flight (speed, rate of climb, configuration, etc.);
• the takeoff weight being close to the maximum with a high aerodrome altitude and high temperature;
• the rocky environment around the aerodrome, unsuitable for an emergency landing.

4 - RECOMMENDATIONS

The accident to 7T-VEZ brought to light failings in the area of task-sharing, joint control and handing over the controls.

Consequently, the Commission of Inquiry recommends that:

• Air Algérie, along with other operators, ensure that their CRM-type training programs effectively heighten crew awareness of the strict respect required for handover procedures and task-sharing.

The Commission of Inquiry recommends that:

• The Civil Aviation and Meteorology Directorate set up an organisation to inspect the application and conformity of the procedures in training programs for flight crews.

Considering that flight safety can only benefit from the input provided by a system of feedback entirely based on events that have occurred during operations, the Commission of Inquiry recommends that:

• The Civil Aviation and Meteorology Directorate ensure that Air Algérie and other operators set up a flight safety program that associates feedback and the systematic analysis of flight data.

Finally, considering the importance of a permanent dialogue on safety issues with a national feedback system, the Commission of Inquiry recommends that:

• The Ministry of Transport set up a permanent organisation for the investigation of civil aviation accidents and incidents.

APPENDIX 1

Decision No 283/cab/M of 6 March 2003

APPENDIX 2

Transcript of radio-communications

Remarque : les temps enregistrés sur la piste UTC présentent des variations semble-t-il aléatoires. Ils ne peuvent donc représenter une référence absolue.

APPENDIX 3

CVR Transcript

FOREWARD

The following is a transcript of the elements which were comprehensible at the time of the readout of the cockpit voice recorder (CVR) of the Boeing 737-200 registered 7T-VEZ involved in the accident at Tamanrasset on 6 March 2003. This transcript contains conversations between crew members, radiotelephonic messages and various noises corresponding, for example, to the movement of selectors or to alarms.

The reader's attention is drawn to the fact that the recording and transcript of the CVR are only a partial reflection of events and of the atmosphere in a cockpit. Consequently, the utmost care is required in the interpretation of this document.

The voices of crew members are heard via the cockpit area microphone (CAM). They are placed in separate columns for reasons of clarity. One column is reserved for the voices of others, noises and alarms, also heard via the CAM.

GLOSSARY

APPENDIX 4

FDR Graphs

APPENDIX 5

Performance calculations

APPENDIX 6

Engine failure on take-off

(1) Except where otherwise noted, the times shown in this report are expressed in Universal Time Coordinated (UTC). One hour should be added to obtain the applicable time in Algeria on the day of the accident.

(2) It was not possible to determine whether this abnormal value corresponded to the display made before the flight or resulted from manipulation after the accident. (See 1.16.4.3)